Candidate evaluation software and candidate assessment scorecards with the evidence behind every rating

Candidate evaluation software does four jobs: it fixes the criteria for a role before anyone reads a resume, it gathers evidence against those criteria from resumes, tests and interviews, it converts that evidence into ratings on one scale, and it keeps a record of who decided what and why. Tools differ mainly in which of the four they cover.

The first job is the one teams skip. A hiring manager says "strong communicator, senior, knows our stack", the recruiter writes a job ad, and three interviewers each carry a slightly different version of the role into their conversations. Evaluation software forces the version into writing: five to eight criteria, each marked as a must-have or a nice-to-have, each with a weight. Once that exists, every later rating has something to point at.

The second and third jobs are where the categories split. An applicant tracking system collects ratings from interviewers on a scorecard form. A skills testing tool gathers evidence by setting tasks. An interview notetaker records the conversation and summarizes it against the questions. An AI evaluation agent reads the resume or profile itself and rates each criterion with the line that supports it. HireAgent sits in that last group, which is why it is most useful at the top of the funnel, when there are 200 applicants and nobody has spoken to any of them.

The fourth job, the record, is now a legal question as much as an operational one. New York City, Illinois and California all put duties on employers whose tools substantially drive who advances, and a debrief note that says "felt like a better fit" does not survive a challenge. A stored scorecard with criteria, ratings and evidence does.

Most teams end up with two of these, not four. The common pairing is an AI evaluation layer at the top of the funnel and ATS scorecards for the interview loop, with a skills test added only for roles where a task genuinely predicts the job. Buying a test for every role is the most frequent way evaluation budgets get wasted: completion rates fall on senior roles, and the strongest passive candidates simply do not take a 45-minute assessment for a job they have not decided they want.

To compare candidates objectively, agree on weighted criteria before you see anyone, rate every candidate on the same scale against written evidence, and total the weighted scores only after each criterion is rated on its own. Reading the whole resume and then picking a number is the habit that lets a strong first impression decide everything.

Here is a worked example for a customer success manager role with five criteria. Weights add to 100.

Candidate C has the best onboarding and tooling ratings and the most impressive resume to skim, yet ranks last, because the must-have carries a 2. Candidate A looks strongest on the must-have and loses to B on renewals. That is the conversation a debrief should have, and it only happens when the ratings are written down criterion by criterion before anyone argues.

Two rules make the method hold. A missing must-have caps the total instead of being averaged away; in HireAgent a candidate without a must-have cannot rank above a candidate who has all of them. And every rating needs a sentence of evidence next to it. A 4 with "managed 52 mid-market accounts at a payroll SaaS, 2021 to 2024" is a rating. A 4 with nothing next to it is a feeling.

A candidate evaluation scorecard should include five to eight job-related criteria, a must-have or nice-to-have flag and a weight for each, a rating scale with a written definition of every point, a space for the evidence behind each rating, and an overall recommendation recorded separately from the criterion ratings.

The rating definitions are what most scorecards leave out, and they are what makes two interviewers comparable. "3 = meets the bar" means nothing on its own. "3 = has owned a book of accounts of this size for at least a year, with named outcomes" means the same thing to everyone in the loop.

Keep the criteria about the work. Years of experience is a weak criterion on its own; what the person did in those years is a strong one. Avoid criteria that act as proxies for protected characteristics, such as graduation year, a specific university list or "culture fit" with no definition, because those are exactly what an adverse-impact review looks for.

Finally, separate the overall recommendation from the ratings and collect it last. If interviewers give a hire or no-hire first and fill in the criteria afterwards, the criteria get bent to match the verdict. HireAgent applies the same discipline to its own output: it rates each criterion on evidence, totals them, and leaves the recommendation and every advance or reject to a person on your team.

Yes. US employers can use AI to evaluate candidates, but where a tool substantially drives who advances, several jurisdictions require notice, bias audits and record keeping, and the rules follow where the candidate is located.

New York City Local Law 144 requires an independent bias audit within the past year, a published summary and candidate notice at least 10 business days before an automated employment decision tool is used for a New York City role. Illinois requires notice when AI is used in employment decisions and bars using zip code as a proxy for a protected class. California regulations on automated-decision systems under FEHA, in effect since 1 October 2025, prohibit discriminatory use and require four years of record retention. Title VII adverse-impact analysis applies to any selection procedure, automated or not.

There is also an open question about consumer reporting. A proposed class action filed in California on 20 January 2026 alleges that one large talent-intelligence vendor compiled data on candidates and delivered match scores to employers in a way that makes it a consumer reporting agency under the Fair Credit Reporting Act. The case is unresolved and the allegations are only allegations, but the practical lesson for buyers is clear: ask every vendor what data it evaluates beyond what the candidate submitted, and whether candidates can see and dispute it.

HireAgent evaluates the application, the resume and the profile the candidate or your sourcing brief points to, stores the criteria version, the rating and the evidence for every score, and leaves every decision to a human. That is the record each of these rules asks for.

Candidate evaluation software costs from about $15 per user a month for ATS scorecards with AI scoring, to $215 a month and up for a skills-testing account, to $299 a month for an AI agent that evaluates and ranks candidates, to $100,000 a year and more for enterprise talent-intelligence platforms. The figures below are read from each vendor's own pricing page or marketplace listing.

The enterprise end deserves a number because almost nobody publishes one. Eightfold lists a Starter Edition on the SAP Store at USD 25,000 per quarter with a one-year minimum, priced by the customer's employee count, which puts the entry point at $100,000 a year before anything is configured. Our Eightfold AI pricing breakdown covers what that figure includes and what it leaves out.

HireAgent publishes every plan. Solo is $299 a month for one seat and up to three open roles, or $1,794 a year billed annually. Growth is $799 a month, or $4,794 a year, for three seats, up to ten roles, autosend outreach and ATS integration with Greenhouse, Lever and Ashby. Scale is $1,999 a month for ten seats and SSO. Every plan includes sourcing, evaluation, ranking, outreach and scheduling.

The comparison most buyers actually face is time, not another tool. If a recruiter spends 6 minutes on each of 250 applicants for one role, that is 25 hours of reading before the first phone screen. Cutting it to the top 40 with evidence already attached gives back most of a working week per role. The AI recruiting software pricing page sets every vendor we have priced side by side.

In-house talent teams buy evaluation software to make a debrief defensible. A mid-size US company running 30 open roles has dozens of interviewers who were never trained on the same bar, and the evaluation layer is how the recruiter keeps the loop honest. For high applicant volume, the high volume hiring software page covers the extra controls.

Staffing and search firms buy it for the submittal. A client asks why these three candidates and not the other forty, and a criterion-by-criterion card with evidence answers that in one page. The AI recruiter for staffing agencies page walks through that workflow.

Hiring managers without a recruiter buy it to avoid reading 300 resumes at night. A founder or department head can write the brief, set the must-haves and get a ranked, explained shortlist the next morning; the AI recruiter for hiring managers page shows how that runs.