Search the guides

Google Ads

Audit a Google Ads page from both sides

Combine a real visitor walkthrough with a server-side access check without pretending that browser emulation is Google's review.

One destination, two useful views

Your laptop opens the page, but Google Ads reports a destination problem. Test the customer's task and the server response separately. One working visit does not explain what happened to another request.

Choose the exact advertised action and URL. A homepage check says little about an old product path used by the campaign. Define success before testing: for example, one valid quote request received.

Walk through as a new visitor

Start on a phone in a fresh session, without administrator access. Read the ad, then the page. Can you explain the offer and its price without knowing the business already?

Complete the task using an agreed test procedure. Label test enquiries and clean them up. For orders or bookings, use the supported test process so you do not consume real stock or appointments.

  • Read the first screen with no prior knowledge of the business.
  • Check essential terms before the primary action.
  • Open navigation, contact and policy links.
  • Submit invalid and valid form data.
  • Confirm both the visible success state and the destination record.
Two approachesWhat each view can establish

New customer

  • Can recognise the advertised offer.
  • Can finish and confirm the action.

Investigated request

  • Receives an accessible response.
  • Reaches the intended destination through its redirects.

A working customer session and an accessible automatic request answer different questions.

Inspect what the server received

Ask the operator to find the request by its path and time in the access log—the server's record of incoming requests. Check its response and redirects. For a review failure, investigate the relevant request rather than disabling protection across the site.

AdsBot is Google's automated requester for advertising-related checks. Googlebot serves Search. A Search Console result therefore does not replace investigation of the exact error in the advertising account.

A user-agent is the request's text label for its client. Changing it can expose a rule tied to that label, but does not create a verified Google request. Verify the source by the service's documented method before granting an access exception.

A worked example: the mobile challenge never ends

Hypothetical case: desktop booking works, but a fresh phone visit stays on an extra security check. The account reports a destination error around that time. These observations may be related; they do not yet establish the same cause.

Record the phone's route and time. Find which service returned the security screen and whether the request reached the booking application. If it stopped earlier, changing the form will not repair the blocked step.

Diagnostic routeTurn the symptom into a test
  1. Observe

    A fresh phone visit stays on a challenge.

  2. Locate

    Match the time and path in access records.

  3. Isolate

    Find the component that stopped the request.

  4. Repair

    Adjust the demonstrated failing rule.

  5. Retest

    Repeat the phone task and inspect the reported issue.

The example separates an observed phone failure from the platform's independently reported issue.

After a narrow fix, repeat the phone journey and the previously working desktop journey. Confirm receipt of the request. Check the platform's diagnostic separately before claiming its reported problem is resolved.

Use a small evidence matrix

Keep evidence tied to the failure. In the hypothetical phone case, record the blocked address, security response and device state. 'Google dislikes the site' gives an engineer no step to reproduce.

Assign each issue to the person who can change it: offer wording, access rules or form delivery. Include reproduction steps so the next owner can verify the observation rather than restart the investigation.

Choose a test without overstating it

Treat a changed client label as a narrow experiment. The network origin and other request conditions remain different. Verify the suspected rule in the configuration; the experiment is not a copy of the platform's entire review.

Inspect the content as well as the response status. A maintenance screen with a success status is not the offer. Conversely, a screenshot of a working final page cannot rule out a redirect loop on another route.

State what remains unknown. For example: the fresh phone visit now works, but the earlier platform request is missing from retained logs. That record is more actionable than 'everything fixed'.

Keep a small audit record

For each test, record the address, device or request context, time, task and result. Keep the evidence needed to reproduce it; remove personal and secret values before sharing.

Name who verifies the page and who verifies the received business record. Otherwise the design and server response can both be approved while nobody checks whether the enquiry arrived.

Retest the smallest changed path

Repeat the failed case and one case that worked before. This catches a fix that restores mobile access but breaks desktop routing. Include the ad URL and tracking settings in the record because they can change separately from the page.

If the platform still reports an issue, use its review process with the collected evidence. Avoid unrelated rewrites just to provoke another result. The audit should leave a narrower, reproducible problem even when the internal review decision is not visible.

Questions answered

Does changing my user agent recreate AdsBot?

No. It changes one request label. Use it to investigate label-based rules, not to claim that you reproduced Google's network origin or review process.

ADSTRAFFIC / ENAll guides →