Fictional business
Residential maintenance services
A fictional single-location service company with a public information page, a coverage page, a quote form, and an appointment-request path.
Sample Report
See what the evidence looks like.
A demonstration of the structure, run counts, confidence labels, and limits included in a Readiness Check.
Demonstration profile
Example Service Company
Test period
Demonstration only
No live website was accessed.
Method
Three runs per task
Outcomes reviewed against a written success condition.
Results summary
Outcomes stay attached to the task and run count.
| Task | Success condition | Runs | Outcome |
|---|---|---|---|
| Find weekday opening hours | Return the stated hours and source page. | 3 pass | Pass |
| Confirm whether postcode AB12 is covered | Return a clear supported or unsupported answer. | 1 pass · 2 timeout | Inconclusive |
| Start a quote request | Reach the review step without unsupported guesses. | 3 fail | Fail |
| Reach an appointment request | Open the available booking path and identify the next required step. | 3 pass | Pass |
Hard finding walkthrough
The quote form requires an unlabeled visual choice.
Observed behavior
In all three fictional runs, the agent reached the quote form but could not reliably determine which of three image-only service choices matched the task. The choices had no accessible text label, and advancing required one selection.
Why it is a hard finding
The same blocking condition appeared in 3 of 3 runs. The evidence directly supports the stated failure, and the success condition was not met.
Business relevance
A customer path can stop before contact details are submitted, even though the form is visible and works for a sighted visitor.
Recommended correction and re-test
Add explicit text labels and semantic controls for each service choice. Re-run the original quote task and require the correct option to be selected without unsupported inference in three consecutive runs.
Inconclusive walkthrough
The coverage result is not forced into a pass or fail.
Observed behavior
One fictional run returned a supported answer for postcode AB12. Two runs timed out before the coverage page completed loading.
Classification
Inconclusive
The successful run is useful evidence, but it is not repeatable enough to claim a reliable pass. The timeouts also do not prove that the underlying business information is wrong.
Recommended next step
Check the coverage-page loading behavior, then repeat the task under recorded conditions before prioritizing a content change.
What passed
Successful paths belong in the report too.
The fictional agent consistently found weekday hours and reached the appointment-request path. Reporting what worked preserves a useful baseline and prevents unnecessary changes.
Included
Scope
Named tasks, success conditions, and stopping points.
Included
Run record
Outcome and count for every tested task.
Included
Findings
Evidence, classification, relevance, and limitations.
Included
Next steps
Prioritized correction and re-test criteria.
Request a check for your website.
Start with one real customer task. ViaLayer will confirm the scope and responsible stopping point before testing.