Before the AI Visibility Campaign: A Google Search Preflight for Agencies
Check the foundations in a sensible order before commissioning more content or markup: access, indexing, page identity and the buyer's actual question.
Before an agency commissions another “AI-optimised” page, it should establish whether the page already meant to answer the customer's question is ready for Google Search.
Google says SEO practices remain relevant to AI Overviews and AI Mode. Supporting pages must be indexed and eligible for a Search snippet; inclusion is not guaranteed. Google's AI features guidance.
That creates a sensible starting point for delivery. Run a short preflight on a small set of commercially important pages, then decide which problem deserves the next hour of work.
The sequence below is a proposed agency checklist. It is specific to Google Search; it should not be presented as the access policy for every AI platform.
Priority 1: check whether the page can enter the process
Google's minimum technical requirements include access for Googlebot, a successful HTTP response and indexable content. Google Search technical requirements.
For each selected URL, record the response, access controls and indexing instructions. Investigate an unexpected login wall, error response or blocking rule before making content refinements.
Do not remove a restriction simply because it prevents search visibility. It may protect private information or reflect a deliberate business policy. Ask the site owner what the page is intended to do, then obtain approval for any change.
Classify the result as an intentional restriction, an accidental blocker or no blocker found. This keeps the preflight from becoming an indiscriminate instruction to expose more content.
Priority 2: confirm which URL you are investigating
Clients often send a convenient URL rather than the version that search systems treat as the main page. Check whether you are dealing with a duplicate, redirect, filtered variation or genuinely separate page.
Google describes redirects and canonical annotations as strong canonicalisation signals, while sitemap inclusion is weaker. These communicate preferences rather than providing a simple command that every chosen URL must be selected. Google's canonical URL guidance.
Your working sheet should distinguish the business's intended URL from the currently observed canonical. If they differ, diagnose the reason before changing tags across a template.
The practical deliverable is a clear page identity. Everyone on the account should know which URL the content brief, measurement record and release ticket refer to.
Priority 3: separate the live page from indexed evidence
Use Google's indexed report and live test for different questions: processed information versus the page currently available. URL Inspection documentation.
Record when each observation was made. A recently corrected page and an older indexed version can coexist in the evidence without anyone having done something wrong.
Avoid marking the issue resolved because the live page looks right in an ordinary browser. Write a narrower status, such as “live correction checked; indexed state awaiting confirmation.”
Priority 4: read the page as a prospective customer
Now ask a colleague unfamiliar with the account to identify:
- What is being offered?
- Who is it appropriate for?
- What does it include and exclude?
- Which statements require supporting evidence?
- What should a qualified visitor do next?
This is an editorial test, not a claimed ranking formula. It reveals whether the page actually answers the question the agency wants it to serve.
If the answer requires account knowledge that is absent from the page, get that knowledge from the business. Do not fill the gap with plausible wording and treat the result as verified expertise.
Priority 5: choose the next action, not a giant score
End the preflight with one of four recommendations: resolve an access problem, investigate page identity, improve the answer or continue observation.
Attach the evidence and the owner needed for that next action. A score can summarise a review, but it should not hide a critical blocker behind several easy passes.
After an approved change, repeat the checks relevant to that change. Keep implementation status separate from observed AI appearances and ordinary search performance. That gives the client a defensible explanation of progress even when visibility has not moved.
Kloovy describes human review before its agent executes changes. Use that decision point to check that the proposed fix addresses the evidence, rather than simply adding more tasks. See how Kloovy works.
Sign up for Kloovy to explore an approval-led workflow for the fixes your preflight identifies.