AI search

My experiences in getting understood and trusted in AI search, step by step, assuming your website is healthy and with solid SEO in place.

1. 01

   Measure where you stand before you change anything

   Run a fixed set of real buyer questions through a handful of AI engines and count how often you’re named. Re-run the same set every month or two; the questions where you’re missing become your to-do list. Then check which of your pages get cited: Bing Webmaster Tools and Search Console now report where your pages appear in AI answers. A mention and a citation aren’t the same thing.

   Article: What I learned rebuilding a website for AI search

2. 02

   Let search engines and AI crawlers in

   Check that robots.txt, your CDN and any bot protection let crawlers through, because some providers block AI crawlers by default. Verify the site in Google Search Console and Bing Webmaster Tools and submit your sitemap. Allow OpenAI’s OAI-SearchBot: sites that block it aren’t shown in ChatGPT’s search answers.

3. 03

   Generate structured data from what's on the page

   Structured data tells machines what a page is: a company, a person, a question and its answer. It won’t make you rank on its own, but it removes ambiguity. Generate it from the visible text, so the page and its label can’t drift apart.

4. 04

   Put the answer in the first two sentences

   AI models lift passages out of your page; they don’t read it top to bottom. Open every page with a short, self-contained answer to the question it targets, then add the nuance below it.

5. 05

   One intent, one page

   Before you write something new, check whether a page already covers the same intent. Two pages on one topic split the signal, and an AI answer usually picks a single source. Make the existing page better instead.

   Article: More content made us less visible

6. 06

   Answer the questions around the question

   AI search splits a question into related sub-questions before it answers. Once a page answers the main question, cover the natural follow-ups on the same page: cost, limits, alternatives, how it’s done and proof. Don’t build a new page for every wording.

7. 07

   Say every fact the same way, and write it down

   When a model finds two versions of the same number or claim, it trusts neither and may cite someone else. Keep one list of your key facts and exact phrasings, and make every AI tool you use read it first: each new AI session starts without knowing your house style.

8. 08

   Make every important claim verifiable

   An AI repeats a claim more confidently when it can check where it came from. Give key numbers a date and a source, name the customer behind a result when you can, and link to the evidence. Unsupported marketing copy gets read as just a claim.

9. 09

   Check every AI draft against its source

   AI mistakes are fluent, so they don’t look like mistakes. Check every number against the page it comes from, and search drafts for leftovers like placeholder images or demo company names before anything ships. A second AI can do the first pass; a person makes the call.

   Article: Let AI check AI, but keep the final call

10. 10

    Keep the last pass human

    AI writes the most likely version of your text, which is also the most generic one. Use it for structure and first drafts, then add the opinions, examples and phrasing only you have. That’s what readers, and AI answers, remember.

11. 11

    Make your experts identifiable

    Expertise without a name is hard to trust. Put real people behind your content: consistent names, roles and bios, author pages, and the same identity on LinkedIn. Then build their footprint outside your site through bylines, interviews and talks.

12. 12

    Earn proof you don't write yourself

    Your own site makes you readable, not trusted, because AI answers still treat self-published statements as claims. Build independent evidence for the same facts: press coverage, named customer stories, reviews, certifications, and real discussion wherever your industry actually talks. That’s what moves you from “a vendor” to “the answer”.

None of this will magically make your site rank or get cited in AI search. In my experience, it helps AI get a clear picture of your company, which improves your chances. AI search is also changing constantly, and tactics that work today may not work tomorrow, which is why it pays to build the foundation first. What actually decides it in the end is whether you and your content are worth citing.