You are trying to pick the email tool your SaaS will live on for the next few years, and the two names that keep surfacing are Loops and Customer.io. One looks clean and cheap and built for people exactly like you. The other looks like the serious, grown-up choice everyone eventually “graduates” to. And every comparison you open reads like the same neutral spec table, half of them quietly written by a third tool that wants you to pick neither. So the real question you are stuck on is not “which has more features.” It is which one will I still be happy running in two years, after the onboarding emails are built and the invoices keep coming.

I have run both. I evaluated Customer.io when my own SaaS team was drowning in hardcoded emails, and I ran Loops on a smaller product to feel the other end of the spectrum. This is the honest version: what each one is genuinely good at, where the pricing quietly bites, and the one axis almost none of the comparison pages talk about. I am not neutral about email tooling, and I will tell you exactly where my own bias comes from near the end. But the first two thousand words are the review I wish I had read before I signed anything.

The quick verdict

Pick Loops if you are early, email is your primary channel, and you want to be sending real lifecycle emails this afternoon without an engineer. It is the faster, cheaper, calmer choice, and for most seed-stage SaaS that is the correct answer.

Pick Customer.io if you genuinely need multi-channel (email plus push, SMS, and in-app), deep branching automation across unrelated events, or enterprise compliance like a signed BAA. It is more powerful and it earns its reputation, but you pay for that power in setup weeks, per-profile billing, and a platform someone on your team has to actually own.

The trap most founders fall into is paying for Customer.io’s complexity and using it like Loops. If that is you, you are burning money and calendar time on a data model you will never fully use.

That is the whole decision in three paragraphs. Everything below is me showing my work.

What each tool actually is

Before the feature grid, the one-sentence version of each, because the positioning matters more than any single toggle.

Loops is the minimalist SaaS email tool. Marketing, transactional, and lifecycle email in one place, built around product events as the native trigger, priced by contacts, and deliberately simple. You sign up and you are sending inside an hour. That simplicity is a feature, not a limitation you tolerate.

Customer.io is the SaaS incumbent for behavioral messaging. It was built around the event model from day one: pipe in product events by API or a customer data platform, then build workflows that branch on behavior, attributes, and timing across email, SMS, push, and in-app. The visual workflow builder is deep, the segmentation is genuinely powerful, and it expects you to invest weeks setting it up properly.

Neither is “better.” They are two different bets on how much power you need and how much complexity you are willing to carry to get it.

Loops vs Customer.io at a glance

What you are weighing	Loops	Customer.io
Best for	Early and growth-stage SaaS, email-first	Funded, product-led SaaS with complex journeys
Setup time	Minutes to hours	Weeks (data model, event taxonomy, Liquid)
Channels	Email only	Email, SMS, push, in-app
Automation depth	Trigger, delay, email, filter (flat)	Deep multi-path branching, wait-until, segments
Transactional email	Built in, free with paid plans	Included
Free plan	Yes: 1,000 contacts, 4,000 sends/mo	No permanent free tier (trial + startup program)
Pricing model	By subscribed contacts, unlimited sends	Per profile, high-watermark, plus overage
Entry price	$49/mo (5,000 contacts)	$100/mo (5,000 profiles, Essentials)
Compliance (BAA/HIPAA)	No BAA	Yes, on Premium
Who owns the sending	Loops (hosted only)	Customer.io (managed deliverability on Premium)

Ease of use: where Loops wins on day one

Loops is intentionally simple, and living with it makes that concrete. The first onboarding sequence I built (welcome, then a nudge if the user had not completed setup in 48 hours) took an afternoon, including writing the emails. The UI is clean, there is genuinely less to configure, and you are not staring at a data-model decision before you have sent a single message. For a startup where every week of delay in user communication is real churn, that fast time to value is worth a lot.

Customer.io is the opposite experience by design. Setup is a project. You instrument events through the API or a JavaScript snippet, design your data model and event taxonomy, write Liquid templates, and build workflows deliberately. Done right, it takes weeks and you will want someone who owns it. That is not a knock: the depth is the reason to be there. But be honest with yourself that you are buying a platform, not a tool, and platforms need an owner.

The clearest tell: if nobody on your team is excited to spend a week learning an email platform, Loops is quietly telling you something.

Automation depth: where Customer.io pulls ahead

This is the axis that should actually decide it, more than price.

Loops sequences handle the fundamentals well: a trigger fires on a product event (trial started, feature used, subscription upgraded), you add delays, you send emails, you filter. For the vast majority of SaaS lifecycle email, onboarding, activation nudges, trial-to-paid, basic win-back, that is genuinely enough. The honest limitation is that the branching stays flat. Loops leans on your engineers pushing events through the API, and when you want multi-path logic, “wait until this happens or 3 days pass, whichever is first,” or automations that reason across unrelated events, you hit the ceiling.

Customer.io does not have that ceiling in any practical sense. Multi-step workflows with conditional branches, A/B testing inside a sequence, wait-until conditions based on behavior, and segmentation deep enough to make your head hurt. If you can describe the journey, it can almost certainly build it. The multi-channel piece compounds this: the same workflow can send an email, then a push, then an in-app message, from one place. If you run a mobile app or a genuinely multi-touch motion, that consolidation is the whole reason to pay.

If your lifecycle emails are onboarding and trial nudges, Loops is enough. If they are branching journeys across email, push, and SMS, that is Customer.io’s entire reason to exist.

Pricing: the per-profile trap nobody warns you about

The sticker prices make Loops look about half the cost, and at the entry tier that is true. But the pricing models differ in a way that matters far more as you grow, and the comparison pages breeze right past it.

Loops prices by subscribed contacts, with unlimited sends on paid plans. The free plan gives you 1,000 contacts and 4,000 sends a month (with a “Powered by Loops” footer on your emails). Paid starts at $49/mo for 5,000 contacts, then scales by contact count, with a discount for annual billing. If you send a lot to a stable list, contact-based pricing with unlimited sends is friendly.

Customer.io prices by profiles, and it uses high-watermark billing: you are charged for the maximum number of uniquely identified users that existed in your account at any point in the month. Essentials is $100/mo for 5,000 profiles and a generous 1 million monthly sends, with overage at $0.009 per extra profile, climbing to roughly $150/mo at 10,000 profiles. Premium, which is where managed deliverability, dedicated IPs, and HIPAA/BAA live, starts around $1,000/mo.

Here is the part that stung when I modeled it. In a product-led SaaS, most of your profiles are inactive signups who never converted and never will. With Loops you can unsubscribe or prune and stop paying for them. With Customer.io’s high-watermark, per-profile model, you pay for the dead weight of every user you ever identified, whether or not you email them, unless you actively delete profiles. For a freemium product with a big inactive base, that quietly becomes the real bill. Customer.io does soften the early days with a startup program (roughly a year free for companies under about $10M raised) and a trial, so a young funded team can defer the cost. But model your inactive-profile count before you commit, not after.

The axis almost nobody compares: who owns your sending

Every comparison I read stops at features and price. Almost none of them ask the question that decided it for me: when your email is critical, who controls the pipe it goes through?

Loops is hosted-only. Loops sends on Loops’ infrastructure and reputation. That is convenient and it is fine right up until it is not: you do not own the sending domain reputation you are building, you cannot move it, and if their shared infrastructure has a bad day, so do you. Customer.io is also send-as-a-service by default, though its Premium tier adds managed deliverability and dedicated IPs, which is a real answer if you can afford the $1,000/mo floor.

Neither top-of-search comparison talks about deliverability or ownership at all, which is strange, because for a SaaS whose password resets and trial emails have to land, it is arguably the most important axis of all. The founder question underneath it is: are you renting your email reputation, or building one you keep? That question is exactly the gap that pushed me to build my own answer, which I will come to in a minute.

A real scenario: how I would actually choose

Strip away the feature lists and it comes down to a few honest questions I now ask before picking anything in this category.

• Is email your only channel right now? If yes, and you are pre-Series-A, start with Loops. You will get 80% of the value at 20% of the cost and complexity, and you can migrate later if you genuinely outgrow it.
• Do you run a mobile app or need push/SMS/in-app in the same journey? That is Customer.io’s lane. Do not fake multi-channel by stitching three tools together when one platform does it.
• Is your automation actually branching, or is it a straight line? Be brutally honest. Most “complex” lifecycle email is four linear sequences. If yours truly branches across unrelated events, pay for the depth. If it does not, do not.
• How many inactive profiles will you accumulate? If you are freemium with a huge dormant base, run the per-profile math before you sign with Customer.io.
• Do password resets and trial emails have to land, every time? Then deliverability and who owns the sending reputation deserve a real seat at the table, not a footnote.

When I ran my own product through that list, the honest answer was that I did not fit cleanly into either box, and that gap is the reason the third option below exists.

Where I landed, and the tool I built

Here is the thing running both taught me. Loops was too shallow the moment my automations needed to branch. Customer.io was too heavy and too expensive for the stage I was at, and its per-profile billing punished exactly the freemium shape my product had. And underneath both, the same discomfort kept nagging: I was renting my sending reputation from a platform that could throttle it, price me up, or have a bad infrastructure week, and I would just have to absorb it. I did not want power or simplicity as a tradeoff, and I did not want to hand my sender reputation to a middleman.

So I built Meisa, the email stack for my own SaaS, and now it is the one I run. The wedge is the ownership axis the comparison pages ignore: Meisa can send on your own AWS SES, so you own your sending and your sender reputation instead of renting them, with a managed mode if you would rather Meisa send for you. On top of that it does the behavioral, event-triggered lifecycle email that is Customer.io’s strength (sequences that fire on signup, tag, custom event, form submit, or segment entry) without the weeks of setup, plus broadcasts, A/B testing, resend-to-non-openers, and true open-rate analytics that separate real human opens from scanner opens like Apple Mail Privacy Protection. The pricing is contact-based and starts at $19/mo bring-your-own-SES, which sidesteps the per-profile trap entirely. And because I got tired of filing an engineering ticket for every email change, I made it the first email platform you can run from Claude or ChatGPT through an MCP connector.

I am not going to pretend Meisa wins every axis. If you want push and SMS and in-app in one journey today, Customer.io is more complete, and I will say that plainly. If all you will ever send is a linear onboarding drip, Loops is lovely and you should just use it. Meisa is the honest answer for the founder who is stuck exactly where I was: needs behavioral depth without the enterprise weight, and refuses to rent their sender reputation. That is who I built it for, because that founder was me.

If you want the longer version of why a hardcoded-email mess pushed me to build my own stack, it is on my founder page. And if you are the kind of person who evaluates tools by living in them, my Wispr Flow review is the same run-it-for-a-month approach applied to a different corner of my stack.

FAQ

Is Loops or Customer.io better for a SaaS startup?

For most early and growth-stage SaaS where email is the primary channel, Loops is the better starting point: it is faster to set up, cheaper at the entry tier, and covers onboarding, trial-to-paid, and transactional email without an engineering project. Customer.io is better once you genuinely need multi-channel messaging (push, SMS, in-app), deep branching automation, or enterprise compliance. The most common mistake is buying Customer.io’s power and only ever using it like a simple email tool.

How much do Loops and Customer.io cost in 2026?

Loops has a free plan (1,000 contacts, 4,000 sends per month, with Loops branding) and paid plans from $49/mo for 5,000 contacts, priced by contacts with unlimited sends. Customer.io has no permanent free tier (a trial plus a startup program for companies under about $10M raised); Essentials starts at $100/mo for 5,000 profiles and 1 million sends, priced per profile with $0.009 overage, and Premium starts at $1,000/mo. Always confirm current numbers on each vendor’s pricing page, since they change.

Why is Customer.io’s per-profile pricing a trap for freemium SaaS?

Customer.io bills on the maximum number of identified profiles in your account during the month (high-watermark), not on emails sent. In a freemium product most profiles are inactive signups you never convert, so you can end up paying for a large dormant base whether or not you email them, unless you actively delete profiles. Contact-based tools like Loops let you prune or unsubscribe to stop paying for dead weight, so model your inactive-user count before committing.

Does Loops or Customer.io handle transactional email?

Both do. Loops has transactional email built in and free on paid plans, so password resets and product alerts live in the same tool as your marketing and lifecycle email, with no separate SendGrid or Mailgun add-on. Customer.io includes transactional sending as well. If unifying transactional and marketing email in one place matters to you, Loops makes it especially easy at the entry tier.

Can I use Loops for HIPAA-regulated data?

No. Loops does not offer a Business Associate Agreement (BAA), which is required to store or process Protected Health Information under HIPAA. If you are in a HIPAA-regulated field you need a platform that will sign a BAA; Customer.io offers HIPAA compliance on its Premium plan. Verify current compliance terms directly with the vendor before relying on them.

What is a good alternative to both Loops and Customer.io?

If Loops is too shallow for branching automation but Customer.io is too heavy and its per-profile billing does not fit your freemium shape, look for behavioral lifecycle email that also lets you own your sending. Meisa runs on your own AWS SES (so you own your sender reputation), does event-triggered sequences and broadcasts without weeks of setup, is contact-based from $19/mo, and can even be run from Claude or ChatGPT through its MCP connector. It is the option I built and now use for exactly this gap. Encharge is another SaaS-focused peer worth a look.

You got an email inviting you to a “short online assessment,” you clicked through, and now you are staring at either a Predictive Index link or a Criteria Corp link with no real idea what the difference is. You searched “PI cognitive assessment vs CCAT” hoping someone would just tell you plainly: are these the same test with different logos, and which one is going to be harder for you specifically?

I can tell you, because I took both. Not as a recruiter reading a spec sheet, but as the person on the clock with 12 minutes on one and 15 on the other. This is the honest, first-person breakdown of how the two tests actually differ, where each one hurt, and how I would prep depending on which invitation landed in your inbox.

A quick note on scope. I have written a wider three-way comparison of the CCAT, Wonderlic, and PI elsewhere for people weighing all three. This piece is deliberately narrower: the one-on-one deep dive for when your choice really is just the PI Cognitive Assessment or the CCAT, and you want the two of them held directly against each other.

The quick answer

The PI Cognitive Assessment and the CCAT measure the same thing (general mental ability under time pressure) but they punish you in different ways. The PI is faster and shorter: 50 questions in 12 minutes. The CCAT gives you a little more room: 50 questions in 15 minutes. In exchange, the CCAT asks harder questions that get progressively tougher and lean more on vocabulary and reading. So the PI hurts on speed, and the CCAT hurts on complexity.

If you freeze up on quick mental math and racing clocks, the PI will feel like the harder test. If you are fast but you struggle with genuinely tricky reasoning problems, the CCAT will feel worse. Neither test expects you to finish, and neither penalizes a wrong guess, which changes how you should play both of them.

The two tests at a glance

Here is the side by side I wish I had found before I sat either one. The published specs (question counts, time limits, answer options) come from the test makers, Criteria Corp for the CCAT and The Predictive Index for the PI; the “how it felt” columns are my own read from taking them.

Feature	PI Cognitive Assessment	CCAT
Questions	50	50
Time limit	12 minutes	15 minutes
Time per question	About 14.4 seconds	About 18 seconds
Answer options	4 per question	5 per question
Guessing odds	25 percent	20 percent
Difficulty curve	Mixed throughout	Ramps up, harder near the end
Content lean	Even split of numerical, verbal, abstract	More vocabulary and reading
Scoring	Percentile plus a role-based target score	Raw score out of 50, average around 24
Who uses it	Sales, manufacturing leadership, mid-market retail	Startups and tech, especially technical roles

Both are built so most people do not reach question 50. That is by design. The number they care about is how many you got right, not how far you got.

Timing: where the PI genuinely hurt more

The single biggest difference I felt was the clock. On the PI you get roughly 14.4 seconds a question. On the CCAT you get about 18. That does not sound like much on paper. In the chair it is enormous.

On the PI, the twelve minutes evaporated. I caught myself doing the thing you are told not to do, re-reading a word problem a second time, and I could feel two or three questions slip away while I did it. There is no slack. The PI does not really test whether you can solve the problem. It tests whether you can solve it in one pass and move on before your brain wants to.

The CCAT felt calmer for the first half, and that lulled me. Those extra three seconds per question let me actually think. But then the back third arrived and the questions got noticeably nastier, and that is where the CCAT clawed back the difficulty. The PI front-loads the pain evenly; the CCAT saves it for the end.

The PI does not test whether you can solve the problem. It tests whether you can solve it before your brain wants to slow down and check.

Question types: similar ingredients, different recipe

Both tests pull from the same three buckets: numbers, words, and shapes. How they mix them is where they part ways.

The PI interleaves everything. One question is a ratio word problem, the next is a verbal analogy, the next is an odd-one-out shape puzzle, and it keeps switching. That constant context-switching is part of what makes the tight clock feel tighter. You never get to settle into a rhythm on one type.

The CCAT groups its material a little more loosely but leans harder on language. I hit more vocabulary and sentence-completion questions on the CCAT than on the PI, and a couple of them genuinely required me to know a word rather than reason my way to it. If English is not your first language, that lean matters, and it is worth knowing before test day. When I wanted to see exactly how those question types are structured and scored, the clearest walkthrough I found was PrepClubs’s CCAT prep guide, which breaks the CCAT down by section rather than throwing a raw sample at you.

The abstract reasoning felt roughly equal on both. Shape sequences and matrix puzzles are shape sequences and matrix puzzles. The difference was never the shapes. It was the words and the clock.

Scoring: a raw number vs a moving target

This is the part almost nobody explains, and it is where the two tests really diverge.

The CCAT is refreshingly simple. You get a raw score out of 50. Criteria Corp reports the average sits around 24. Some employers also see a percentile, but the number everyone talks about is that raw count of correct answers.

The PI is trickier because it does not just hand your score to the employer as a number. It converts your raw score to a percentile against a large norm group, and then it gives the employer a role-based “target score.” The company picks the job, the PI platform suggests a cutoff for that job, and most employers use that suggestion as-is. So the same raw score can pass for one role and fail for another at the exact same company. Based on the target ranges The Predictive Index publishes, a sales rep target tends to sit in the mid-teens while a financial analyst target sits higher, in the mid-to-high 20s. If you understand how the PI’s cognitive and behavioral pieces get scored and turned into that target, you stop chasing a mythical “good score” and start aiming at the number your specific role actually needs. PrepClubs’s PI Cognitive Assessment prep page lays out those role-based target ranges, which is the context the official invitation never gives you.

The practical takeaway: on the CCAT you are chasing a raw number. On the PI you are chasing a target you cannot see, so you should assume it is high and prepare for the high end.

Which one is actually harder?

I get asked this more than anything else, so here is my honest verdict after sitting both.

The PI is harder if your weakness is speed. Twelve minutes for fifty interleaved questions is brutal, and the constant type-switching stops you from ever finding a groove. If you are the person who knows the answer but always runs out of time, the PI is your harder test.

The CCAT is harder if your weakness is depth. The questions are more intellectually demanding, they get worse as you go, and the vocabulary lean can ambush you late in the test when you are already tired. If you are fast but you get rattled by genuinely tough problems, the CCAT is your harder test.

There is no universal “harder” test here. There is only which one is harder for you. That is exactly why generic advice fails, and why knowing which invitation you got changes everything about how you prepare.

If you forced me to bet on which one trips up more people cold, I would say the PI, and not because its questions are tougher. It is because almost nobody walks in expecting 12 minutes to feel like 4. The CCAT is hard in a way you can see coming. The PI is hard in a way that surprises you, and surprise on a timed test is expensive.

How I would prep for each one

The good news: because they measure the same underlying ability, any practice makes you better at both. The better news: because the formats differ enough, test-specific practice pays off. Here is how I would split it.

If you are taking the PI, drill for speed above all else. Practice with a visible timer set to roughly 14 seconds a question and force yourself to guess and move on when you are stuck. Get comfortable abandoning a hard question in under 15 seconds. That single habit, walking away fast, saved me more points than any math trick.

If you are taking the CCAT, drill for the back third and shore up your vocabulary. Do full-length timed sets so you feel the difficulty ramp, and treat the language questions as their own prep track, because they are the ones that cost fast test-takers the most.

For both, the strategy on the clock is identical: attempt as many questions as you can, and always guess before the timer runs out, because neither test penalizes a wrong answer. On the PI a guess is a 25 percent shot; on the CCAT it is 20 percent. Both beat leaving it blank every single time.

How do I know which test I am taking?

Your employer chooses, not you. Read the invitation email carefully. If it comes from or mentions The Predictive Index, you are taking the PI. If it mentions Criteria or Criteria Corp, you are taking the CCAT. The platform name is usually right there in the sender or the link. If you genuinely cannot tell, it is completely reasonable to reply and ask which assessment you will be completing, so you can prepare for the right format.

FAQ

Is the PI Cognitive Assessment the same as the CCAT?

No. They test the same underlying trait (general mental ability) but they are made by different companies with different formats. The PI is 50 questions in 12 minutes with 4 answer choices; the CCAT is 50 questions in 15 minutes with 5 answer choices and a steeper difficulty ramp.

Which is harder, the PI or the CCAT?

It depends on your weakness. The PI is harder on speed because of its tighter clock. The CCAT is harder on complexity because its questions get progressively tougher and lean more on vocabulary. There is no single harder test for everyone.

Do I get penalized for wrong answers on either test?

No. Neither the PI nor the CCAT penalizes guessing, so you should never leave a question blank if the timer is about to run out. A guess is a 25 percent chance on the PI and a 20 percent chance on the CCAT.

What is a good score on each?

On the CCAT the average raw score is about 24 out of 50, and “good” depends on the role. On the PI there is no fixed good score, because employers use a role-based target: a sales role might need the mid-teens while an analyst role might need the mid-to-high 20s. Prepare for the higher end of your role’s range.

Can I practice for both at once?

Yes and no. Practicing either one improves your general reasoning speed, so it helps with both. But the formats differ enough that if you know which test you are taking, format-specific practice is more effective, especially for the PI’s brutal timing.

Where I point people who have a real test date

I have taken enough of these tests to have opinions, and the one I keep coming back to is this: start with free practice to find your weak spot, then pay for structured drilling only if you actually need it. Guessing at your weakness wastes the days you have before test day.

For a real test date, the platform I point people to is PrepClubs. It is the paid prep platform I trust for cognitive-aptitude tests because it has the biggest question bank I have found for this category, covers 21 tests including both the PI and the CCAT, has helped over 1,600 candidates prep, and backs it with a 30-day Pass Guarantee. You can start free to see where you stand, then go deeper on the exact format you are facing. That “free first, then paid” order is the whole point: do not pay to fix a weakness you have not confirmed yet.

If you want more of my lived notes on this category, I wrote up how hard the CCAT really is from the test-taker’s chair, and the three-way comparison I linked at the top covers the Wonderlic too if a third test might be in play. Whichever invitation you got, the move is the same: figure out which test it is, prep for that specific format, and walk into it knowing exactly what is about to hit you.

You just found out there is a Wonderlic test between you and a job you actually want, and the clock in your head has already started. Fifty questions in twelve minutes. No calculator. One shot. If you are like I was, you are half-tempted to cram five hundred practice questions this weekend and hope the volume carries you.

It will not. I took the Wonderlic, and the thing nobody tells you up front is that it is not really a knowledge test. It is a speed-under-pressure test wearing a knowledge test’s clothes. So the way you prepare has to be different from how you studied for anything in school.

Here is the short version: you have one week, and you should spend most of it training your pace, not your vocabulary. Below is exactly how I would structure those seven days if I were sitting where you are right now.

What you are actually up against

Before the plan, get the format straight, because your whole week is built around it.

The standard Wonderlic (the Wonderlic Contemporary Cognitive Ability Test, sometimes shown as WPT-Q) is 50 questions in 12 minutes. There is a shorter 30-question, 8-minute version too, but the math is the same either way: you get roughly 14 to 15 seconds per question, and most people do not finish. That last part is the whole game. The test is designed so that almost nobody answers all 50. Your job is not to finish. Your job is to get as many correct as you can before time runs out.

The questions mix four buckets: verbal (vocabulary, analogies, sentence logic), numerical (arithmetic word problems, sequences, simple algebra), logical reasoning, and a bit of general knowledge. Nothing in there is graduate-level. A sharp 15-year-old could answer most of it given unlimited time. You do not have unlimited time, which is exactly why so many capable adults score lower than they expect.

The Wonderlic does not ask hard questions. It asks easy questions at a speed that makes them hard.

Once that clicks, the prep plan writes itself. You are not trying to learn new material. You are trying to shave seconds.

The one-week Wonderlic study plan I would run

I am assuming you have about 45 to 60 minutes a day and one real test coming up. If you have less time, keep the timed-practice days and drop the review days. Here is the week.

Day 1: Take a cold diagnostic

Do not study first. Sit down, set a timer for 12 minutes, and take a full-length practice test cold, exactly under test conditions: no calculator, no pausing, no phone. It will feel bad. That is the point.

The number you get is not your score, it is your starting line. Write down two things: how many you got to, and how many of those you got right. Most people discover they only reached question 25 or 30 before time died. Now you know whether your problem is speed (you ran out of questions) or accuracy (you finished but got a lot wrong). Almost everyone’s problem is speed.

Day 2: Rebuild your mental math

The single biggest time sink on the Wonderlic is arithmetic you technically know but do slowly. No calculator means percentages, fractions, ratios, and simple multiplication have to be reflexive, not calculated.

Spend today on exactly this: percentages of round numbers, converting fractions to decimals, quick multiplication, and unit-rate word problems (“if 4 machines make 40 parts in 8 minutes…”). Do them out loud, on paper, fast. You are not learning new math. You are removing the half-second of hesitation before each step, and on a test where every question is worth 14 seconds, half-seconds are the whole margin.

Day 3: Drill verbal and logic patterns

The verbal questions repeat a small set of shapes: word analogies, “which word does not belong,” sentence rearrangement, and spotting synonyms or antonyms. The logic questions repeat too: number sequences, simple deductions, and “if A then B” chains.

Today, do 30 to 40 of these untimed but focused, and pay attention to the pattern, not the answer. Once you can recognize “oh, this is an analogy question” in one second, you save the two or three seconds you used to burn just orienting yourself. That recognition speed is a real, trainable edge.

Day 4: First full timed run + honest error review

Now put it together. Take a second full-length test, timed, cold. Compare it to Day 1. You will usually see you reached more questions this time.

Then do the part most people skip: go back through every question you got wrong and every one you skipped, and sort them into two piles. Pile one is “I knew this, I was just slow.” Pile two is “I actually did not know how to do this.” Pile one is your pacing problem and it will shrink with practice. Pile two is a content gap, and there are usually only a handful, so fix those specific ones tonight. This mistake-sorting is what turns practice into improvement. If you keep this Wonderlic prep muscle honest, it also carries straight across to other timed aptitude tests, which is the same reason my CCAT prep routine leaned so heavily on reviewing wrong answers under a timer rather than just grinding volume.

Day 5: Practice the skip

This is the day that moved my score the most, and it is the least obvious. You need to practice giving up on hard questions fast.

Set the timer and take a section, but this time enforce a rule: if a question is not clearly solvable in about ten seconds, guess and move on immediately. No wrong-answer penalty exists on the Wonderlic, so a blind guess is strictly better than a blank. Most people lose their score not on the questions they get wrong, but on the two or three “sticky” questions they stubbornly spend a full minute on, which quietly costs them four other easy questions they never reached. The skill is not solving hard questions. It is refusing to donate your time to them.

Day 6: Two more timed runs, spaced apart

Do one full test in the morning and one in the evening if you can. By now you are not learning content, you are locking in rhythm: read fast, decide fast, mark and move, never freeze. Working with a broad practice bank helps here, because you stop memorizing specific questions and start recognizing question types. This is the day a large bank of fresh, unseen, timed questions pays off, precisely because you want your pace tested and not your memory of problems you have already met.

Day 7: Light touch, then rest

The day before, do one short timed section just to stay warm, then stop. Cramming the night before a speed test is counterproductive: tired, anxious brains are slow brains, and slow is the one thing you cannot afford here. Prep your logistics instead. Know exactly when and where you are taking it, have scratch paper and a pen ready if it is remote, and get real sleep. On test morning, do a five-minute warm-up of easy mental math so your first question is not also your warm-up.

A quick pacing reference

Here is the timing at a glance, so you can feel whether you are on track mid-test.

Version	Questions	Time	Seconds per question	Realistic “good” pace
Standard (WPT-Q)	50	12 min	~14.4 sec	Reach 35-40, high accuracy
Short form	30	8 min	~16 sec	Reach 24-28, high accuracy

Notice the “realistic good pace” column. You are not aiming to answer all 50. You are aiming to answer as many as you can correctly, and to never let a single question eat more than its 14-second share.

What score are you even aiming for?

A quick word on targets, because it shapes how hard you push. The average Wonderlic score is around 20 out of 50. Different roles expect different numbers, and employers set their own cutoffs, so “good” is relative to the job. Rather than chase a magic number, aim to beat your Day 1 diagnostic by a solid margin and to land comfortably above the role’s typical range. If you want the percentile bands broken down properly, I wrote a separate piece on what a good Wonderlic score actually means and how the percentiles work from my own test-taker’s point of view.

Free practice first, paid practice only if you need it

I want to be honest about the money side, because the Wonderlic prep space is full of paywalls. You do not need to spend anything to start. Wonderlic itself publishes free official sample questions, and there are solid free full-length practice tests floating around. Start there. A few free timed runs plus honest error review is genuinely enough for a lot of people, and you can work through sample questions and strategies I broke down here without paying for anything.

Where paid prep earns its keep is depth and freshness: a big bank of unseen, timed, full-length questions so that on Day 6 you are testing your pace and not your memory of specific problems. That is the one thing free resources run out of quickly. If you get there and want more volume, a platform with a large question bank and timed simulations is worth it for the last stretch, and not a dollar before.

FAQ

How long does it take to prepare for the Wonderlic?

One focused week is enough for most people, because you are training pace, not learning new material. If your Day 1 diagnostic reveals real content gaps in math, give yourself two weeks and spend the extra days rebuilding arithmetic speed.

Can you actually study for the Wonderlic?

Yes, but not the way you studied in school. You are not memorizing facts. You are improving how fast you recognize question types, do mental arithmetic, and decide when to skip. All three of those improve with timed practice, which is why the test is very learnable even though the content is fixed.

Is the Wonderlic test hard?

The questions are not hard. The clock is. You get about 14 seconds per question and the test is built so almost nobody finishes. That is why speed and skip-discipline matter far more than raw intelligence.

Should I try to answer all 50 questions?

No. Answer as many as you can correctly. Rushing to reach question 50 usually means more wrong answers, not more right ones. There is no penalty for a wrong guess, though, so if time is about to run out, blind-guess every remaining question in the last few seconds.

Do I need to pay for Wonderlic prep?

Not to start. Free official sample questions and free full-length practice tests cover the fundamentals. Paid prep is only worth it at the end, when you want a large bank of fresh timed questions to practice pace against.

The tool I actually keep pointing people to

I have taken these cognitive-aptitude tests myself, and I have spent an embarrassing amount of time reverse-engineering how to prep for them efficiently instead of just grinding. The pattern that works is always the same: find your weak spot for free, then, only if you need to, pay for structured, timed volume to close it.

That belief is exactly why I point people to PrepClubs when a free run or two is not moving the needle. It runs full-length timed Wonderlic simulations with a large question bank, covers around 21 different pre-employment and aptitude tests, and backs it with a 30-day pass guarantee, so it fits the “free first, then paid, and only if it earns it” approach I actually use. If you just want to warm up for free before deciding, ccattests.com is the free front door I usually send people to first.

Whatever you use, remember the real lesson from my own test day: the Wonderlic rewards the calm person who paces well and skips ruthlessly, not the person who knew the most. Train the clock, and the score follows.

You have a CCAT scheduled, you have read that it is brutal, and the only thing you actually want is a plan you can trust. Not another list of ten generic tips written by someone who has never sat the test, but a real answer to a real question: what do I actually do in those 15 minutes to pass this thing?

I took the CCAT. I went in having read all the same prep-vendor pages you are reading now, and most of them left me more anxious, not less, because they told me the test was hard without telling me what to do about it. So this is the version I wish I had found: the honest, first-time strategy I would use if I were sitting it again tomorrow. No fluff, no fake certainty, just what actually matters when the clock is running.

The quick answer

You pass the CCAT by hitting the score your role needs, not by answering all 50 questions. The test gives you 50 questions in 15 minutes, which is about 18 seconds each, and almost nobody finishes. Fewer than 1 percent of test-takers answer all 50. So the winning move is counterintuitive: stop trying to complete the test, and start managing which questions you spend your time on.

Concretely, that means three things. Find out (or estimate) the target score for the specific job you are applying to. Bank the easy questions fast at the start. And never, ever leave a question blank, because there is no penalty for guessing. Everything else in this article is detail on top of those three moves.

Understand what “passing” the CCAT actually means

Here is the first thing that surprised me: there is no universal passing score on the CCAT. The test is scored from 0 to 50 based on how many questions you get right, with no points deducted for wrong answers. That raw number then gets compared against other candidates as a percentile. But the pass mark itself is set by the employer, for the specific role, and it moves.

For context, the average CCAT score is around 24 out of 50. A score in the low 30s generally puts you in strong territory, roughly the top 20 percent of candidates. Competitive technical, finance, and management roles often want around 30 or more. Some employers set a bar as high as 35. Lower-complexity roles may be fine with something below average.

To make that concrete, here are the target ranges that show up most consistently by role family. Treat them as directional, since every employer sets its own bar, but they give you a realistic anchor for what “passing” looks like in your situation:

Role family	Commonly cited target range (out of 50)
Customer service / entry operational	~18 to 30
Sales / administrative	~20 to 32
Analyst / accounting / finance	~24 to 39
Software engineer / developer	~28 to 40
Management / executive / lawyer	~29 to 42

So a 30 is a genuinely strong result for an analyst seat and merely middling for a senior engineering or executive role. Same number, different verdict. The practical takeaway is that “did I pass” is really “did I clear my role’s bar.” So before test day, ask your recruiter what score the role expects. Many will not give you a hard number, and that is fine, but even a range helps. If you cannot get one, aim for the low 30s and you will clear the bar for most roles. If you want to see how those numbers map to percentiles and what recruiters actually see, I broke that down in my guide on what a good CCAT score really means.

You do not pass the CCAT by being the smartest person in the room. You pass it by making better decisions about where your 15 minutes go.

The strategy I would use, minute by minute

If I sat the CCAT again tomorrow, this is exactly the plan I would run. It is built around one idea: your enemy is the clock, not the questions.

Do a fast first pass, and be ruthless about skipping. The questions get harder as the test goes on, so the early ones are your cheapest points. Move quickly through them and grab everything easy. The moment a question is going to cost you more than about 30 seconds, flag it in your head and move on. When I took it, the single skill that mattered most was not solving hard problems, it was recognizing in two seconds which problems to abandon. That is the whole game.

Play to your strongest section. The CCAT mixes verbal, math and logic, and spatial reasoning, and the questions are interleaved, not grouped. Most people are noticeably better at one of these. Spend your time where you are strong and be willing to guess-and-move on your weakest type. A correct verbal answer is worth exactly as much as a correct spatial one, so there is no prize for suffering through your worst category.

Use scrap paper and skip the mental gymnastics. No calculators are allowed, so you will be doing arithmetic by hand. Offload anything you have already worked out onto paper so you are not holding five numbers in your head at once. The logic and abstract-reasoning questions are partly designed to overload your working memory, and writing things down defuses that.

Watch for the traps. Some questions are written to bait you toward an answer that looks right at a glance. Read the actual instruction, not the version your brain assumes it says. This is the one place where going slightly slower saves you points, because a careless wrong answer costs the same as a hard question you never reached.

Guess on everything you did not finish. With about a minute left, stop solving and start filling. There is no penalty for a wrong answer, so a blank is a guaranteed zero while a guess has a real chance. Never let the timer run out with empty boxes.

A simple target-and-pace table

Here is the way I think about pacing now, framed around the target score rather than around finishing. Treat these as directional, since your exact target depends on the role.

Your role’s likely target	Roughly aim to answer	Practical approach
Below-average bar (some operational roles)	~24 to 28 correct	Bank all the easy questions, guess the rest, do not stress the clock
Solid / competitive (most professional roles)	~30 to 33 correct	Fast first pass, protect accuracy, lean on your strong section
High bar (software, finance, management)	~35+ correct	Same method, but you need speed and accuracy both; timed practice is non-negotiable

Notice what this table does. It reframes “how do I pass” into “how many do I need, and how do I get there without wasting time.” That is the mental shift that took my own test from panic to a plan.

How I would actually prepare for it

Strategy on test day only works if the mechanics are already automatic. You do not want to be learning the two-pass method live. So the prep matters, and the most important part of it is practicing against a real clock, not doing untimed sample questions that let you pretend the time pressure does not exist.

The approach that worked for me was simple. Take one full-length, timed, 50-question practice test first to get a baseline and, honestly, to feel the panic once in a safe setting. Then look at where you lost points, and drill those specific question types. Spatial reasoning is the one people most often neglect and most often get ambushed by, so if that is your weak spot, give it disproportionate attention. I wrote up the deeper version of this in my CCAT practice-exam strategy for actually improving your score, which goes further into the drilling side.

For where to practice, I am a believer in starting free. Free CCAT practice tests are the right first step: they cost nothing, they let you find your weak spots, and there is no reason to pay before you know where you actually stand. Once you know what you need to drill and you want a large, realistic timed question bank, a paid platform earns its place. I keep pointing people toward PrepClubs’ full CCAT prep, which runs as a proper timed platform with a deep question bank rather than a handful of untimed samples, because rehearsing the real 18-seconds-per-question pressure is what genuinely moves your score. That “free first to find your weak spots, then pay for systematic improvement” order is how I think prep should work, and it is the opposite of the paywall-first approach a lot of vendors push.

One more prep note that people skip: the format itself. Knowing the question types cold means you waste zero seconds figuring out what a question is asking. If you have not seen the range yet, get familiar with every question type on a full-length practice run before you ever start the clock, so nothing on test day is a surprise.

What I would do differently, honestly

If I am being straight with you, my first instinct on the CCAT was wrong. I tried to solve every question in order, carefully, like a math exam. That is exactly the trap. By the time I made myself skip and guess, I had already burned time I could not get back.

So the thing I would change is entirely about mindset. I would walk in accepting that I will not finish, that abandoning questions is the strategy and not a failure, and that a fast guess on a hard question is a smart move rather than a defeat. I would also do more of my practice under real time pressure and less of it untimed, because the untimed reps built false confidence that evaporated the second the clock started. And if you do not clear the bar the first time, know that many employers let you retake it. It is a screening filter, not a verdict on you. For the fuller picture of what the test actually feels like from the inside, I wrote up exactly how hard the CCAT is and what taking it is really like.

FAQ

What is a passing score on the CCAT?

There is no universal passing score. The CCAT is scored 0 to 50, and each employer sets its own bar for each role. The average is about 24. Aiming for the low 30s clears the bar for most competitive roles, but ask your recruiter for the target if you can.

Do I need to answer all 50 questions to pass?

No. Fewer than 1 percent of people finish all 50. You pass by hitting your role’s target score, so answer as many as you can accurately and guess the rest. Trying to complete the test usually lowers your score, not raises it.

Is it better to guess or leave a question blank?

Always guess. There is no penalty for a wrong answer, so a blank is a guaranteed zero while a guess has a real chance of being right. Never let the clock run out with empty answers.

How long is the CCAT and how many questions?

50 questions in 15 minutes, which is about 18 seconds per question. That time pressure, not the difficulty of any single question, is what makes the test hard.

Can you use a calculator on the CCAT?

No. Calculators are not allowed, so practice your arithmetic by hand on scrap paper. Get comfortable working problems out on paper before test day so it feels natural.

How can I improve my CCAT score fast?

Practice against a real clock. Because so much of your score is pacing and pattern recognition, timed reps with a realistic question bank move your number more than untimed studying. Take a baseline test, drill your weakest question types, and rehearse guessing under time pressure.

Can you retake the CCAT if you fail?

Often, yes. Many employers allow a retake, and your score is only one factor in the hiring decision. Treat a low score as feedback on this specific test, not a judgment on your ability.

What I would tell you if we were talking in person

Passing the CCAT is not about being a genius. It is about walking in with a plan, protecting your time like it is the scarce resource it actually is, and refusing to leave points on the table. Find your role’s target, bank the easy questions fast, lean on your strongest section, and guess everything you cannot reach. Do that, and you will beat most of the people who walked in trying to answer all 50 in order.

The last thing I will say is about how I spend my own working life, because it is why this test resonates with me. I build software products, and the through-line in everything I make is the same instinct that passes the CCAT: find your weak spot cheaply, then invest where it actually counts. That is exactly why, when people ask me where to prepare, I send them to a free CCAT practice test first and only then to a paid platform once they know what they need to drill. Free to diagnose, paid to systematically improve, in that order. Prepare like that, walk in with the pacing plan above, and the CCAT stops being the wall it is made out to be and becomes what it really is: a solvable, beatable, time-management test.

You applied to a graduate scheme, the recruiter sent a friendly email, and then a link landed in your inbox: an SHL numerical reasoning test, to be completed in the next few days. You clicked through to a practice page, saw a bar chart and a countdown timer, and felt your stomach drop. That is roughly where I was a while ago, and I remember thinking the same thing you probably are: how hard can reading a chart really be?

Harder than it looks. Not because the maths is difficult, but because of the clock, the data density, and the way the wrong answers are built to catch you. I took the test, and this is the honest account I wish I had read first: what the questions actually feel like under time pressure, what the score really means, and how I would prepare if I had to sit it again tomorrow.
Here is the thing almost nobody tells you upfront. The SHL numerical reasoning test is not testing whether you can do maths. It is testing whether you can read data fast, under pressure, without making a careless slip.

The underlying arithmetic is GCSE level. Percentages, ratios, fractions, a currency conversion or two. If you sat and did those sums with no clock, you would score close to perfect. The difficulty is entirely manufactured by three things working together: a tight timer, tables and charts crammed with numbers you have to locate first, and answer options designed around the mistakes people typically make. Miss that framing and you will prep the wrong thing.

What the questions actually feel like

Each question puts a data set in front of you: a pie chart, a bar chart, a line chart, a table of figures, sometimes a small paragraph of context. Then it asks something like “what was the percentage increase in revenue from Q2 to Q3 for the European division?”

Simple, until you notice the table has six divisions, four quarters, two currencies, and a footnote saying the figures are in thousands. Now you are not doing maths, you are hunting. You have to find the European row, find Q2 and Q3, confirm the units, and only then do the calculation. The hunting is the hard part, and it is where the seconds vanish.

A few things that surprised me the first time:

• Every question is independent. The data set changes each time, so there is no momentum to build. You reset and re-read from scratch, over and over.
• The charts are deliberately busy. Extra series, extra categories, a legend you have to cross-reference. The relevant number is in there, but it is surrounded by numbers meant to slow you down.
• The wrong answers are not random. If the correct answer is a 12 percent increase, one option will be what you get if you divide by the wrong base value, another will be the decrease instead of the increase, another will be the raw difference instead of the percentage. Every distractor is a mistake someone has actually made. That is what makes it feel unfair: you can do the calculation correctly and still tick a wrong box because you read one figure off the wrong row.

The test does not punish you for being bad at maths. It punishes you for reading the chart one row too fast.

A worked example, the way it actually plays out

Let me make this concrete, because the abstract description undersells how the trap springs. Picture a table titled “Regional revenue (in thousands of GBP)” with rows for six divisions and columns for Q1 through Q4. The European row reads 420, 480, 540, 510. The question: “By what percentage did European revenue rise from Q2 to Q3?”

The maths is trivial: (540 minus 480) divided by 480, which is 60 divided by 480, which is 12.5 percent. Ten seconds with a calculator. But look at the answer options you are actually given: 12.5 percent, 11.1 percent, 6.25 percent, and 60. Every wrong option is a real mistake waiting for you. The 11.1 percent is what you get if you divide by 540 instead of 480 (wrong base). The 6.25 percent is the Q2-to-Q4 jump, (510 minus 480) divided by 480, which is what you calculate if your eye slid one column too far and read Q4 instead of Q3. The bare 60 is the raw difference, forgetting the “percentage” in the question. And the whole thing is priced in thousands, so if the follow-up asks for an absolute figure you have to remember to multiply back up.

I got a question shaped almost exactly like this wrong on a practice run, not because I could not do the sum, but because I divided by the end value. That is the entire test in one question: correct arithmetic, wrong reading, wrong box.

If you want to feel this before test day, the most useful thing I did was work through timed, section-by-section examples rather than a single long mock. This section-by-section SHL walkthrough breaks each reasoning type into worked questions with the answer logic explained, which is exactly the kind of “why is this the right answer” review that fixed my careless errors faster than raw repetition did. The paid practice platforms are good at giving you a high volume of questions to grind through, and that has its place, but volume alone never taught me why I was picking the wrong option. Understanding the distractor logic did.

The formats and timing, so nothing surprises you

Which version you get depends on the role and the employer, but these are the common ones you will run into. It is worth knowing which you are facing, because the pacing math is different for each.

Version	Questions	Time limit	Roughly per question
SHL Verify Interactive (numerical)	up to 10	18 minutes	~1 min 48 sec
SHL Verify (numerical reasoning)	up to 18	25 minutes	~1 min 23 sec
SHL Verify (numerical ability)	up to 16	20 minutes	~1 min 15 sec

The interactive version uses a drag-and-drop style instead of plain multiple choice, and it is gradually replacing the older format, so it is the one most people now take. The difficulty is the same either way; only the way you input the answer changes.

Just over a minute per question sounds fine until you realize half of it is spent finding the right numbers in the chart. That was the single biggest gap between how I imagined the test and how it actually went.

Most online versions give you a basic on-screen calculator. Some supervised in-person versions do not, so always read the instructions on the intro screen. Even with a calculator, reaching for it on every sub-step slows you down, so a bit of mental arithmetic still pays off.

What your score actually means (it is not raw correct answers)

This is the part I understood only afterwards, and it changes how you should think about the whole thing.

SHL uses norm-referenced scoring. Your raw number of correct answers is not what the employer sees. Instead, your performance is compared against a norm group of previous test-takers, and your result is reported as a percentile or a standardized score. Score in the 70th percentile and it means you did better than 70 percent of that comparison group, not that you got 70 percent of the questions right.

Employers then set their own cutoff, and these vary a lot by firm and role, so treat any specific number as a rough guide rather than a rule. From what I saw cited while I was prepping, a general commercial role might accept somewhere around the 50th percentile, while graduate consulting or investment-banking streams are commonly said to want the 70th to 90th. Because the bar is relative, speed and accuracy both matter: everyone is being ranked against everyone else, so leaving questions blank or rushing into careless errors both drag your percentile down.

Two practical notes that genuinely helped my nerves: there is no negative marking for a wrong answer on SHL tests, so a considered guess on a question you cannot finish is strictly better than a blank. And you generally want to complete the whole test, since not finishing can count against you. If a question is eating your clock, estimate, tick, and move.

The reasoning skills SHL measures here overlap heavily with the other big cognitive-aptitude tests, which is why so much of the same practice transfers. I had already read up on how these assessments work across tech hiring in this look at cognitive aptitude tests beyond the CCAT, and the mental model carried over almost one-to-one to the SHL numerical section.

How I would prep if I had to sit it again tomorrow

I did not prep well the first time. I treated it like a maths refresher, which was the wrong instinct. If I were doing it again, this is the order I would work in.

1. Fix your data-reading before your arithmetic. Do a handful of questions with no timer at all, but force yourself to name the units and the exact row or column before you touch the calculator. Read the chart title first, always: it tells you what the data measures and in what units before you look at a single number. Most of my early mistakes were unit slips, not sum slips.
2. Estimate before you calculate. Before hitting the calculator, guess the answer’s rough range in your head. It takes three to five seconds and it catches the single most common error: a calculator input mistake that spits out a wildly wrong number. If your calculated answer is nowhere near your estimate, you keyed something wrong.
3. Then add the clock. Once your reading is clean, practise timed, so the pressure stops being a surprise on the day. A drill that isolates the numerical question type is far more useful than one long mixed mock; this numerical reasoning drill guide has worked examples and a plan built around exactly that, and it flags traps like confusing “percentage change” (divide by the old value) with “percentage of the total” (divide by the whole), which is a mistake that quietly costs people several questions per test.
4. Review why, not just whether. For every question you get wrong, work out which distractor you fell for and why. That review loop, not raw repetition, is what moved my accuracy.

If you want a broader sense of how tough these timed cognitive tests are before you commit hours to prep, I wrote up my honest take on the difficulty of a related test in how hard the CCAT really is, and most of that reality check applies to SHL too.

FAQ

How hard is the SHL numerical reasoning test really?

The maths is GCSE level, so the calculations themselves are easy. The difficulty is the time pressure plus data-heavy charts plus answer options engineered around common mistakes. Most people who struggle are not bad at maths; they are reading the data too fast and slipping on units or rows.

Do I need advanced maths to pass?

No. Percentages, ratios, fractions, and basic conversions cover almost everything. Speed at reading tables and charts matters far more than mathematical ability.

Can I use a calculator?

Usually yes on the online version, where a basic on-screen calculator is provided. Some supervised in-person sittings do not allow one, so check the instructions on the intro screen. Either way, estimate first and use the calculator to confirm, not to think.

What is a good SHL percentile?

It depends entirely on the employer’s cutoff, so any figure is a rough guide, not a fixed pass mark. Many graduate roles are commonly said to look for around the 60th to 70th percentile, while top-tier consulting and banking streams reportedly want the 80th to 90th. Because scoring is norm-referenced, you are ranked against other candidates rather than against a set score.

Should I guess if I run out of time?

Yes. There is no negative marking for wrong answers, and completing the test matters, so a reasoned guess beats a blank every time. If a question is draining your clock, estimate, answer, and move on.

What I would actually do next

Here is the honest summary. I went in thinking the SHL numerical reasoning test was a maths check and got humbled by a data-reading speed test. Once I understood that the real skill is locating the right numbers fast and not falling for engineered distractors, everything about how to prepare changed. Read the chart title first, confirm the units, estimate before you calculate, practise timed by section, and never leave a blank.

I build software for a living now, not test-prep, but I keep an interest in this space because I genuinely took these tests and remember how opaque they felt. When people ask me where to practise, I point them at PrepClubs, because it does the thing that actually moved my scores: timed drills by question type with the answer logic explained, rather than another wall of untimed questions. Start with the free material to find your weak spot, then decide whether systematic paid practice is worth it for the role you are chasing. That “free first, then pay only if it earns it” order is how I would approach any of these assessments, and it is the honest advice I would give a friend the night before their test.

You finished the Wonderlic, you have a two-digit number in front of you, and the only thing you actually want to know is simple: is this good or not? Nobody hands you a key. The test gives you a raw score out of 50 and leaves you to figure out whether 24 means you nailed it or barely scraped by.

I have been exactly where you are. I took the Wonderlic as part of a hiring process, and the moment the timer stopped I was already doing the thing you are probably doing now: staring at a raw number with no context and trying to reverse-engineer whether it was good. So I did what everyone does. I opened a browser and typed “what is a good Wonderlic score,” and I got a different answer on every page. One site said the average was 20. The next said 21. A third said 22 with total confidence. Half of them then tried to sell me a course before they had even told me what my number meant. So let me do for you what I wish those pages had done for me: give you the honest version, with the myths flagged, the role-by-role numbers laid out, and the percentile confusion cleared up.

The quick answer

A Wonderlic score of 21 or higher is above average, and anything in the high 20s or above is genuinely strong. The test has 50 questions and a 12-minute limit, one point per correct answer, no penalty for guessing, so scores run from 0 to 50. The working-adult average sits somewhere around 20 to 22 depending on whose sample you trust.

But “good” is not a single number. It depends entirely on the job. A warehouse role might expect a 15. A software or finance role might want a 30. There is no official pass mark on the Wonderlic. The employer sets the bar, and that bar moves with the role. So the useful question is not “what is a good score” in the abstract, it is “what is a good score for the job I am applying to.” I will get you both.

How the Wonderlic is scored

The Wonderlic Personnel Test, now often branded as the Wonderlic Contemporary Cognitive Ability Test, works like this:

• 50 multiple-choice questions.
• 12 minutes total. That is roughly 14 seconds per question, which is the entire point of the test.
• One point per correct answer. No points deducted for wrong answers.
• Final score is simply your number of correct answers, on a 0 to 50 scale.

Almost nobody finishes all 50. The clock is the real opponent, not the questions themselves, which mix verbal, numerical, and logic items that are individually not that hard. The moment that stuck with me was hitting a word-problem I knew I could solve, doing the mental math to realize it would eat 40 seconds, and physically making myself skip it because 40 seconds was three or four easier questions I would lose. That is the whole test in one decision. You are not really being asked “can you solve this,” you are being asked “can you tell, in two seconds, which questions to abandon.” That time pressure is why a score of 25 on the Wonderlic is more impressive than it looks on paper.

One thing worth knowing: there is also a shorter variant, the Wonderlic WPT-Q or QuickTest, which is 30 questions in 8 minutes and is faster for employers to administer. The scale is different, so if you took the short form, do not compare your number directly against the classic 0 to 50 benchmarks below. The “good score” numbers people quote, and the ones in this article, refer to the full 50-question test.

What counts as a good Wonderlic score

Here is the honest banding, drawn from how the score actually gets read and cross-checked against multiple sources. Treat these as directional, because different sources cite slightly different cutoffs:

Score	What it means
10 or below	Well below average. Struggling with the pace or the material.
11 to 20	Below to right around average.
21 to 25	Above average. A solid, employable result for most roles.
26 to 30	Strong. Top-decile territory, preferred for technical and professional roles.
31 to 39	Very strong. Well into the top few percent.
40 to 50	Exceptional and rare. A score of 40+ is roughly the top 1 percent, and 45 to 50 is genuinely uncommon (some estimates put a perfect 50 at around one in 30,000).

If you scored 21 or above, you did better than the average test-taker. If you cracked the high 20s, you should feel good about it regardless of what job you were sitting for. And if you are staring at a number in the teens, do not spiral: for plenty of roles that is exactly the range the employer expects.

A good Wonderlic score is not the highest possible number. It is a number that clears the bar for the specific role you want.

What is a good score for your job

This is the part the score chart alone never tells you. Employers benchmark the Wonderlic against the cognitive demands of the role, so the “good” line is different for a warehouse associate than for a systems analyst. Here is the pattern that shows up consistently across sources, again directional rather than gospel:

Role type	Typical target score
Manual / lower-complexity (warehouse, security guard, cashier)	~15 to 20
Skilled trades and clerical	~19 to 23
Service and mid-office (nurse, teacher, sales)	~22 to 26
Technical and professional (engineer, accountant, systems analyst)	~27 to 32
Highly analytical (chemist, senior analyst, some management)	~30+

If you want the specific job-title version rather than the tiers, here are the benchmark scores that show up most often across published Wonderlic norm tables. Again, these are approximate and vary by source, so read them as a target range, not a cutoff:

Job	Commonly cited target score
Security guard	~17
Warehouse / material handler	~15
Cashier	~21
Bank teller	~22
Nurse	~23
Salesperson	~24
Teacher	~24
Accountant	~28
Programmer / software engineer	~29
Electrical engineer	~30
Systems analyst	~32
Chemist	~31

So a 24 is a genuinely good score for a nursing or sales role, and only okay for an engineering seat where the benchmark sits near 30. Same number, two verdicts. When people ask me “is a 24 good,” my honest answer is “good for what job?” That is not a dodge, it is how this test actually gets used.

If your target feels far away, the good news is that the Wonderlic is very trainable, because so much of your score is pacing and pattern recognition rather than raw intelligence. What moved the number for me was practicing against a real clock with a full question bank, not untimed sample questions, and platforms like PrepClubs’ Wonderlic prep exist for exactly that, letting you rehearse the 12-minute pressure before it counts.

The percentile vs. raw score confusion (read this before you panic)

Here is the single thing that tripped me up, and it trips up almost everyone: your raw score and your percentile are not the same number, and most articles blur them together.

Your raw score is how many questions you got right, out of 50. Your percentile is what share of people you scored higher than. Related, but not interchangeable. A rough, commonly-cited mapping looks like this:

• A raw score around 20 lands near the 50th percentile (you beat about half of test-takers).
• Around 26 to 30 lands somewhere in the 80s to 90th percentile.
• Around 35+ approaches the 98th percentile, which is roughly Mensa-qualifying territory.
• 40+ is up near the 99th percentile.

These percentile mappings vary by source and by year, so treat them as approximate, not official. Wonderlic does not publish one universal, fixed percentile table that everyone shares. Different normative samples give slightly different numbers. Any page that hands you a precise, to-the-decimal percentile for every raw score is overselling its own certainty. So if your raw score is 26 and one site tells you that is the 67th percentile while another says the 80th, neither is lying, they are just reading off different norms. Focus on the band, not the decimal.

The IQ conversion, and why the “times five” rule is a trap

You will run into the claim that you can convert a Wonderlic score to an IQ by multiplying by five. Score a 24, so your “IQ” is 120. It is a tidy rule and it is repeated everywhere.

It is also only loosely true, and most pages just state the multiplier and move on. The times-five shortcut is roughly reasonable near the average, but the real relationship is not a clean straight line: it underestimates at the low end and overestimates at the high end. The Wonderlic is a speeded test taken under extreme time pressure, and a full IQ assessment is not, so equating the two too confidently is a mistake. Use times-five as a rough ballpark if you want, but do not go telling people your exact IQ off a 12-minute test. I almost did, then I looked into it, and I am glad I did not.

Why the “average” keeps changing depending on where you look

Remember how one page told me the average was 20 and another said 22? That contradiction is real. There is no single universal Wonderlic average, because it depends on which population you measure. Wonderlic’s own commonly-cited figure is around 20. Other reputable sources report a working-adult mean closer to 21 or 22, with a standard deviation of roughly 7. None of them is wrong. They are sampling different groups.

The practical takeaway: do not obsess over whether the “real” average is 20 or 22. The difference is inside the noise. Anything comfortably in the low-to-mid 20s puts you above the typical test-taker, and that is the bar most people are actually trying to clear.

The NFL Wonderlic story, and what it teaches you

You have probably heard of the Wonderlic because of the NFL. The league used it at the Scouting Combine from 1968 until 2022, when it was dropped in favor of an in-house assessment. Quarterbacks and offensive linemen historically scored highest as a group, with QB averages commonly cited in the mid-20s, while skill positions tended to score lower, though most positions clustered somewhere in the teens to mid-20s. Individual player numbers circulate online (you will see figures like Tom Brady around 33 and Patrick Mahomes around 24), but treat those specific numbers with a grain of salt, since leaked individual scores are frequently unconfirmed and disputed.

Here is the part I find genuinely useful. Studies looking at whether a quarterback’s Wonderlic score predicted his actual NFL performance mostly found little to no relationship. Some players with modest scores had long, excellent careers. The lesson that carries straight over to your own situation: the Wonderlic score matters for getting hired, not for whether you are good at the job. It is a screening filter, not a verdict on your worth. Clear the bar, get the interview, and then let the rest of you do the talking.

FAQ

What is the average Wonderlic score?

Roughly 20 to 22 for working adults, depending on the source and sample. Wonderlic’s own commonly-cited figure is around 20. Do not read too much into the exact number.

What is considered a good Wonderlic score?

Anything above 21 is above average. High 20s and up is strong. But “good” depends on the role, so check the target for your specific job in the table above.

Is there a pass or fail score on the Wonderlic?

No. There is no official cutoff. Each employer sets its own threshold based on the role, so the same score can pass for one job and fall short for another.

How is the Wonderlic scored?

One point per correct answer across 50 questions in 12 minutes, no penalty for wrong answers, giving a raw score from 0 to 50. Your score is simply how many you got right.

Is it better to guess or leave a question blank?

Guess. There is no penalty for a wrong answer, so a blank is a guaranteed zero on that question while a guess has a real chance of scoring. Never leave the last few questions empty when the clock runs out.

What is a good Wonderlic score for a software engineer or analyst?

Aim for around 30. Technical and analytical roles benchmark higher than the general average, typically in the high 20s to low 30s.

Does the NFL still use the Wonderlic?

No. The NFL used it at the Combine until 2022 and then replaced it with its own assessment.

Can I improve my Wonderlic score with practice?

Yes, meaningfully. Because so much of the score is pacing under time pressure, timed practice with a realistic question bank tends to move your number more than untimed studying does.

What I actually did about it, and what I would tell you

When I got my Wonderlic number, my mistake was treating it like a fixed verdict on my intelligence. It is not. It is a snapshot of how many trainable, time-pressured puzzles I could clear in 12 minutes. Once I understood the banding, checked my number against the role I was chasing, and stopped conflating my raw score with a percentile or an IQ, the whole thing got a lot less stressful.

If you have not sat the test yet, or you want to push your number up before it counts, the highest-leverage thing you can do is practice against a real clock. I keep pointing people toward PrepClubs for this because it runs Wonderlic prep as a proper timed platform with a large question bank rather than a handful of untimed samples, and it is free to start, which is how I think prep should work: find your weak spots for free first, then pay for systematic improvement if you need it. Rehearsing the 12-minute pressure is what actually moves the score.

For the fuller picture of the test itself, I have also written a complete Wonderlic test guide covering what to expect before you take it, and a rundown of the most common Wonderlic FAQs worth knowing. But if you only take one thing from this piece: a good Wonderlic score is the one that clears the bar for the job you want, and that bar is almost always more reachable than the number on your screen makes it feel.

I Took the PI Cognitive Assessment: What 12 Minutes of 50 Questions Actually Feels Like

If you just got an email asking you to sit the PI Cognitive Assessment before your next interview, you probably have two questions. What is actually on it, and how bad is 50 questions in 12 minutes going to feel. I had the same two questions the morning I took it, and most of what I read beforehand was written by people trying to sell me a prep course, not by someone who had recently sat in the chair and watched the timer.

So here is the version I wish I had found. I took the Predictive Index Cognitive Assessment, and this is what the test is, what the 12 minutes genuinely feel like, and the handful of things that would have changed my score if I had known them going in.

The quick answer: what the PI Cognitive Assessment actually is

The PI Cognitive Assessment is a timed, 50 question aptitude test built by The Predictive Index. You get 12 minutes total. That works out to roughly 14 seconds per question if you wanted to answer every single one, which almost nobody does.

It is not an IQ test and it is not a knowledge test. There is no trivia, no reading you were supposed to have done, nothing role specific. It measures how quickly you can pick up a pattern, do a small piece of reasoning, and move on. Employers use it because general cognitive ability is one of the more reliable predictors of how fast someone ramps into a new role, so it tends to show up early in hiring for analyst, sales, operations, and graduate style positions.

Here is the single most important thing to internalize before you start: you are not expected to finish all 50 questions, and finishing is not the goal. Only about 1 percent of test takers correctly answer more than 40. The test is designed so that the clock beats almost everyone. Once I understood that, the whole thing got less intimidating.

What is on the test: the three question types

The 50 questions are drawn from three reasoning categories, mixed together rather than grouped into neat sections. You will bounce between them.

Numerical reasoning. Number series where you work out what comes next in a sequence, short word problems that are really just workplace flavored arithmetic, and value comparison questions where you decide which of a few fractions or figures is largest or smallest. No calculator in most sittings, so you are doing this in your head or on scratch paper if you are allowed any.

Verbal reasoning. Antonyms, analogies, and short logical or deductive statements where you decide what must be true. These felt like the fastest points for me because they do not require any calculation, just quick reading.

Abstract reasoning. Visual pattern series. You get a row of shapes that change according to some rule and you pick the next one. These are the questions that either click in two seconds or eat thirty seconds you do not have.

Each question has four answer options, except a few verbal ones that only have three. You can move backward and forward between screens, so a question you skip is not lost, you can circle back if time allows.

Here is roughly how the format breaks down.

Feature	Detail
Questions	50
Time limit	12 minutes
Average time per question	About 14 seconds
Categories	Numerical, verbal, abstract reasoning
Answer options	4 per question (3 on some verbal)
Calculator	Usually not allowed
Guessing penalty	None
Raw score	Number correct out of 50
Scaled score range	100 to 450

What 12 minutes actually feels like

The number on paper is 12 minutes. What it feels like is a countdown that starts faster than you expect and never lets up.

The first two minutes felt fine. The questions were easy, my confidence was high, and I remember thinking this was going to be manageable. Then I hit an abstract pattern I could not read, spent what felt like a moment on it, glanced at the timer, and a chunk of my time was simply gone. That is the trap. The questions are individually not that hard. The difficulty is entirely in the pacing, and the pacing punishes the exact instinct that makes you good at your job, which is to keep working a problem until you crack it.

The interface shows you the time remaining in minutes, not seconds, which somehow makes it worse. You do not get a precise countdown, you get a blunt reminder that another minute is gone. Around the halfway mark I made a decision that I think saved my score: I stopped trying to be right and started trying to be efficient. If a question did not resolve in my head within a few seconds, I picked the answer I leaned toward, marked it mentally as a guess, and moved on.

The people who struggle most on the PI are not the ones who are bad at the questions. They are the ones who refuse to leave a question unsolved, and the clock quietly bankrupts them one stubborn problem at a time.

I did not finish. I got somewhere in the mid 30s for questions attempted, and I made peace with the ones I never saw. That is a normal, even good, outcome.

How the scoring works, and what a good score is

Your raw score is simply the number of questions you answered correctly. That raw score gets converted to a scaled score somewhere between 100 and 450, which is then compared against a large norm group of past test takers to produce a percentile.

The commonly cited benchmark is that a scaled score around 250, which is roughly 20 correct answers, sits near the average, around the 50th percentile. A scaled score of about 320, which is roughly 27 correct, moves you into the top 20 percent of candidates. Push toward a scaled score in the 350 range and you are into genuinely high performer territory that few candidates reach. Many professional roles look for something in the mid 20s of correct answers and up, though the truth is that the target depends heavily on the specific role and employer. A pattern heavy analyst job and a relationship heavy sales job are not going to weight the same score the same way.

If you want the full breakdown of how the cognitive score sits alongside the behavioral half of the Predictive Index and how employers actually read the two together, this detailed look at how the Predictive Index cognitive and behavioral results are scored lays out the bands and cutoffs more thoroughly than the vendor pages tend to. Understanding that a score is a percentile against a norm group, not a pass or fail line, took a lot of the pressure off for me.

One more scoring detail that changes how you should play the test: there is no penalty for guessing. A blank and a wrong answer cost you exactly the same amount, which is nothing beyond the point you did not earn. So a blank is strictly worse than a guess. Which leads directly to the next section.

The things I wish I had known before I started

None of these are secrets, and none of them require weeks of prep. They are just the handful of adjustments that would have earned me more correct answers on the day.

Guess on everything, always. With no wrong answer penalty, the correct strategy is to never leave a question blank. In the last 20 seconds, if you have unanswered questions, mark an answer for every one of them even if you are choosing at random. On a four option question, random guessing still lands you correct answers over enough tries. Leaving them blank guarantees zero.

Set a hard personal timer per question. My rule after the fact would be simple: if a question is not resolving within about 20 to 30 seconds, take my best guess and move. The single biggest score killer is sinking a minute into one stubborn abstract pattern. That minute could have been four other questions.

Do the fast points first. The verbal questions, the antonyms and analogies, took me the least time per point. If you find one category comes easily, lean into it and do not let a slow category rob you of the easy points sitting later in the test.

Practice the format, not the content. You cannot really study cognitive ability, but you can absolutely remove the surprise. The reason the number series and value comparison questions cost me time was that I was seeing the format for the first time under a clock. A few timed practice runs beforehand would have made those feel routine. If you have any lead time at all, a short set of timed practice questions across all three categories is the highest return thing you can do, mostly because it trains the guess and move instinct that the real test demands.

Take it in your strongest language if offered. Reading speed directly controls how many questions you get through, so if there is a language option, pick the one you read fastest in.

A quick note on the behavioral half

If your employer uses the full Predictive Index, you will likely also get the PI Behavioral Assessment. That one is completely different in feel. It is untimed, it is a two part word choice exercise about how you naturally work, and there are no right or wrong answers. Do not confuse the two. The cognitive test is the one with the clock. The behavioral one you should just answer honestly, because it is trying to map your working style, not grade you.

FAQ

How hard is the PI Cognitive Assessment?

The individual questions are not hard. The difficulty is the time limit. 50 questions in 12 minutes means almost nobody finishes, and the test is built that way on purpose. If you go in expecting to leave questions unanswered, it feels a lot more manageable.

Am I supposed to answer all 50 questions?

No. Only about 1 percent of people correctly answer more than 40. You are meant to get through as many as you accurately can, not to complete every one.

What is a good PI Cognitive score?

Roughly 20 correct is around average. About 27 correct pushes you into the top 20 percent. The right target genuinely depends on the role and employer, since it is scored as a percentile against a norm group rather than a fixed pass mark.

Can I use a calculator?

Usually not. Most sittings do not allow one. You may be permitted scratch paper depending on the instructions you are given, so read them before you start.

Should I guess?

Yes, on everything. There is no penalty for a wrong answer, so a blank is always worse than a guess. Fill in every question before time runs out.

How should I prepare in a short amount of time?

Do a few timed practice sets that cover numerical, verbal, and abstract questions. You are not trying to learn content, you are trying to make the format familiar and to rehearse guessing and moving on so the clock does not surprise you.

Where I landed on it

I spend most of my time building software as a founder, and I have taken more of these pre employment cognitive tests than I would like to admit, from the CCAT and Wonderlic to the PI and beyond. The pattern is always the same. The people who do best are not the smartest in the room, they are the ones who accepted the clock and optimized for volume of correct answers instead of perfection.

That realization is also why I care about how people prepare for these. My honest advice is to start free. Get a feel for the question types and time yourself before you spend a cent, because a lot of what passes for prep is an overpriced version of what a free timed practice set already teaches you. When you do want structured, systematic practice across the full set of pre employment tests, including a dedicated PI track, that is where a platform like PrepClubs earns its place, because it puts the PI alongside the CCAT, the Wonderlic, and the rest of the tests employers actually use, with the biggest question bank I have found for practicing under real time pressure. Free first to find your weak spots, then paid to fix them systematically, is the order I would follow again.

The PI is not a test you can cram for the night before and transform your raw ability. But it is absolutely a test where knowing the format, respecting the clock, and guessing without guilt can move your score by a meaningful margin. Walk in expecting the timer to win, play for correct answers rather than a clean sweep, and you will do better than the version of you that walked in expecting to finish.

You have already decided that built-in dictation is not enough. You want to talk and get finished text back, and the two names that keep coming up are Wispr Flow and Superwhisper. So you do the sensible thing and read a comparison, except every comparison you find is either a vendor page that happens to prefer itself or a rival dictation app that happens to crown itself the third option nobody asked about. What you actually want is simpler: which one survives a real month of real work.

So I ran both for a month. Not a benchmark, not five neat rounds scored in an afternoon. Thirty days of using each as my only dictation tool on the days it was assigned, across the actual work I do as a founder: investor emails, Slack replies, product requirements docs, code comments, and the messy voice notes I leave myself at 11pm. This is what the month taught me, where each one earned its place, and where each one quietly cost me time. I paid for both. I sell neither, which matters, because near the end I do name a tool I built, and you should be able to trust everything before that.

The one-line answer, if you are in a hurry

Wispr Flow is the polished cloud tool that works everywhere. Superwhisper is the private local tool that lives on a Mac. If you bounce between a Mac, a PC, and a phone and you want zero fuss, Wispr Flow. If you handle sensitive work, live inside the Apple ecosystem, and enjoy configuring a tool to your taste, Superwhisper. Almost every difference in the month came down to one decision each team made: where your audio gets processed.

Here is the whole thing at a glance before I walk you through the month:

	Wispr Flow	Superwhisper
Where it runs	Cloud only	Local on Apple Silicon, cloud optional
Works offline	No	Yes (local models)
Free tier	2,000 words/week (desktop)	Does not expire, smaller local models
Paid	$15/mo, or $12/mo billed annually	$8.49/mo, $84.99/year, or $249.99 lifetime
Platforms	Mac, Windows, iOS, Android	Mac, iOS, Windows (no Android)
Custom vocabulary	Learns your terms automatically	Manual, per-mode, more control
Best for	Polish and cross-platform reach	Privacy, offline, one-time cost

Now the month.

Week one: the setup gap is real, and it decides your first impression

I started the month on Wispr Flow, and it was working before my coffee cooled. Install, grant a permission, pick the hotkey it suggests, talk. The onboarding is the smoothest in the category and the defaults are sane. Within ten minutes I was dictating a client email that needed no cleanup. This is the experience you would hand a non-technical teammate and walk away from.

Then I switched to Superwhisper, and week one felt like assembling furniture. It does not hand you a finished tool. It hands you a system: local models to download and choose between, optional cloud models to wire up with your own API key, and custom “modes” you configure for email versus code versus notes. That is the entire appeal for a power user, but the first hour is spent managing the tool rather than using it, and the larger local models took a real several seconds to spin up the first time each session.

Wispr Flow wins the first day. Superwhisper makes you earn the second week. If your patience for setup is thin, that gap alone may decide it.

The moment the free tier ran dry

Here is something the spec-sheet comparisons never tell you, because you only learn it by living in the tool. Wispr Flow’s free plan gives you 2,000 words a week on desktop. That sounds generous until a heavy dictation day. I hit the wall on a Tuesday afternoon, mid-draft on a long product doc, and the tool politely stopped. On a real workday where dictation is the point, 2,000 words a week is roughly one good morning.

Superwhisper’s free tier works differently: it does not expire, and it runs the smaller local models indefinitely, so you can dictate all day without a meter. The catch is that the free local models are the smaller, less accurate ones, so you are trading a word cap for a quality floor. Neither free tier is a place you can actually live long term, but they run out on you in completely different ways, and knowing which failure you can tolerate is more useful than a feature checklist.

Output quality: polished-by-default versus tuned-if-you-bother

Both tools do the thing that makes them AI dictation rather than plain transcription. You ramble, and they hand back something punctuated, de-ummed, and shaped. This is the whole category. Raw transcription is a solved, boring problem.

Wispr Flow is excellent out of the box. It takes a filler-heavy thought and returns tidy prose, and its habit of adapting tone to the app you are writing in is genuinely good: a Slack message comes back casual, an email comes back composed. For everyday writing I rarely edited it.

Superwhisper can match that and, in narrow cases, beat it, because you control the model and the cleanup prompt behind each mode. If you set up a mode that runs a strong cloud model against your own instructions, the output is tuned exactly to your taste. The trade is effort: out of the raw local models, Superwhisper leaves more of your thinking-out-loud in the text, so the “um, wait, no, send it to the whole team” survives into the transcript more often, and you either edit it or configure a cleanup layer to fix it. One person who measured this carefully, developer Zack Proser, clocked Wispr Flow near the top for words-per-minute with its AI cleanup on and Superwhisper lower on raw output, and I want to be clear that is his measurement, not a lab result I ran. It matched my felt experience, though: Wispr Flow needed less editing more often.

Custom vocabulary is where this splits along the same line as everything else. My work is full of product names, teammates’ names, and jargon that generic models mangle, and here Wispr Flow leans on convenience while Superwhisper leans on control. Wispr Flow picks up your recurring terms and corrects them for you over time with almost no setup. Superwhisper makes you add vocabulary and replacements per mode by hand, which is more work up front but means you know exactly what it will and will not touch. If proper nouns are a big part of your dictation, that difference is worth more than a raw accuracy score.

One angle worth flagging even though it was not my use case: if you dictate inside a developer workflow, this comparison has a whole extra dimension. Some tools in this space now plug into coding agents and command-line environments, and both Wispr Flow and Superwhisper get discussed for exactly that, dictating into an editor or driving an AI coding assistant by voice. I did not lean on that heavily as a founder who splits time across marketing and code, but if your day is mostly in an IDE, weigh how each one behaves inside your editor before you decide.

The flight that ended the debate for me

Three weeks in, I was on a plane with no wifi, trying to draft a long doc from voice notes. Wispr Flow is cloud-only. There is no offline mode at any price, so with no connection it simply does not work. Dead. Superwhisper, running a local model on my Mac, kept transcribing at 35,000 feet as if nothing had changed.

That single afternoon reframed the whole comparison for me. Everything else is a preference. This is a capability. If any meaningful share of your dictation happens on planes, in bad hotel wifi, in a secure building, or anywhere the network is unreliable, Wispr Flow’s cloud-only design is not a minor footnote, it is a wall you will hit. And it flows directly from where each tool processes your voice, which is the same decision that drives the privacy question.

Privacy: local by default versus cloud by design

Wispr Flow sends every word you dictate to a server, processes it there, and sends it back polished. It offers a zero-retention Privacy Mode and, on Enterprise, SOC 2 Type II and HIPAA, and to be fair Superwhisper carries the same certifications. But “we do not keep it” is a different promise from “it never left your machine,” and the two are not interchangeable if your work is confidential.

Superwhisper is the privacy story here. On Apple Silicon, Whisper-family transcription runs on-device, so your audio stays local. That is a real, structural advantage that Wispr Flow’s architecture cannot match. Read the defaults, though, because they are not clean: Superwhisper saves your audio recordings by default until you turn that off, and the instant you switch to a cloud model for higher quality, your transcript goes to that provider like anyone else’s. The privacy is genuine but conditional. You get it by staying on local models and changing the audio-saving default, not for free.

On-device beats cloud-only for privacy, full stop. Just know that Superwhisper’s cloud modes quietly break the promise, and its defaults do not protect you until you change them.

Platforms: one of these follows you everywhere

Wispr Flow runs on macOS, Windows, iPhone, and Android off a single account, the broadest reach in this matchup. The honest footnotes: the Windows build is a heavier app that some users report can freeze the program they are dictating into, and Android is still filling in features. But if you live across a Mac, a PC, and a phone, Wispr Flow is the only one of the two that even tries to be everywhere.

Superwhisper is Mac-first and unapologetic about it. There is a solid iPhone app, and Windows support now exists where it did not before, but it trails the Mac experience, and there is no Android at all. On a Mac, Superwhisper is superb. Off a Mac, it is either an afterthought or absent. If you had told me a year ago Superwhisper was Mac-only I would have agreed, but that framing is now outdated: Mac and iOS are the home, with Windows added and catching up.

Pricing and the money question over time

This is where a month of use turns into a real number, so let me give you the verified 2026 pricing and then the judgment.

	Wispr Flow	Superwhisper
Free tier	2,000 words/week (desktop)	Does not expire, smaller local models
Subscription	$15/mo, or $12/mo billed annually	$8.49/mo, or $84.99/year
Lifetime option	None	$249.99 one-time
Where it runs	Cloud only	Local on Apple Silicon, cloud optional
Offline mode	No	Yes (local models)
Platforms	Mac, Windows, iOS, Android	Mac, iOS, Windows (no Android)
Best for	Polish and cross-platform reach	Privacy, offline, one-time cost

Month to month, Superwhisper is cheaper, and its lifetime license changes the math entirely if you plan to stick with it. At $84.99 a year against a $249.99 one-time license, the lifetime pays for itself somewhere around year three, and everything after that is free. Wispr Flow has no lifetime tier, so at $12 to $15 a month the meter never stops. As a founder watching burn, that lifetime break-even is the kind of decision I actually like: pay once, own it, stop thinking about it. The counterweight is that a lifetime license only pays off if the tool keeps up with your needs for three-plus years, and dictation is moving fast enough that betting three years on any single tool is its own small gamble.

The trust thing nobody puts on a feature page

One softer note, because it shaped how I felt about each tool more than any single feature. Wispr Flow was involved in a controversy over capturing active-window context, which it walked back to opt-in after a public apology, and a recurring theme in community threads is people feeling the paid product delivered a little less magic than the free trial. I cannot verify that as a defect and your mileage will vary, but it is a real sentiment worth knowing. Superwhisper’s version is quieter and more technical: the audio-saving default and reports of API keys stored in plaintext do not bother a casual user but should give pause to anyone handling sensitive work. Both are trustworthy enough for ordinary use. Both have one specific thing to check before you hand them your most confidential dictation.

The gap the whole month kept pointing at

Look back at what actually happened over thirty days. Every time I preferred one tool, I gave something up. Wispr Flow’s polish cost me offline access and local privacy. Superwhisper’s privacy and offline resilience cost me cross-platform reach and a frictionless setup. Polish or privacy. Reach or local processing. Simplicity or control. I kept having to pick a corner.

That tradeoff is not a law of nature. It is just where these two happen to sit on the map. And the fact that I kept bumping into the same missing combination, polished and private and cross-platform at once, is the reason my team and I built our own dictation tool, Contextli. So read this next part as the founder pitch it is, and check the claims yourself against your own workflow.

The short case: on the privacy question the flight and the confidential-work sections kept raising, Contextli gives you three modes instead of the cloud-or-Mac choice. Cloud when you want speed, bring-your-own-key so your audio goes from your machine straight to your own provider account, or fully offline where transcription and the AI rewriting both run locally and nothing leaves the device. That offline mode runs on Windows and Mac, not Apple Silicon only, which is the exact line Superwhisper cannot cross. On platforms it matches Wispr Flow’s reach across Windows, Mac, iOS, and Android, but the offline mode comes along for the ride, and the screen-context feature Wispr Flow got burned on is off by default. I will not pretend it beats Wispr Flow’s years of polish or Superwhisper’s customization depth. But on the specific compromise this comparison forces on you, Contextli is the option that refuses to make you choose. Try the free tier against a real workday before you take my word for any of it.

How to actually choose

If you have read this far, here is the decision in plain terms.

Pick Wispr Flow if a frictionless experience matters most, you want it on every device including Windows and Android, and your work is not sensitive enough for cloud-only to be a problem. Pick Superwhisper if you are a Mac user who values on-device privacy and offline resilience, you enjoy configuring a tool to your taste, and you can live without Android and remember to switch off the audio-saving default. If dictation is central to your day, the honest move is to run each free tier for a few days on your own real work, because the setup gap, the free-tier wall, and the offline question hit differently depending on how and where you actually work.

If you want this same matchup as a scored, round-by-round breakdown with a scoreboard instead of a diary, I wrote that version up separately in my Wispr Flow vs Superwhisper head-to-head. And if Wispr Flow is the one you are leaning toward, my full Wispr Flow review after a month of solo use goes deeper on the tool on its own.

FAQ

What is the difference between Superwhisper and Wispr Flow?

The core difference is where your audio gets processed. Superwhisper can run transcription locally on-device on Apple Silicon, so your audio never leaves your machine and it works offline. Wispr Flow is cloud-only: every word travels to a server, gets AI-cleaned, and comes back, which means no offline mode but broader platform reach including Android. Pricing differs too, with Superwhisper offering a lifetime license and Wispr Flow being subscription-only.

What is better than Wispr Flow?

It depends on the axis. For privacy, offline use, and a one-time cost, Superwhisper is the stronger pick. For an out-of-the-box experience with the least fuss and the widest platform support, Wispr Flow is hard to beat. There is no universal answer, which is exactly why a versus comparison is more useful than a single winner.

Is Wispr Flow the best dictation app?

Wispr Flow is the best for cross-platform, plug-and-play use with strong AI cleanup, and for many people that makes it the best overall. It is not the best if you need offline dictation, on-device privacy, or you specifically dislike subscriptions, since it has no lifetime option and no local mode.

Does Superwhisper work offline?

Yes, that is one of its defining strengths. On Apple Silicon, Superwhisper runs its local Whisper-family models entirely on-device, so it keeps transcribing with no internet connection, which Wispr Flow cannot do at all. Just note that if you switch Superwhisper to a cloud model for higher quality, you lose that offline benefit for those requests.

Is dictation actually faster than typing?

For most people, clearly yes. Typing averages around 40 words a minute, while research on speech input has measured it at roughly three times that speed with fewer errors once you stop self-editing. Both Wispr Flow and Superwhisper are fast enough that raw speed is not the deciding factor; privacy, platforms, and polish are.

The bottom line

A month with both comes down to one question: do you want polish and reach, or privacy and control? Wispr Flow takes setup, platforms, and out-of-the-box quality. Superwhisper takes privacy, offline resilience, and long-term cost, if you are on a Mac and willing to tune it. There is no universal winner, only the right fit for how and where you actually work.

The reason the month kept ending in tradeoffs is that these two sit at opposite corners of the same map. If you would rather not pick a corner, that is the exact problem I built Contextli to solve: polish and privacy and cross-platform reach, with an offline mode that runs anywhere. Talk one messy sentence into the free tier and see whether you still feel like compromising.

Pricing and features are accurate as of mid-2026 and change often, so verify on each official page before buying.