Screening
Vendor Evidence Request
An HR lead buying or renewing an automated hiring tool, who sends it to the vendor before configuration.
A written request an employer sends an assessment or hiring-tool vendor, asking by name for the twelve documents the employer will have to produce if a regulator, a plaintiff's counsel or its own general counsel asks later. Twelve rows, each naming what to send, why it is asked for, and what an acceptable answer looks like, plus five contract terms that keep the file after the subscription ends. Send it before the tool is configured.
Sending it
Send it before the tool is configured, not after a pilot has started. Every row names a document that either exists or does not, and both answers are decision grade.
Put it in the procurement request rather than in the demonstration. A demonstration is answered by a salesperson in the room. A written request is answered by someone whose answer gets filed.
Name the occupations in the first line, then delete the rows that do not apply to the way the tool will be run. A request covering six job families against one study comes back part answered, and that is the finding rather than a failure of the request.
Date the request and keep the reply, a refusal included. A refusal in writing is a procurement record, and it prices the tool as undocumented for those occupations.
The request
Copy from here.
Send the following for [list the occupations this tool will be used for], with a date on each item. Where a document does not exist in the form named below, say so in writing rather than sending something adjacent. Reply by [date].
| What to send | Why it is asked for | An acceptable answer |
|---|---|---|
| 1. The validity report covering those occupations. | Once a complaining party shows a particular practice causes disparate impact, demonstrate that it is job related for the position in question and consistent with business necessity. 42 U.S.C. 2000e-2(k), effective November 21, 1991. | A dated report naming the occupations it covers, with an author who can be reached by name, address and telephone 1. Not a platform-level statement, and not the one-page summary. |
| 2. The job analysis underneath it. | A report about a job rests on somebody having analyzed that job. An off-the-shelf competency framework analyzed a different one. | The work behaviors, the tasks they sit in, the work products they produce, and how importance was measured 1 2. |
| 3. The map from each task in the tool to a work behavior. | Nobody can defend a difference between two candidates if nobody can say what an item measures. | An item-by-item map, and a comparison of the manner, setting and complexity of the exercise against the work itself 1 2. |
| 4. The derivation of the cutoff to be configured. | A default cutoff was set against somebody else's workforce, and configuring it makes it yours. | How expectations of acceptable proficiency were determined and how the cutoff follows from them 3, what happens to a candidate just below it, and whether the underlying work stays readable to a reviewer. |
| 5. The evidence for the way the result will be used. | Evidence adequate to pass or fail a candidate may not reach a use that sorts them, and the tool will be run one of those ways 3. | Evidence addressed to the use being bought, named as such. Where the result sorts candidates, evidence that a higher result is likely to mean better performance of the work 1. |
| 6. The alternatives investigated. | Looking for procedures, and for ways of using them, with less adverse impact is part of a validity study rather than a courtesy beside it 4. | The scope, the method, the findings and the conclusions of that search 1. That none were investigated is an answer, and it goes in the record. |
| 7. Impact figures broken out by occupation and by the threshold to be run. | Treat a selection rate below four-fifths of the highest group's rate as evidence the enforcement agencies weigh, not as a legal threshold you pass or fail. 29 CFR 1607.4(D). | Rates and ratios for each group, per occupation, at the threshold to be configured, with the counts behind them 5. One model-level figure is not this. |
| 8. The data behind the audit. | An impact ratio is a property of a population. A pool of high-volume retail applications carries its own base rates and its own conventions, and none of them travel to a small senior pipeline. | Whose applicants, which occupations, how many, over what period. Where test data stood in for applicants, why historical data was not used 6. |
| 9. The auditor, the engagement letter and the scope note. | A published audit reflects the scope its purchaser chose, which is why the scope note is worth more than the ratio 7. | The auditor named, their relationship to the vendor and to any employer using the tool stated in writing, any financial interest in either disclosed, and a scope note saying what was not tested. Dated later than the last change to the model 6. |
| 10. The summary of results in a publishable form, and the invite mechanics behind it. | A bias audit within the prior year, a public summary of its results, and notice to the candidate 10 business days before the tool is used. NYC Admin. Code 20-870 et seq., effective January 1, 2023. | A summary that can be posted as it stands, and written invite mechanics: how far ahead a link can be issued, how long it stays valid, and what a candidate reads before anything is captured 6. |
| 11. The accommodation and alternative-process path. | Select and administer any test so the result reflects the skill it measures, not a candidate's impaired sensory, manual or speaking skills, unless those are what it measures. 29 CFR 1630.11. | The written route a candidate takes to ask for a change, which changes exist, who decides, and how the request reaches that decider without a diagnosis reaching the hiring manager 6. |
| 12. Where a study run elsewhere is relied on, the basis for relying on it. | A study travels only as far as job similarity carries it, and the similarity has to be shown rather than asserted 8. | Job analyses on both jobs, an account of why the work behaviors are the same, a comparison of the two applicant pools against the relevant labor market, and the fairness evidence 8 1. |
Copy to here.
Terms to add to the agreement
A document arrives once. These keep it.
- Each item above named as a dated deliverable, with a right to terminate if a named document misses its date.
- A legal hold your own staff can trigger, without asking the vendor first.
- An export of the underlying candidate records in a form that opens without the vendor. A dashboard view is not a record you can produce.
- A retention window set by this agreement rather than by the product's deletion default, long enough to cover the period your counsel names 9.
- A cooperation clause binding the vendor to support an audit or a charge response, at a stated rate, and reaching past the end of the subscription.
A vendor holding the file will sign those. A vendor that negotiates the delivery dates instead of the price has answered the request.
What this does not do
Olive's own bias audit has not been performed. Nothing above reports that any vendor's has been performed either. Every row asks for a document, a document is not a finding, and twelve of them are not a clearance.
It is not a procurement policy and not a contract. Your counsel edits the clause text above before any of it goes into an agreement.
Nothing here reads the report for you. A file that arrives on time and says nothing is still a file that says nothing, and the reading is where the answer is.
Take it
The file and the credit
The publishing entity legal name and postal address are not filled in yet, and both sit inside the disclaimer every packaged format renders. No file is emitted until they are.
Credit line, to paste beside anything you quote from this document.
<!-- Olive template. Licence: https://olive.is/licence/template/1-0 -->
<p><a href="https://olive.is/answers/tools/vendor-evidence-request/" rel="nofollow">Olive</a>, Vendor Evidence Request, version 1.0.0, checked 2026-08-26.</p>Important notes
Not a substitute for the advice of an attorney. This is a starting document for your own attorney to edit. It applies no law to your facts. No attorney has reviewed it for your state or your facts, and Olive is not your lawyer.
Published by Olive Independent Study, Inc., [DELAWARE INCORPORATING ADDRESS], United States. Contact hello@olive.is. A person reads every complaint and answers within ten working days. All concerns that Olive has engaged in the unauthorized practice of law are referred to the North Carolina State Bar, wherever the complaint came from.
Checked 2026-08-26 against the sources listed in this file. Version 1.0.0.
This text disclaims no warranty, caps no liability, waives no remedy, and names no court or state for a dispute. Those absences are deliberate.
Olive assesses how a person works with AI. It does not detect AI-written documents, and it never produces a score, a ranking, or a match percentage for a person. Candidates read the same report the employer reads.
Packaged files
- The publishing entity legal name and postal address are not filled in yet, and both sit inside the disclaimer every packaged format renders. No file is emitted until they are.
Checks
- Sources last re-opened August 26, 2026.
- Next review due February 26, 2027.
Where it came from
Read the argument behind it
- What Do You Ask a Vendor Who Says They're Bias Tested? Ask which roles the audit ran on, whose data it used, and who signed it. Twelve questions with the answer that should worry you, ready for a vendor review.
- How Do You Evaluate an AI-Skills Assessment Before Buying? Ask what the assessment is benchmarked against, what the candidate sees, whether it outputs a single number, and what evidence sits behind a divergent result.
- What Do You Need From an Assessment Vendor for a Bias Audit or EEOC Inquiry? The validity report, the job analysis, the impact figures, the auditor's letter. What each proves, who at the vendor writes it, and the reply that should stop a purchase.
References
- 29 CFR 1607.15 - Documentation of impact and validity evidence Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov The documentation list itself: the essential items in a content-validity report, the job-analysis description, the item-to-work-behavior mapping, the cutoff derivation and the evidence for the use made of the result, the alternatives investigated, the named contact person with mailing address and telephone number, and the transportability documentation in section 15E.
- 29 CFR 1607.14 - Technical standards for validity studies Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov Content validity requires a job analysis of important work behaviors; a content strategy is not appropriate for procedures purporting to measure traits or constructs such as aptitude and judgment; the manner, setting and complexity of the procedure should closely approximate the work situation.
- 29 CFR 1607.5 - General standards for validity studies Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov Section 5G separates the evidence sufficient for a pass or fail use from the evidence needed to sort candidates by the result. Section 5H: cutoff scores should normally be reasonable and consistent with normal expectations of acceptable proficiency within the workforce.
- 29 CFR 1607.3 - Discrimination defined: Relationship between use of selection procedures and discrimination Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov A selection procedure with adverse impact is considered discriminatory unless validated, and the investigation of suitable alternative procedures with less adverse impact is part of the validity study.
- 29 CFR 1607.4 - Information on impact Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov Records disclosing impact by sex and by the listed race and ethnic groups, and the four-fifths rule with its two caveats in both directions.
- Notice of Adoption of Final Rule: Automated Employment Decision Tools (6 RCNY 5-300 to 5-304) NYC Department of Consumer and Worker Protection, 2023. rules.cityofnewyork.us The independent-auditor definition, the historical-versus-test-data rule and the justification it requires, and the employer's own duties to publish the summary of results and to give candidates at least ten business days' notice with alternative-process and accommodation instructions.
- Null Compliance: NYC Local Law 144 and the Challenges of Algorithm Accountability Wright et al., ACM FAccT (arXiv:2406.01399), 2024. arxiv.org Of 391 New York City employers checked by 155 investigators in 2024, 18 posted an audit report and 13 posted a transparency notice; the authors read the pattern as employer discretion over audit scope rather than as clean tooling.
- 29 CFR 1607.7 - Use of other validity studies Uniform Guidelines on Employee Selection Procedures, eCFR, 1978. ecfr.gov Borrowing another user's validity study requires job similarity shown by job analyses on both jobs plus fairness evidence, and cannot be relied on where other variables are likely to affect validity significantly.
- 29 CFR 1602.14 - Preservation of records made or kept EEOC recordkeeping regulations, eCFR, 2024. ecfr.gov One year from the making of the record or the personnel action, whichever is later; where a charge is filed, all relevant personnel records until final disposition, expressly including test papers completed by the unsuccessful applicant and all other candidates for the same position.
9 sources, numbered by first appearance. How Olive sources claims