Start with a useful, indexable website that answers a buyer’s real questions and backs its claims with evidence. Google says established SEO practice still applies to AI search features, and no special AI schema guarantees inclusion.
- 01Make your service, customer fit, working model and commercial starting point explicit, and separate capabilities from demonstrated outcomes.
- 02A case study should state the problem, the intervention, what was measured, the period and the limits.
- 03Measure qualified enquiries, not just mentions. Assistant answers vary between runs and are not a rank.
The foundation is still a useful website.
Google’s guidance says established SEO practices remain relevant to AI search features. A page needs to be indexed and eligible to appear with a snippet. There is no special AI schema that guarantees inclusion, and technical eligibility does not guarantee a citation.
From useful page to possible citation
- 01Follow established SEO practiceStill relevant to AI search features
- 02Page is indexed
- 03Eligible to show with a snippetNo special AI schema guarantees inclusion
- 04May be citedTechnical eligibility is not a guarantee
Answer the questions a buyer actually has.
Our recommendation: make your service, customer fit, working model and commercial starting point explicit. A prospective client should be able to determine whether you solve their kind of problem without decoding a technology list. Separate your capabilities from outcomes you have actually demonstrated.
Publish evidence, not a wall of claims.
A useful case study explains the problem, the intervention and what was measured. State the measurement period and limits. If client approval or baseline data is missing, publish a narrower project overview instead of a result you cannot substantiate. Assistants and human buyers both need something they can verify.
Case study or narrower overview
| Aspect | Case study | Project overview |
|---|---|---|
| Use when | Client approval and baseline data are in place | Client approval or baseline data is missing |
| What it explains | The problem, the intervention and what was measured | A narrower account, without a result you cannot substantiate |
| State clearly | The measurement period and limits | Not covered in this note |
Give every important answer a home.
Use stable URLs, descriptive page titles, ordinary crawlable links and text present in the page itself. Keep service details consistent. Structured company facts, an XML sitemap and an article feed help organise the information; they are not promises of a higher ranking.
A practical first-week audit.
Choose five commercial questions a good prospect asks. Check whether your site answers each one clearly, with evidence and a relevant next step. Then inspect indexing and broken links. Record representative assistant queries over time, remembering that answers vary between runs and are not a deterministic rank.
- Can a buyer identify who you help?
- Can they understand the engagement model?
- Can they verify your claims?
- Can they find the right next action?
The first-week audit
- 01Five commercial questions a good prospect asksPass when: Each is answered clearly, with evidence and a relevant next step
- 02IndexingPass when: Important pages are indexed
- 03LinksPass when: No broken links
- 04Representative assistant queriesPass when: Recorded over time, not read as a deterministic rank
What to measure.
Connect discovery to qualified enquiries, not just mentions. Ask prospects how they found you and record that answer alongside campaign context. Review whether those enquiries match your intended customers. This is Noesis’s implementation guidance, not a claim that a search provider will recommend any particular business.
Sources & context
Sources checked 2026-09-10. Recommendations and illustrative scenarios are Noesis’s analysis, not claims of client results.