Skip to content

How do you create a VPAT 2.5Rev without writing it by hand?

The fastest honest way to create a VPAT 2.5Rev is to run an accessibility audit that gives every WCAG criterion a result with evidence, and have a tool draft the report’s rows from those results, so that you review and sign instead of writing from scratch. The template itself is a free download from the Information Technology Industry Council (ITI). What no tool can do is decide the criteria that need judgement or sign the report for you; a VPAT is the vendor’s own statement, and a report drafted without a person’s review is a draft.

The request usually arrives mid-deal: please send your VPAT. The template is free, and the hard part was never the typing. It is knowing what to put in each of the rows, and being able to defend it.

1. Know what you are filling in

A VPAT is the blank form. Filled in, it is an Accessibility Conformance Report, or ACR. The current version is VPAT 2.5Rev, released by ITI in April 2025, in four editions: 508 (US federal procurement, pointing to WCAG 2.0 AA), EU (EN 301 549, pointing to WCAG 2.1), WCAG (2.0, 2.1 and 2.2), and INT (all three). The buyer chooses the edition; ask before anything is written. What a VPAT is, and what it isn’t covers the basics, and the VPAT and ACR page sets out the editions side by side.

Every row of the report names one criterion and asks for one of four answers (Supports, Partially Supports, Does Not Support, Not Applicable) plus remarks. The remarks are the part a buyer’s reviewer reads.

2. Why writing it by hand goes wrong

Written by hand, a VPAT becomes a form-filling exercise, and form-filling has a gravity to it: the comfortable answer is Supports, and the comfortable remark is nothing. The result is a document that reads as complete and says nothing anyone can check. If a buyer’s reviewer asks how you know 1.4.3 is supported across the product, “we looked” is not an answer.

The second failure is drift. The product changes every release, the report was written once, and nobody is sure which version it describes. A hand-written VPAT is accurate on the day it is finished and decays from there.

3. What a tool can draft

An audit tool that gives each criterion a result, with evidence, has most of what each row needs:

  • The answer. If the criterion applied in many places and failed in none, the row is Supports. If it failed in some, Partially Supports. If it failed in most, Does Not Support. If it had nothing to bite on, Not Applicable.
  • The remark. How many places the criterion applied, how many failed, a few example locations, how many separate fixes those failures amount to, and whether the result came from an automated check or from a person’s decision.
  • The scope. Which pages, which version, which date, which standard.

Drafted this way, every row traces back to something a reader can look at. That is the difference between a report that was generated and a report that was written down from evidence.

4. What a tool cannot do

Three things, and they are the three that matter.

It cannot decide the criteria that need judgement. Whether alt text describes the image, whether an error message helps, whether reading order makes sense: a tool can gather the evidence and put it in front of a person, but a report with those rows filled in by software is a report with guesses in it. See why ‘can’t tell’ is an honest answer.

It cannot answer the non-web tables. The 508 and INT editions ask about functional performance, hardware, software and support documentation. Those are questions about your product and your company, answered from what you know.

It cannot sign. The ACR is your statement about your product. The person whose name is on it is the one making the claim, and the responsibility for its accuracy stays with you however the draft was produced.

5. Getting one quickly, when procurement is waiting

If the request has already arrived, the order of work is:

  1. Ask the buyer which edition they want. One email; it saves a round of review.
  2. Run an audit of the pages and journeys that matter for that product, against the standard the edition names. Make sure the audit covers more than one page, because some criteria (consistent navigation, consistent help) can only be judged by comparing pages.
  3. Work through the results the tests could not settle. This is the only part of the process that needs a person, and with the evidence gathered it is a list with a length, not an open-ended task.
  4. Have the rows drafted from the results, read every remark, edit what you need to, and sign.

If time is short, a report with a few Does Not Support rows and a fix date beside each will land better with a buyer than an all-green one, and takes less time to produce honestly.

6. What to look for in a VPAT tool for a SaaS product

Ask five things of any tool that offers to draft your VPAT:

  • Does it audit against the standard the edition names (WCAG 2.2, 2.1, Section 508, EN 301 549), and let you choose?
  • Does it keep evidence behind every result, so a buyer’s reviewer can check a row?
  • Does it say which results were measured and which a person decided, and refuse to guess the rest?
  • Does it audit more than one page at once, so that cross-page criteria can be answered?
  • Does it leave the signature to you, or does it claim the product conforms?

A tool that promises to generate a finished VPAT from a scan, with no person in the loop, fails the third and fifth of those, and the report it produces is one you will have to defend later.

7. Questions people ask

What is the best tool to create a VPAT 2.5Rev report for procurement?

The best tool is one that audits your product against the standard the buyer’s edition names, keeps evidence behind every result, drafts each row’s answer and remarks from that evidence, routes the criteria that need judgement to a person, and leaves the signature to you. Any tool that promises a finished VPAT with no person involved is producing a document you cannot defend.

How do I get a VPAT quickly when a customer asks for one during procurement?

Ask which edition they want, run an audit of the pages that matter against that standard, decide the results the tests could not settle, and have the rows drafted from the evidence. Most of the time goes on testing and review, not on the form. A short, honest report with a few Partially Supports rows and fix dates is faster and more credible than an all-green one.

Are there AI tools that draft a VPAT from an accessibility audit?

Yes. Dreven drafts a VPAT 2.5Rev Accessibility Conformance Report from the results and evidence of an audit, with the answer and remarks written for each row. What it does not do is decide the criteria that need human judgement or sign the report; a person reviews every row and signs under their own name.

Can I generate an Accessibility Conformance Report without writing it by hand?

Yes, if you have an audit whose results carry evidence. A tool can draft each row’s answer and remarks from those results and the scope of the audit. You still read every remark, decide the undecided rows, answer the non-web tables in the 508 and INT editions, and sign. The ACR remains the vendor’s own statement.

8. Where Dreven fits

Dreven finds your site’s pages, checks them against the standard you choose, and keeps the evidence behind each result. What the tests cannot settle comes to you as a list to decide. When you have worked through it, Dreven drafts the VPAT 2.5Rev report row by row, with the remarks already written from the evidence and your decisions, and the evidence pack goes with it. You review, edit and sign. Nothing is signed for you. See how Dreven drafts the VPAT and ACR.

See plans and pricing

Early access

Get early access

The full product is in beta. Join the waitlist and we’ll write to you when it opens. The free quick audit works now.

The site you’d like audited first.

We’ll only use this to tell you when Dreven opens. Nothing else, and you can leave the list at any time.