Pricing
Priced per assessment, with the limits written down
Five plans, published numbers, and an honest account of how a plan actually gets set today. The free tier is a real one — twenty-five assessments a month is enough to run a pilot and decide, rather than a demo that expires before you have learned anything.
The plans
Prices in US dollars, per month. Every figure below is the value configured in the product itself rather than an indicative range.
Going over an included allowance costs 40 cents per extra credit on Starter, 30 on Growth and 22 on Scale. Since one assessment is one credit, that is the price of one additional candidate sitting one test. The free plan has no overage — it stops rather than quietly billing you, which is the behaviour a free plan should have.
What one assessment means
The definition matters more than the price, because it is where assessment pricing is usually made deliberately confusing.
One assessment is one candidate starting one sitting. Not an invitation sent, not a candidate created, not a question answered. A candidate you invite who never turns up costs nothing. A candidate who starts, disconnects and resumes the same attempt is one assessment, not two.
Inviting costs nothing, but it is where the allowance is checked
You are billed for the sittings that actually happened, so invitations nobody takes up are free. The plan’s monthly allowance is checked when you invite rather than when they start, which is deliberate: a drive larger than your allowance is stopped at the door instead of halfway through, with candidates already sitting.
Candidates are counted, not charged
A candidate record counts against the plan’s monthly candidate limit and costs no credits.
Reviewing is free
Replay, integrity analysis, similarity scans and diff review are not metered. Looking at your own evidence should not have a price.
Camera analysis is the exception
It calls a paid model per frame, so it is metered per frame — and it is off until an account owner deliberately turns it on.
What else is metered, and at what rate
Published because a usage-based bill you cannot reconstruct is a bill you cannot argue with.
Your usage is visible inside the product against each of these, itemised, for the current billing month. Model token spend and code executions are recorded too, at zero, purely so a bill can be reconstructed later — they are measured, not charged, and they are deliberately never a cap. A limit on code runs would refuse a candidate part-way through an assessment, and a candidate mid-exam is not a billing event.
How limits behave when you reach one
- Monthly allowances reset; point-in-time counts do not. Assessments, AI grades, camera frames and interview minutes are summed within a billing month. Seats, active tests and storage are counted as they are right now, so archiving a test or letting recordings expire gives the capacity straight back. These two are kept separate on purpose: adding up "how many active tests" across a month would tell an organisation with three that they had used ninety of their three.
- A missing limit means unlimited, never zero. If a limit is absent from a plan it is uncapped. The failure mode of a typo or a partial write has to be "we did not cap this", never "we capped it at nothing and locked the account out mid-drive".
- Enterprise agreements are measured, not enforced. Usage is counted and reported; nothing is blocked. A negotiated contract should not have a technical tripwire in the middle of it.
How you actually buy this today
Plainly: there is no card checkout. There is no payment processor wired into this product, and a company administrator deliberately cannot move their own account onto a larger plan — that would be self-service money printing with an audit trail saying it was authorised.
So the path is: create an account, use the free plan, and when you need more, request the plan from inside the product. The request is recorded and a person answers it. Invoicing is arranged directly. That is slower than a checkout button and it is what is true, and a "Buy now" button that opened a contact form would waste your time in a less honest way.
Buying from a small vendor
Worth addressing directly rather than letting you discover it in a security review.
ProctoHire is a young product on a new domain, sold against companies with years of authority and large enterprise teams. That is a real consideration and it cuts in both directions. What you should weigh honestly:
- There is no SOC 2 report and no ISO 27001 certificate. This is stated first here and first on the trust page inside the product, because it is the first thing a security team checks and finding it hidden would be the end of the conversation anyway.
- There are no pre-built connectors to specific applicant tracking systems. There is a scoped API and webhooks, so an integration is something you build rather than switch on.
- Support is direct rather than tiered. For a small team that is usually better; for a procurement process that requires a named support SLA it is a gap you should raise early.
- What you get in exchange is a product that will answer a specific question about how something works, and change something in a week rather than in a roadmap quarter.
The full posture — where data lives, how long it is kept, which other companies touch it, and how a candidate's data is erased on request — is on security and data protection.
- Free plan
- 25 / month
- Assessments started. No card, no expiry, no sales call.
- Cheapest paid plan
- $99 / month
- 250 assessments, 5 seats, public API.
- Extra assessment
- $0.22 – $0.40
- Depending on plan, once the included allowance is used.
- Reviewing evidence
- Not metered
- Replay, insertion analysis, similarity scans and diffs are free to run.
Elsewhere on this site
Buyer FAQ
GDPR, retention, appeals, dropped connections, and what happens when the evidence is inconclusive.
Proctoring and integrity evidence
What a sitting records, what raises a finding, what each finding is worth, and the signals this product deliberately refuses to score.
Session replay
Scrub the whole sitting with the evidence laid over the same timeline — including the findings the guards cleared.
Screening developers
Answers that execute against test cases in 22 languages, with the console output the candidate actually saw.
Start on the free plan and find out
Twenty-five assessments a month is enough to run a real hiring round. If it works, ask for a bigger plan; if it does not, you have lost an afternoon rather than a procurement cycle.