Separate documented platform changes from fashionable claims. Accessible pages, useful answers and trustworthy business details still matter. Evaluate AI-search visibility alongside ordinary search and inquiries without treating a new calendar year as a new ranking formula.
Search now includes more ways to encounter an answer than a list of website links. That changes what a business should observe. It does not remove the need for useful pages, accurate information and a working customer journey.
Use current platform documentation to distinguish a documented change from a proposed tactic. Keep experiments labelled as experiments.
AI features use familiar foundations
Google's AI features guidance says supporting pages must be indexed and eligible for a snippet. It requires no special AI schema or new machine-readable text file.
Check whether important answers are accessible in the page, linked from relevant content and supported by evidence. Give each article a useful purpose. Do not create dozens of slight variations because a prompt tool can generate them quickly.
Search access and training access are different decisions
OpenAI documents separate search and training crawlers. Do not change a site's training preference merely to make it discoverable in ChatGPT search.
A successful request with a copied bot user-agent does not establish genuine crawler access. Security controls may treat verified bot traffic differently. Use available server evidence or native inspection tools, and label what remains unverified.
For the practical implementation sequence, read the AI-search visibility guide.
Reporting has changed; definitions still matter
Google's June 2026 announcement describes dedicated generative-AI visibility reports alongside overall performance reporting. Check the reports available in the actual property and read their metric definitions.
An impressions figure is not a separate count of AI-generated sales. Keep observed citations, identifiable referrals, valid inquiries and qualified leads distinct. Missing attribution should remain missing rather than being assigned to a fashionable channel.
Content quality is not a detector score
Google's spam policies address scaled content abuse. The useful review is whether the page adds value and makes accurate claims, not whether a detector guesses how a paragraph was drafted.
Read the article against its promise. A how-to should provide enough prerequisites, steps and validation for its audience. An analysis should separate evidence from interpretation. Verify author identity and remove experiences or results the author cannot substantiate.

Worked priority decision
Imagine a business has an elaborate AI text file but a service page with an outdated phone number, unsupported claims and no clear inquiry route. Correct the business facts and buyer journey first. Then verify crawling and indexing, improve the supporting answers and record search observations.
The text file may remain a useful navigation aid. It is not evidence that any search system has indexed or cited the business.
What to stop buying on faith
Be cautious with guaranteed AI placement, fabricated authority scores, bulk city pages and date changes presented as content refreshes. Ask what the proposed work changes, what source supports the claim and how its result can be observed.
Use service-aligned blog categories to organize distinct articles. Connect related guidance to relevant services without forcing every paragraph into a sales pitch. The content hub workbook provides the planning structure.
Match the evidence to the claim
| Observation | Supported conclusion | Still not established |
|---|---|---|
| A live page can be fetched | That request reached the current page | Indexing, genuine bot access or citation |
| A search report records an impression | The report observed visibility under its definition | A visit or qualified inquiry |
| An AI answer cites a URL | That sampled answer cited it at that time | Consistent citation across prompts or users |
| A referral reaches a service page | A recorded visit followed that path | Whether the visit caused a sale |
| A qualified inquiry is recorded | A prospect met the agreed qualification criteria | Attribution to an unobserved earlier interaction |
If the gap is access or page quality, use the SEO service. For an AI-discovery review, our AI Search Optimization service starts with the documented controls and observable results above.
Leave the review with an action
Choose one verified defect or useful improvement, name its owner and define completion. Download the search review workbook for a record of sources, checks and follow-up. Review search visibility and business response over time; neither is established by publishing the change.
Sources
- AI features guidancedevelopers.google.com
- separate search and training crawlersdevelopers.openai.com
- June 2026 announcementdevelopers.google.com
- spam policiesdevelopers.google.com
