TestGorilla alternatives: what they cost, and what they predict

Start with the plans as published. The prices below are the US dollar view.[1] That qualifier matters more than it should, and the next table explains why.

Three things the price alone does not tell you

It is an annual contract, shown as a monthly rate. TestGorilla's own billing FAQ states that Core and Plus are annual upfront plans with no monthly payment option, and that there is no free trial.[1] The page still leads with "$142/mo". Nearly every billing complaint in the review sites traces back to that gap.

Credits expire. One credit covers one candidate taking an assessment of up to five skills tests, which is generous per candidate. But unused credits do not roll over at the end of the contract period on any plan, free or paid.[2] If your hiring is uneven, you are buying capacity you will lose.

Integration is a tier away. ATS integrations and the API are on Plus, which starts at $4,800 a year.[1] Core gives you two seats that can spend credits. Hiring managers can view for free, but a third recruiter cannot run anything.

The same plan costs a different amount, and includes a different bundle

The pricing page carries a currency switcher, and it does not simply convert. The credit allowance changes too.

What everyone else says it costs

This is worth showing, because if you have spent the morning reading comparison pages you have been given at least six different prices. Prices do move, and pages go stale. Here is what each said on the day we checked.

We will be wrong eventually too. TestGorilla has changed pricing model four times in about three years, most recently overhauling credits in January 2026 and retiring the last legacy plans in May. The date on this section is the only claim we can really stand behind, which is why it is there.

At full utilisation, $1,704 across 400 credits is $4.26 a candidate, which is a good price. Almost nobody runs at full utilisation. A company hiring two roles a year with thirty candidates each uses sixty credits and loses the other 340.

Move the sliders to your own numbers. The comparison underneath uses published list prices only.

That spread is the most useful thing on any review site here. G2 and Capterra are where employers go. Trustpilot is where the people who were made to sit the test go. A page quoting only the 4.5 is telling you half the story, and it is the half that does not have to take the assessment.

Stay if you hire steadily and at volume. Between roughly 250 and 400 candidates a year, on two recruiter seats, without an ATS integration, Core works out near $5 a candidate for a library of 381 tests.[3] Very little in this market beats that, and switching would cost you more in disruption than you would save.

Stay if you need breadth more than depth. Their own category counts come to 162 role-specific tests, 83 programming, 59 software, 37 language, 17 cognitive and 13 situational judgment.[3] If you hire across many different roles, that breadth is genuinely hard to replace.

Stay if proctoring matters and budget is tight. This is the thing competitor pages least like to mention: TestGorilla's anti-cheat features are available on every plan, including the free one. Most rivals put proctoring behind a paid tier.

Stay if you are already integrated and mid-contract. The cost of moving assessment history, retraining a panel and rebuilding scorecards is real, and it rarely appears in anyone's comparison table.

Credit where it is due on one more point. TestGorilla publishes per-test reliability figures with sample sizes, which is more than most skills-first vendors do at all. Their abstract reasoning test reports a Cronbach's alpha of .89 on 617 people.[3] The limits of that disclosure are covered in the fairness section, but the disclosure itself is above average for this category.

We make one of these, and it is in the table in its place rather than at the top. On published prices at low volume, Equip and TestDome undercut us and nearly everyone else.[18] If cost per candidate is your binding constraint and you only need skills screening, start there rather than with us.

TestGorilla's library is 381 tests by its own category counts. Six of them are personality and culture. That ratio is not a criticism, it is a description of what the product is for: it is a skills library with a small behavioural corner attached, and it is sold to people whose problem is "can this person do the work".

If your problem is that people you hire can clearly do the work and then leave, clash or stall, a bigger skills library will not touch it. That is a different measurement.

What the behavioural side honestly adds, and what it does not

We should be careful here, because we sell a personality instrument and the evidence does not support overclaiming. Once you already have a structured interview, biodata and an integrity check in place, a general conscientiousness measure contributes about 3% of the predictable variance in job performance.[6] Anyone telling you a personality test is the main event is selling.

Where it earns its place is narrower and more useful. Personality measures predict counterproductive behaviour, which a skills test cannot see at all: agreeableness is the strongest predictor of interpersonal deviance and conscientiousness of organisational deviance.[9] And how you ask matters. Conscientiousness measured with work-framed items reaches .25 with essentially no variation across studies, against .19 for generic wording.[4][8]

There is also a live disagreement worth knowing about. In 2007 six former journal editors argued that self-report personality tests should be abandoned for selection outright, and were answered in the same issue.[12] That argument has not been settled. We think the honest position is that a personality measure is a useful second input and a poor first one.

If you want to see what the behavioural read looks like in practice rather than take our word for it, there is a complete employee personality assessment sample that computes in your browser, and an instrument-by-instrument comparison in best personality tests for hiring, compared.

The 1998 estimates everyone quotes came from Schmidt and Hunter, who applied a range-restriction correction uniformly across studies.[7] Most of those studies did not warrant it. In 2022 Sackett, Zhang, Berry and Lievens redid the analysis and published a corrected matrix.[5][4]

Two things follow. Structured interviews, not tests, are now the strongest single method. And the largest downward revision of all landed on work samples, which is the category most of this page's alternatives sell. Codility, to their credit, cite the corrected .33 on their own validity page rather than the flattering older number.[19]

Grade A means a meta-analysis or a replicated finding. Grade B means one strong study. Score gap is the standardised Black and White difference on the method itself, which the next section is about.

Do not read this as "assessment does not work". Sensible combinations of several methods still average about .47, barely down from .51 under the old estimates.[4] The correction hits single-method marketing claims, not the practice.

One more finding worth the detour, because it is probably the cheapest improvement available to most buyers. In a study where people predicted students' grades, adding an unstructured interview made their predictions worse than using prior grades alone. In one condition the interviewee answered at random. Interviewers did not notice, and formed confident impressions anyway.[15] If you run unstructured interviews after your assessment, that is where your money is going.

Work samples produce a Black and White score gap of 0.67, against 0.79 for cognitive ability and 0.23 for a structured interview.[4] A skills test is roughly three times the gap of the interview it is often sold as an improvement on. Integrity tests, at 0.10, are the fairest thing on the chart and the method candidates like least.

This is not an argument against skills testing. It is an argument for knowing the number before your general counsel does. The trade-off is also smaller than it used to look: reaching the four-fifths ratio with a sensible weighting of methods costs about .09 of validity,[6] not the collapse the older literature implied.

So the question to ask a vendor is narrow and answerable: do you publish criterion validity and adverse impact for the specific tests I will use? Reliability alone does not answer it. A test can be highly consistent and still predict nothing.

Applied to TestGorilla, the fair reading is mixed. They publish per-test reliability with sample sizes, which most skills-first rivals do not. On the two tests we checked, criterion validity and the race and ethnicity analysis both read "pending", and there is no technical manual or published bias audit. Alva Labs and Bryq publish more on this than most, and it is reasonable to ask the others why they publish less.

Straight from TestGorilla's own candidate help centre, on 16 August 2026:[2]

• Most assessments take about 60 to 90 minutes, unpaid.
• You can take an assessment once.
• One question per screen, and you cannot go back after submitting, even if you skipped it.
• Full screen is required. Leaving hides the question, the timer keeps running, and the employer sees it.
• Whether you ever see your score is the employer's decision.

None of that is unusual for the category, and the anti-cheat rationale for most of it is real. It is still an hour and a half of somebody's evening, and half of the complaints in this category are about being asked for that before a human has spoken to them.

Shorter is not the fix everyone assumes

Every vendor in this market competes on assessment length. The research does not support it. Assessment length did not predict candidates quitting; most drop-off happens in the first twenty minutes whatever the total; and giving a longer, more honest estimate upfront was associated with fewer people quitting, not more.[16] Some of the drop-off you would remove by shortening is weaker candidates removing themselves, which is the test doing its job.

What candidates do respond to is whether the task looks like the work. Across 86 samples and nearly 49,000 people, interviews and work samples are rated most favourably, cognitive tests lower, and personality inventories lower still.[13] That ordering held across 17 countries.[14] Face validity, the sense that this is obviously relevant, is the strongest driver of whether people think a process was fair.

The practical read: a relevant 25-minute assessment beats an abstract 8-minute one on both completion and goodwill. Which is awkward for a spec war fought on minutes.

How the prices were checked

Every price on this page was read from the vendor's own website on 16 August 2026, in a browser, not taken from a comparison page. Where a vendor publishes nothing, this page says "not published" rather than repeating a third-party figure, because in this category those figures are wrong more often than they are right. TestGorilla's US pricing page, its billing FAQ, its credits help article and its candidate help centre were all read the same day.

How the evidence was graded

Validity and adverse impact figures come from the 2022 and 2023 Sackett and 2024 Berry papers rather than from the 1998 estimates still in circulation. Grade A means a meta-analysis or replicated finding, B a single strong study, C a famous but contested one. Nothing graded C appears as fact on this page, and the numbers we rejected are listed in section seven rather than quietly dropped.

Limits, stated plainly

Pricing in this category changes often, and TestGorilla has changed model four times in about three years, so treat 16 August 2026 as the shelf life of section one rather than a permanent claim. The cost calculator uses list prices and cannot account for negotiated discounts, which at the top of this market are normal. We have not run every one of these tools, and the review-site complaints are other people's reported experiences, not our findings. Validity figures are averages across many studies with wide spreads, so they describe methods in general and not the specific product you buy.