What actually transfers when you switch accessibility monitoring vendors?

Almost nothing transfers automatically. Your page inventory and crawl configuration are quick to rebuild. Everything that took time to accumulate is vendor-shaped and does not move: the dated history of when each issue first appeared and when it was fixed, the suppression list of findings your team reviewed and dismissed as false positives or accepted risks, the severity overrides, the assignment history, and any stored screenshots or DOM snapshots used as evidence. A new platform starts you at an empty history with a fresh full-site backlog, including every item the old one had already been told to ignore.

Why did our accessibility score drop the week we changed vendors?

Because the number was never a property of your site. It is a ratio the vendor defined: which rule set and engine version ran, how many pages were sampled, whether an issue is counted once per element or once per page or once per rule, and how severity is weighted into a single figure. Two competent tools can look at an identical site and report figures that differ by a wide margin without either being wrong. Expect the number to move on the first scan, tell whoever reads the dashboard in advance that it will move, and never present a cross-vendor before-and-after as evidence of progress or regression.

Should we run both platforms at the same time?

For at least one full scan cycle, yes, and it is the cheapest insurance in the whole migration. An overlap month gives you a same-week reading from both tools on the same pages, which is the only way to translate the old numbers into the new ones for anyone who has been tracking them. It also catches the failure that otherwise appears a month later: the new crawler cannot reach your authenticated area, your SPA routes or your PDF library, so its clean report is not an improvement but a smaller sample. Budget one month of double cost and negotiate it into the new contract as a free onboarding period rather than paying twice.

What should we get in writing before giving notice?

Four things from the outgoing vendor. First, the export format and scope in concrete terms — a CSV or JSON of findings with URL, selector, success criterion, first-seen date and status, not a set of PDF reports. Second, how long the account stays readable after the final invoice, because some platforms revoke the dashboard on the cancellation date and the export you meant to take becomes impossible. Third, whether any conformance statement, VPAT or accessibility statement they host on your behalf will remain reachable or will 404 the moment the account closes. Fourth, the actual notice period and auto-renewal date, since annual accessibility subscriptions commonly renew on sixty or ninety days notice and the renewal is the only moment you hold leverage.

Do we lose our legal position by switching vendors mid-remediation?

Not if you keep the evidence. What matters in a demand letter or a consent-decree review is a documented, dated record of what was found, what was fixed and when — not the continuity of a single subscription. The risk in a migration is purely evidentiary: if the dated findings and the fix timestamps live only inside the old dashboard and the account closes, you have lost the paper trail for that period. Export it, store it somewhere you control, and say in your accessibility statement that monitoring is continuous rather than naming the tool. If you are under an active agreement with a specific reporting obligation, check whether it names a methodology before you change one.

Is it worth switching at all, or is every platform the same?

The automated rule engines overlap heavily — most are built on the same open-source core — so if a change is motivated purely by hoping for better detection, the gain is usually small. The differences that justify the disruption are structural: whether it can authenticate and crawl the parts of your product where the real failures live, whether a finding exports in a shape your issue tracker can ingest, whether the pricing unit matches how your site grows, and whether it produces evidence a lawyer or a procurement officer will accept. Decide against those four and the migration is worth the lost history. Decide against a score, and you are paying to reset a number.

Switching Accessibility Monitoring Vendors in 2026: What Does Not Come With You

The Renewal Date Is the Only Leverage You Will Get

Accessibility monitoring is sold on annual terms with automatic renewal and a notice period measured in months, which means the window in which you can negotiate anything at all — price, scope, export rights, an overlap period — opens once a year and closes quietly. Most teams discover their notice period by missing it. Put the renewal date and the notice deadline in a shared calendar with a sixty-day warning the day you sign, and treat the eight weeks before that deadline as the evaluation window, not the week after the invoice arrives.

This also changes what a competing vendor will do for you. A trial that runs inside your notice window, against a real deadline, gets onboarding attention and concessions that the same conversation in month four does not. The concession worth asking for is rarely a discount; it is a free overlap month, migration of your suppression list, or a written commitment on export format.

What Moves, What Rebuilds, and What Is Simply Gone

It is worth being specific, because the parts that are cheap to rebuild are the parts people worry about, and the parts that are unrecoverable are the parts nobody mentions in the sales call.

Rebuilds in an afternoon: the page inventory, crawl depth and exclusions, user seats, alert recipients, and the integration into your issue tracker or chat tool.
Rebuilds painfully: the suppression list. Every false positive your team triaged and dismissed, every accepted risk with a written justification, every severity override. A new platform re-raises all of it as a fresh backlog, and the second triage pass is the single biggest hidden cost of a migration.
Does not move at all: first-seen and fixed-on dates per issue, the trend line, stored screenshots and DOM snapshots, and any audit trail of who dismissed what and why.
May vanish at cancellation: a hosted accessibility statement, a VPAT or conformance report served from the vendor's domain, and any public certification badge. If a URL you have published elsewhere points at their infrastructure, it will 404.

The practical response is to export before you give notice rather than after. Ask for findings as structured data — URL, element selector, success criterion, severity, first-seen date, current status, resolution date — and store the file somewhere you control alongside a dated copy of the final report. A folder in your own document store is a perfectly good archive and it costs nothing. A set of PDFs you cannot query is not an archive.

Your Score Will Change and Your Site Will Not Have

The first scan on the new platform will produce a different number, usually a noticeably worse one, and the reason is structural rather than diagnostic. Vendors differ on the rule set and engine version they run, on how deep they crawl, on whether an issue is counted once per failing element or once per page or once per rule, and on how severity is folded into a headline figure. A single missing label in a shared header component is one finding in one tool and four thousand in another. Neither is lying.

Two things follow. Tell the people who read the dashboard — usually a manager who has been watching a line go down for a year — before the switch rather than after, because an unexplained jump reads as a regression and burns the credibility of the whole programme. And never publish a cross-vendor before-and-after as evidence of anything. If you need continuity in a board update or a compliance report, the honest framing is two separate series with a stated break, not one line.

The Overlap Month Is the Cheapest Insurance in the Migration

Run both platforms across one complete scan cycle before the old one lapses. It costs a month of duplicate subscription and it buys two things you cannot get any other way. The first is a translation table: the same pages, in the same week, measured both ways, so you can say what the new number means in terms of the old one. The second is detection of the quiet failure mode — the new crawler that cannot log in, cannot render your client-side routes, times out on the pages behind a bot filter, or skips your document library entirely, and therefore reports a clean site because it looked at a third of it.

Check coverage explicitly rather than trusting the summary. Compare the crawled-URL count between the two tools, then spot-check a page you know is broken and confirm the new platform found the specific defect. A scanner that reports fewer issues because it saw fewer pages looks identical, on a dashboard, to one that reports fewer issues because your site improved.

A Migration Order of Operations

The sequence matters more than any individual step, because several of these become impossible once notice is served.

Find the renewal date and the notice deadline; work backwards eight weeks.
Export the full findings history, suppression list and final report while the account is live.
Confirm in writing how long the outgoing dashboard stays readable after the final invoice.
Identify any hosted artefact — statement, VPAT, badge — and rehost it on your own domain first.
Stand up the new platform and reproduce the crawl scope, including authenticated areas.
Run one full cycle in parallel; compare crawled-URL counts, not scores.
Re-apply the suppression list, reviewing rather than bulk-importing — some of it was wrong.
Announce the metric break to everyone who reads the number, before the first new report.
Only then serve notice.

When Not to Switch

If the complaint is that the current tool finds too few issues, a migration will probably disappoint you. The automated layer of accessibility testing is largely built on a small number of shared open-source engines, so detection differences between mainstream scanners are narrower than their marketing implies, and the criteria that need a human remain out of reach of all of them. Switching to solve a detection gap is usually switching to solve a coverage or a methodology gap that a new subscription does not touch.

The reasons that do justify the disruption are the structural ones: it cannot reach the authenticated product where your real users live, a finding cannot be exported into a shape your developers can act on, the pricing unit punishes the way your site grows, or the output is not evidence anyone outside the team will accept. Those are permanent properties of a platform, and no amount of tuning fixes them.

Questions Teams Ask Before Changing Vendors

How long does a realistic migration take?

Setup is hours. The honest figure is one to two scan cycles, because the work is not configuration — it is re-triaging the backlog the new tool raises, including everything the old one had been told to ignore. For a mid-sized site, plan a week of someone's attention spread across a month, and do not schedule it against a compliance deadline. Teams that treat it as a same-day swap end up with an untriaged backlog that nobody reads, which is functionally the same as having no monitoring.

Can the new vendor import our issue history?

Ask, but expect a partial answer. Some platforms will ingest a CSV of your suppression list so your team is not re-dismissing the same false positives, which is the import actually worth having. Almost none will reconstruct first-seen dates or a trend line, because those are derived from their own scan record and cannot be faked without inventing data. Treat any promise of a full history import with suspicion and get the specifics of what maps to what before it becomes a decision criterion.

We are mid-way through fixing a backlog. Is now a bad time?

Mid-remediation is the most expensive moment to move, because the thing you lose — the dated record of what was found and when it was fixed — is exactly what the current project is producing. If the move cannot wait, export first and keep the old findings file as the evidence for that period, then treat the new platform's first scan as a new baseline rather than as a continuation. If it can wait, finish the batch, capture the closing report, and switch at a natural boundary.

What does an auto-renewal clause typically look like here?

Annual term, automatic renewal, and thirty to ninety days written notice before the anniversary. Several platforms also price the renewal against your current page count or seat count rather than the number you originally signed for, so the renewal quote can rise without any change in the contract. Read the notice mechanism specifically — some require notice by a named channel, and an email to your account manager is not always sufficient to stop a renewal.

Should we tell the outgoing vendor we are evaluating alternatives?

Yes, and early enough for it to matter. A retention offer is a legitimate outcome of an evaluation, and the concessions that come at renewal — a fixed price across a growing page count, authenticated crawling included rather than upsold, an export commitment in the contract — may solve the original problem more cheaply than a migration does. What you should not do is start the conversation after the notice deadline has passed, when the only available answer is next year.

How do we keep our accessibility statement accurate through the change?

Write the statement so it does not name the tool. A statement that says monitoring is continuous, states the standard applied, gives a dated last-review and names a contact route stays true across a vendor change; one that names a platform and a score needs rewriting twice and invites a question about why the score moved. If your current statement or VPAT is hosted on the vendor's domain, copy it onto your own before cancelling, because any link you have given a customer or a procurement officer will otherwise break.

Take Your Own Reading Before the Account Closes

The one artefact that survives any migration is a dated scan you took yourself, in a format you control. Run one across the templates you care about before you give notice, and you have a neutral reference point that belongs to neither vendor — which is what makes the first report from the new platform readable at all.

Then judge the replacement on four structural lines: can it reach your authenticated pages, can a finding become a ticket, does the pricing unit match how your site grows, and will the output convince someone outside your team.