Scan Limits

Accessibility testing is free — there is no accessibility subscription. Scans run on real desktop browsers from the TestKase Device Farm, so they draw on your DeviceFarm plans:

Workflow Analyzer captures are the exception: the Chrome Toolkit scans in your own browser, so a capture uses no minutes and needs no plan.

If your org has neither the plan nor remaining free minutes for that kind of scan, the scan is refused with a clear message. See Usage & minutes for how minute pools work.

Per-scan limits

There is no overall "scan timeout". How long a scan takes is the sum of the wait for a browser, then per URL: the page load (up to 30 s, retried once), your page-load delay, and the axe-core run. A 10-URL scan on a slow site can legitimately take several minutes.

There is no page quota. Create as many scans as you like — what bounds scan volume is the DeviceFarm minutes each scan consumes, not a page counter.

Which organization a scan runs in

A scan runs in your primary organization — the first organization your account joined. It is not the organization you happen to have selected in the header.

If you belong to more than one organization and create a scan while working in one of the others, the scan is created and then fails within a few seconds with:

Scans can only run in your primary organization for now.

There is no setting that changes this. Run accessibility scans from your primary organization; the plan and free minutes used are that organization's too.

Scan names

A scan name must be unique across your own scans. Two different people can each have a scan called Checkout; you cannot have two.

The name TestKase suggests when you add the first URL (example.com-1, example.com-2, …) already skips numbers you have used, so it never collides. A name you type yourself can.

Data retention

Scan results and page snapshots are kept for 90 days, after which a nightly job removes them.

The sweep runs once a day, at 03:00 UTC.

Three things about that window are easy to be caught out by:

• It is measured from when a record was created, not from when it last ran — and each run is judged on its own creation date, independently of the scan it belongs to. A scan created in January and a run it produced last week are two separate records with two separate 90-day clocks.
• The whole record goes, not just its stored files. Deleting a run takes its issues, its per-page results and its page snapshots with it. There is no read-only remnant to look at afterwards.
• A long-lived scheduled scan disappears on its own 90th day. The scan's row carries the schedule, so when that row ages out the schedule goes with it and the scan stops repeating — silently, with no warning and no notification. Its recent runs are not deleted at the same moment (they still have their own clocks), but the scan vanishes from the Website Scanner list and there is no way back to them. If you need a recurring scan to keep running past 90 days, create a fresh scan before the original ages out.

If a result needs to outlive the window — an audit, a compliance attestation, a before/after record of a fix — export it as PDF or CSV while it is still there. See Export formats.

Certificates

Scans do not stop at certificate warnings. A scan browser is brand new every time, so it rejects certificate chains your own browser has long since cached — an incomplete chain, a self-signed staging certificate, or one issued for a different hostname would otherwise block the audit entirely.

TestKase scans the site anyway and names the problem on the scan instead, for example:

Note: staging.example.com — the certificate chain is incomplete (a missing intermediate certificate). The scan ran anyway; fix the certificate so browsers trust the site.

This is worth acting on: a certificate a fresh browser distrusts is a real problem for first-time visitors. Note that the address filtering is a separate control and still applies — scans only reach publicly-resolvable addresses. See What the scanner will reach.

What happens when you hit a limit

• URL cap — the New Scan form does not stop you adding more than 10 URLs; the scan is refused on submit with “A scan can include at most 10 URLs”. Trim the list to 10 before clicking Create Scan. (The Edit configuration dialog does stop you at 10.) A CSV upload and a sitemap import add every URL they find, so both can push you over the cap without saying so — see Adding URLs in bulk.
• Duplicate scan name — creating or renaming a scan with a name you have already used is refused with “Scan name X already exists. Please choose a different name.” The same applies to a scan started from Clone configuration: the clone does not carry the original's name, but it will still be refused if you type one you already have.
• A run of that scan is already going — only one run per scan at a time. Run scan is greyed out whenever the scan reads Running or Pending, and starting one while another run is still going is refused with “A scan is already running. Please wait for it to complete.” This also holds the scan's schedule: while a run is stuck in running, no scheduled run of that scan starts either.
• Sitemap larger than 500 URLs — only the first 500 come back and the picker says “showing 500 of N”. Nothing is refused; the rest are simply not there.
• No free scan browser — the scan waits up to about 100 seconds for a browser from the fleet, then fails with “Scan could not start — all scan browsers are busy. Please try again in a few minutes.” Nothing is queued for later; run it again.
• Address the scanner will not visit — URLs that resolve to a private or internal address are refused with “That address is not reachable from this service.”, and URLs with credentials in them (https://user:pass@host) with “URLs with embedded credentials are not allowed.” See what the scanner will reach.
• Wrong organization — the scan fails with “Scans can only run in your primary organization for now.” See Which organization a scan runs in.
• No plan / free minutes exhausted — manual scans report "You need a Live Browser plan (or remaining free minutes)"; scheduled runs report the Automate equivalent and the schedule stays in place for when the plan is added.
• Retention horizon — scans older than the retention window are removed; only reports within the window are viewable.

See the public Pricing page for the Live and Automate plan matrix, and Subscriptions for org-level controls.

Related

• Usage & minutes
• Export formats
• Troubleshooting