Recruitment assessment tools test every candidate the same way, so you compare like with like. We checked 28 of them. Eight publish a price. Below are the real numbers, the pricing unit that catches buyers out, and what each type of test actually predicts.
What does good look like for this role?
Fit scores every respondent against a target you set once. It takes about a minute, and you can change it later without re-inviting anyone.
Skip this and every dimension starts at the midpoint. You can set it any time.
Free · No signup · Live link in about 10 seconds
Every figure below came from the vendor's own pricing page on 17 August 2026. Nothing here is quoted from another comparison post. Sorted by what a small employer pays to start, not by preference.
| Tool | What it measures | How it bills you | Published price | Cheapest way in |
|---|---|---|---|---|
| SeeMyPersonality (ours) | Personality, worded for work | Monthly, by scored respondents | $99/mo, or $948 a year | Build and send free on this page, no account |
| TestGorilla | Skills, plus some personality | Credits, burned per candidate action | Free tier · Core $1,704/yr · Plus from $4,800/yr | Free plan, 10 credits a month, no card |
| TestDome | Technical skills | Per candidate, packs never expire | $20 each for 5, falling to $7 each at 600 | $100 for five candidates |
| Bryq | Personality and cognitive | Flat monthly, banded by your headcount | Go $69/mo · Pro $168/mo, billed yearly | $828 a year, one-year commitment |
| Xobin | Skills, psychometric, AI interview | Credit bundle. One credit runs 30 assessments | $1,000 per 100 credits | Free 5 credits for 14 days |
| HackerRank | Technical skills | Subscription plus per-attempt overage | Starter $1,990/yr · Pro $4,490/yr · $20 an attempt over first-party, but taken from their comparison articles rather than the pricing table | Starter at $1,990 a year |
| Prevue HR | Personality and ability | Flat monthly, banded by headcount | Essentials $195/mo for 1 to 25 staff | Free trial, terms not published |
| McQuaig | Personality | 90-day term, not annual | From $1,500 per 90 days | $1,500, which is roughly $6,000 a year if you hire all year |
| Clevry | Personality, ability, situational judgement | Per test | Personality £135 · ability £40 · SJT £40 · £500 onboarding | Pay as you go, after the £500 onboarding fee |
| Alva Labs | Personality and logic | Per open job per year, not per candidate | Extra positions €370, or €320 on the larger plan base subscription fee loads in the browser and never rendered | 30-day trial, one job, up to 50 candidates |
| Predictive Index | Personality and cognitive | Annual licence, priced by company headcount | From $10,000 a year | $10,000 floor, annual only |
Two of the names you will meet in older comparison posts are gone. Plum was bought by Phenom and no longer sells separately. Berke is now HighMatch, and HighMatch says plainly that it does not offer fixed pricing. If a list still recommends either as a current option, it has not been checked recently.
Headline prices are easy to compare and rarely what you pay. What decides your bill is the unit, and there are three that catch small buyers out.
Credits that expire. TestGorilla's free plan gives you ten credits a month and they do not bank. Hiring in bursts means buying capacity you will lose.
Minimum buys. Xobin has the most generous rate here: one credit runs thirty assessments, so fifty candidates costs about two credits. The catch is you cannot buy two. The floor is a hundred, at $1,000, so you spend a thousand dollars to use twenty of it.
Headcount pricing. Bryq, Thomas and Predictive Index price on how many people you employ, not how many you assess. A forty-person company hiring five a year pays the same as one hiring two hundred. If your hiring is light relative to your size, this is the unit that hurts most.
The number that settles it is cost per candidate, and almost nobody prints it. We built a calculator that works it out from published rates, on our TestGorilla alternatives page. It is the same maths whichever tool you are weighing.
Operational validity for predicting job performance, on the corrected 2022 estimates[1]. Higher is better. The group gap is the Black-White difference in applicant samples, and it is where legal challenge lands.
| Type | Predicts | Group gap | What that means |
|---|---|---|---|
| Structured interview | .42 | .23 | Not a tool you buy. The strongest method there is, and you already do interviews. |
| Job knowledge test | .40 | .54 | Strong where the role has a real body of knowledge to know. |
| Skills test / work sample | .33 | .67 | Fell furthest in the 2022 correction, from .54. Still useful, no longer top. |
| Cognitive ability test | .31 | .79 | Was .51. Carries by far the widest group gap, which is where challenges start. |
| Personality, worded for work (ours) | .25 | −.07 | The most consistent predictor in the table, and almost no group gap. |
| Personality, worded generally | .19 | −.07 | The same questionnaire without "at work" in the items. A third less signal. |
Most vendor science pages still quote the older figures, where cognitive tests scored .51 and work samples .54. Those were corrected downward in 2022, and we tell the longer version of that story on TestGorilla alternatives.
Read the table as a buyer and two things stand out. The strongest method is the interview you already run, so the best thing a tool can do is make that interview structured. And the two rows with almost no group gap are the personality rows, which is the opposite of what most people assume about fairness in testing.
Since we sell one of these, here is our own row read honestly. A questionnaire worded for work predicts at .25, which is below a skills test and well below a structured interview. What the single number hides is consistency: it has the narrowest spread of anything in the table[2], so it is the least likely to disappoint you on a given hire. It also carries essentially no group gap. We would still not use it to rank your final three. We would use it to decide what to ask them.
Hiring a handful of technical people. Buy per candidate. TestDome at $16 to $20 a head beats every subscription until you are past roughly a hundred candidates a year, and the packs do not expire.
Hiring across a lot of roles. A platform earns its keep. TestGorilla's free tier covers a genuine trial, and Xobin's rate is the best in the table if you can use the volume.
You want to understand how people work, not just what they can do. That is the personality category. Word the items for work, keep the result out of the reject decision, and use it to drive the interview. Ours is free to build and send, and there is one on this page.
You are buying for a large organisation. Register for SHL or Talogy and get the real price list, or expect a demo cycle. Predictive Index publishes a $10,000 floor, which is useful mostly as a signal about who the product is for.
Whatever you choose, the cheapest improvement is not on this page. Write your interview questions down, ask every candidate the same ones, and score as you go. That change is worth more than any tool here, and it costs nothing.
They are software products that test candidates in a consistent way: skills tests, personality and ability questionnaires, situational judgement tests, and video or chat interviews that get scored. The point of all of them is the same. Everyone at a given stage answers the same thing, so you compare like with like instead of comparing your impressions.
Of 28 vendors we checked in August 2026, eight publish a price. Among those, a small employer testing 50 candidates a year pays anywhere from nothing to about $10,000. TestDome charges per candidate from $7 to $20. Bryq starts at $828 a year. Predictive Index starts at $10,000 a year regardless of how many people you assess. Most of the rest ask you to book a demo.
It depends on what you are testing and how you are billed, and those two questions matter more than any ranking. For technical roles at low volume, per-candidate pricing wins. For a whole team on a subscription, headcount-banded plans are usually cheaper. For anything where the result affects who gets rejected, pick the type with the evidence behind it and the smallest group gap, then put your real weight on a structured interview.
Accurate enough to help, and not accurate enough to decide alone. The best single method is a structured interview, at about .42. Skills tests come in near .33 and cognitive tests near .31, both lower than the figures most vendor pages still quote. A well-built questionnaire adds about .25 when its items are worded for work. Combine two methods that measure different things and you do genuinely well; stack five and you mostly lose candidates.
Yes, with conditions. The test must not put people with protected characteristics at a disadvantage you cannot justify under the Equality Act 2010, and you must offer reasonable adjustments for disabled candidates. Ask "do you need any adjustments to complete this assessment?" rather than asking about health, which section 60 restricts before an offer. Keep a person in the decision rather than rejecting automatically. None of this is legal advice.
Yes, as one input among several, and not as the thing that produces the rejection on its own. Word the items for work rather than in general, because that difference alone is worth about a third more signal. Use it to shape what you probe in the interview. What it should not do is rank your final shortlist, because that is the part of the scale where candidates present themselves best.
If you want to try one rather than read about them, the builder at the top of the page makes a real assessment and gives you a link to send. No account, and no demo call.
Two prices carry a caveat, stated in the table rather than hidden here. HackerRank's figures are HackerRank's own but come from their comparison articles, because their pricing table renders in the browser and would not return numbers. Alva Labs' base subscription fee does the same, so only their per-position rates are shown. Codility's pricing page returned a server error on every attempt, so no Codility price appears at all.
Validity figures come from the 2022 re-analysis of the selection literature[1] and its 2023 follow-up[2], not from the 1998 table most vendor pages still use. They are averages across many jobs with real spread around them, so treat them as guidance on where to spend effort rather than a prediction about your next hire.