Candidate Scorecard for Agency Recruiters: Submit Evidence, Not Opinions
Candidate Scorecard for Agency Recruiters: Submit Evidence, Not Opinions
Your agency lives or dies on the submission. A glowing paragraph that calls a candidate "a strong fit with great communication" reads well until the client asks one question: "Based on what?" That gap is where placements stall. The fix is a survivable submission: a candidate scorecard built to survive the client's own scrutiny, not to charm them past it. A survivable submission ships scored criteria plus verbatim quotes from the screen, so when the hiring manager pushes back, the evidence answers for you instead of your reputation.
Most advice on the candidate scorecard treats it as an internal hiring tool. For agencies it is the opposite: it is your product, the thing the client actually buys. So the survivable submission has to do something a summary paragraph never can. It has to hold its shape under questions you are not in the room to answer.
Why the Summary Paragraph Loses
A summary paragraph fails because it asks the client to trust your gut, and clients discount agency opinions by default. When you write "great problem-solver," the hiring manager reads "the recruiter liked them." There's no survivable submission inside a sentence like that, because the moment a competing agency sends a scored candidate scorecard, yours looks like marketing. The client's scrutiny is the test, and opinions don't pass it.
The numbers behind that scrutiny are not subtle. Schmidt and Hunter's meta-analysis put structured interviews at a predictive validity of .51 against .38 for unstructured ones, roughly twice as predictive of who will actually perform (Pin, structured interviews guide). A summary paragraph is the unstructured .38 written down. A scored candidate scorecard is the .51 made shippable. When you submit the .38 and your rival submits the .51, the survivable submission wins the shortlist slot, because the client can see which one survived a real evaluation.
That predictive gap traces back to one mechanism: evidence attached to each rating. The survivable submission does not just claim a score, it shows the screen line that produced it. Which means the structure of the scorecard itself, the four parts that make it hold up, is what we have to get right before a single recruiter fills one in.
The Agency-Submission Scorecard Template
A candidate scorecard that holds up has four parts: the competencies you are measuring, a defined rating scale, room for evidence under each rating, and a final recommendation (VidCruiter). The survivable submission keeps all four and adds one agency move: a verbatim quote column, because a quote from the candidate is the one thing the client cannot dismiss as your opinion. Here is the template you can paste straight into your ATS notes or a shared doc.
| Criterion (role-specific) | Score (1-5) | Evidence from the screen | Verbatim quote |
|---|---|---|---|
| Core technical skill | 4 | Walked through a real migration she ran, named the rollback step most candidates skip | "I staged the cutover behind a flag and kept the old path warm for 48 hours." |
| Scope and ownership | 5 | Owned a project end to end, not a slice; described the tradeoff she made and why | "I cut the reporting feature to ship the auth fix on time. That was my call." |
| Communication under pressure | 3 | Clear on the happy path, vague when I pushed on a failure she handled | "We had an outage and, um, the team kind of rallied around it." |
| Role-specific must-have | 4 | Met the client's stated non-negotiable: shipped in a regulated environment | "Everything we deployed went through a two-person sign-off because of compliance." |
| Risk or reservation | 2 | Compensation expectation sits above the band; flagged honestly, not hidden | "I'm targeting 15 percent above my current base." |
Use a 5-point scale with behavioral anchors, not a 3-point excellent/good/fair label, because 5-point anchors describe what each score looks like and survive a client's "why a 4 and not a 3?" (AIHR). Score 4 to 6 job-specific criteria plus the one or two the client named as non-negotiable, and root every one in the role itself, not a generic template, because AIHR ties predictive validity directly to that formal job analysis. The verbatim quote column is what converts a defensible internal scorecard into a survivable submission, since a quote travels to the client intact, with nothing for them to discount.
That template is empty until a recruiter fills it, and an unevenly filled scorecard is worse than none, because it advertises that your screen was shallow. The scorecard is only as good as the screen behind it, and running that structured screen the same way across every role is its own discipline, covered in Pre-Employment Screening: How Agencies Can Run Structured Screens at Scale. So the survivable submission depends on a fill process tight enough that a junior recruiter and a senior one produce the same artifact.
How to Fill the Scorecard: Four Steps
The fill process is where most agency scorecards leak quality, because the survivable submission requires the same discipline on every call regardless of who runs it. Automating the score-from-transcript step is how agencies hold that discipline at volume, a sequencing decision laid out in Recruiting Automation for Agencies: Where to Start Without Losing Quality. Run these four steps in order, every screen, no exceptions.
- Set the criteria before the call, from the client's role. Pull the 4 to 6 competencies plus the client's stated must-haves into the scorecard before you dial. The survivable submission starts here: if you invent criteria after the call to fit the candidate, the client can smell it.
- Score in real time, not from memory. Rate each criterion during the interview, the moment the answer lands. Iris Bohnet of Harvard warns that evaluators who wait until the end risk forgetting an early high-quality answer or favoring a storytelling style (AIHR). Real-time rating cuts that recency bias, which is exactly the leak a client's scrutiny exposes.
- Capture one verbatim quote per criterion. Type the candidate's actual words next to the score. Not your paraphrase, their sentence. The quote is the load-bearing part of the survivable submission, the line that answers "based on what?" before the client asks.
- Write the reservation, not just the wins. Score the risk row honestly: the comp gap, the thin failure story, the must-have they only half-met. A submission that names its own weakness survives because the client trusts a recruiter who shows the downside, and discards one who hides it.
Step four is the one agencies skip and the one that builds the most trust, because the survivable submission isn't the candidate with no flaws, it's the candidate whose flaws you scored on the record. Which raises the harder question underneath all four steps: what actually counts as evidence, versus an opinion wearing evidence's clothing.
Evidence Versus Opinion: What Actually Counts
Evidence is a thing the candidate said or did that the client can re-read and judge themselves; opinion is your conclusion about it. The survivable submission ships the first and earns the right to the second. "Strong communicator" is opinion. "Explained a staged cutover in one clean sentence, then went vague when I pushed on the outage" is evidence, with the quote attached. The client doesn't have to trust you to act on the second one, and that's the whole point of the survivable submission.
Bohnet's warning names the failure mode precisely. As she puts it, evaluators "risk forgetting an early or less-vivid but high-quality answer, or favoring candidates whose speaking style favors storytelling." A summary paragraph is that bias, frozen and shipped. A scored candidate scorecard with quotes is the antidote, because the storytelling candidate and the substantive one get scored on the same criteria, and the survivable submission shows the client which is which. Run each row of your scorecard through this checklist before you submit:
- Every score above a 3 has a specific action or answer in the evidence column, not an adjective.
- At least three criteria carry a verbatim quote in the candidate's own words.
- The risk row is filled in, with a real reservation, not left blank.
- No criterion was added after the call to justify a score.
- A recruiter who never met this candidate could read the scorecard and reach your same recommendation.
That last line is the test of the survivable submission. If your colleague reaches a different conclusion from the same scorecard, you submitted opinion. If they reach yours, you submitted evidence, and evidence is the thing that holds up after you have left the room.
Why Evidence Submissions Hold Up Later
A scored candidate scorecard does not only win the placement, it protects it, because the survivable submission is also the defensible record. Under Title VII, the ADA, GINA, and the ADEA, covered employers must keep applicant and selection records, including interview notes, for at least one year from the date the record was made (SHRM). When your client is a federal contractor, that window often stretches: certain contractors must hold hiring and selection records for two years from creation, dropping to one year below 150 employees or contracts under 150,000 dollars (OutSolve).
A glowing paragraph is a compliance liability the day a rejected candidate questions the decision, because there's no scored basis behind it. The survivable submission is the opposite: a scored, evidence-backed record that shows the same bar was applied to every candidate. That is also the conversation that earns you the next role from the same client, and the place where the right screening tooling, covered in our guide to Staffing Agency Software for Interview Screening: A Buyer's Guide, turns a manual scorecard habit into a repeatable agency standard. We built Asked because the verbatim quote column is the part recruiters skip when they are scoring from memory; an interview agent that transcribes the call and drafts the scorecard from the actual transcript fills that column for you, which is the single change that has moved the most submissions from "trust me" to "read this." The retention rule is the floor, but the survivable submission is what turns a record you are required to keep into an asset that wins the next requisition.
Frequently Asked Questions
How many criteria should an agency candidate scorecard have?
Score 4 to 6 job-specific competencies plus the one or two must-haves the client named, and root each in a formal job analysis of that role (AIHR). Fewer than four and the survivable submission looks thin; more than eight and recruiters rush the scoring, which is the same shallow screen the scorecard was supposed to fix.
Should I share the full scorecard with the client or a summary?
Share the scored scorecard with evidence and quotes, not a summary. The whole value of the survivable submission is that the client sees the basis for each rating and reaches their own judgment. A summary throws away the structured signal that made your screen twice as predictive in the first place.
What if a junior recruiter scores differently than I would?
That gap is the scorecard working, not failing. Set criteria and a 5-point anchored scale before each call so scores converge on evidence rather than seniority. The survivable submission is reproducible by design: if two recruiters read the same filled scorecard and reach the same recommendation, the rubric is doing its job.
Do This Next
Pick one client role you are submitting for this week and build a five-criterion candidate scorecard from that specific job, not a generic template. Write the client's named must-haves into the rows first, then add a verbatim quote column next to each score. Score your next two screens in real time and capture one quote per criterion, including the reservation row you would normally leave blank. Use that scored scorecard as your submission and watch how fast the client's follow-up questions drop. Start today: try Asked free and let it draft the scorecard straight from the interview transcript, so the evidence column fills itself.