What an AI Discovery Assessment Should Include

The assessment question

The question I want an AI discovery assessment to answer is straightforward: can relevant AI systems access, understand, verify, cite, compare and recommend the company for important customer situations?

That is broader than checking robots.txt or adding schema. It also requires understanding what the company offers, who it is for, which claims matter, what independent evidence exists, and whether leadership can measure improvement.

Seven areas to review

1. Accessible: crawler rules, hosting controls, rendering, status codes, sitemaps, canonicals, feeds and APIs.
2. Understandable: entity identity, service or product definitions, information architecture, visible text and structured data.
3. Credible: first-party proof, external coverage, reviews, authorship, profiles and consistency.
4. Citable: stable answer pages, original analysis, direct definitions, dates and source notes.
5. Comparable: audience, use cases, requirements, differentiation, limitations and alternatives.
6. Actionable: contact paths, forms, transactions, policies, handoffs and escalation rules.
7. Measurable: indexing, crawler activity, prompt inclusion, citations, referrals, leads and governance.

The assessment should also compare the company with a small group of realistic competitors. Visibility is relative. A technically solid page can still lose when another source provides clearer evidence.

What the deliverable should contain

The deliverable should explain what is working, where information is missing or inconsistent, and what to fix first. Technical findings and customer questions should support a short list of priorities, with responsibilities and measures agreed for the business. Timing depends on the scope and the people or partners available to do the work.

Every recommendation should state why it matters, what evidence supports it, who owns it, how it will be validated and whether it requires legal, customer or technical approval.

What it should not promise

An assessment cannot guarantee a citation or recommendation from an independent platform. It should not present llms.txt, schema or content volume as a shortcut. Google states that its AI search features use the same foundational SEO requirements and do not require special AI markup.

The goal is to make it easier for a customer to understand your business, with or without an AI assistant. The assessment should leave you with clear decisions and a practical next step.

About the author

Jeff Douglas

I bring 20+ years of ecommerce leadership experience, including revenue and budgeting responsibilities, customer experience and digital marketing leadership, Salesforce Commerce Cloud modernization and high-consideration retail. I advise businesses on customer experience, growth priorities, technology and AI-driven discovery.

Experience