AI search visibility
When a customer asks an AI who to hire, are you in the answer?
More people now ask ChatGPT, Perplexity, Copilot or Google's AI Overviews instead of scrolling a page of links. Those tools can only name a business they can reach, read and trust. I make sure yours is one of them, starting with the technical side most marketing agencies never look at.
How an AI answer gets made
Four stages. A business can drop out at any one of them, and the first two are where most small-business sites fail.
Stage 1: the crawler has to reach your site
OpenAI, Anthropic, Perplexity, Google and Microsoft each send their own crawlers.
Before any of them can read a word, the request has to get past your
robots.txt, your firewall, your CDN and your host.
Where it fails: a robots.txt rule from years ago, a security product that now blocks AI crawlers by default, a bot-protection challenge the crawler cannot solve. Your site loads fine for people and returns nothing to the crawler, and nobody notices.
What I do: test from the crawler's side through every layer, then allow the crawlers you want, by name.
Stage 2: it has to be able to read what it gets
Many AI crawlers fetch the HTML and do not run JavaScript. Whatever is not in that first response may as well not exist.
Where it fails: site builders that draw the page in the browser, prices in images, services hidden behind tabs that load on click, no structured data saying plainly who and where you are.
What I do: make sure the words that matter are in the HTML, and add schema.org structured data for your business, services and common questions. This site works that way: every tool on it reads fully with scripts off.
Stage 3: it has to find you when the question is asked
When an assistant searches the live web to answer, it draws on search indexes and its own. If a page is missing from the index it uses, it is missing from the answer.
Where it fails: a site never verified with Bing, a stale sitemap, new pages that take weeks to be noticed, one page trying to rank for everything.
What I do: verify the site with Google Search Console and Bing Webmaster Tools, publish an accurate sitemap, turn on IndexNow, and give each service a page that answers one question well.
Stage 4: it has to trust you enough to name you
Assistants prefer sources that answer the question directly and agree with what other sources say.
Where it fails: an old address on one directory and a new one on another, a different business name on Google than on the site, no clear statement of what you do and where.
What I do: make your name, address, phone and services consistent across your site, Google Business Profile, Bing Places and the directories that matter, then measure whether it worked.
AI visibility check
Ten questions, about two minutes. Answer honestly, and use Not sure freely: it is the most useful answer here. Nothing leaves your browser.
What is included
- Baseline
- The fifteen to twenty-five questions your customers actually ask, put to ChatGPT, Perplexity, Copilot, Gemini and Google's AI Overviews, several times each. Recorded: who is named, who is cited, and what is said about you.
- Access and rendering
- Every layer between an AI crawler and your pages tested and fixed: robots.txt, firewall, CDN, bot protection, and whether the page has content before scripts run.
- Structured data
- schema.org markup for your business, each service, and your common questions, matched to your public records and validated.
- Index coverage
- Google Search Console and Bing Webmaster Tools verified, an accurate sitemap, and IndexNow so changes are noticed in days rather than weeks.
- Consistency
- Your name, address, phone and services made to match across your site, Google Business Profile, Bing Places and the directories that matter in your field.
- Answer pages
- A short list of pages to write or rewrite so each answers one customer question plainly. I can write them with you, or hand the brief to whoever writes for you.
- Re-measure
- The same questions asked again after six to eight weeks, with a written comparison, and AI assistant referrals separated out in your analytics so you can keep watching.
Fixed fee, quoted after a short call, like everything else here. No monthly retainer unless you want ongoing measurement.
This site is the example
Everything above is applied here. Here is what a crawler receives from this site, fetched live.
Choose one above to see exactly what a crawler receives from this site.
Questions
Is this different from regular SEO?
It overlaps. Both need a site that search engines can crawl and understand. The differences are where the work goes: whether AI crawlers specifically are let in, whether your content is readable without running scripts, structured data, consistent business details across the web, and measuring success by whether assistants name you rather than by a ranking position.
Can you guarantee we will show up in ChatGPT?
No, and nobody honestly can. Each assistant decides for itself what to say, and answers vary from one run to the next. What I can do is remove the technical reasons you are being left out, make your business easy to read and cite, and measure the result against the same questions before and after.
Do I need a new website?
Usually not. Most of the fixes are configuration, structured data, and a handful of pages that answer specific questions. If your site builder genuinely cannot serve readable HTML, I will tell you during discovery, which is when that is cheap to hear.
Will letting AI crawlers in mean my content trains their models?
Not necessarily. Several companies run separate crawlers for search and for training. OpenAI's OAI-SearchBot and GPTBot are separate, and Google-Extended controls use for Gemini training without affecting Google Search. You can allow the ones that put you in answers and decline the ones that only train, crawler by crawler.
How long before anything changes?
The technical fixes take effect when crawlers next visit, which is usually days to a few weeks. Whether assistants start naming you depends on them, so I measure at the start and again after six to eight weeks, and report what moved.
Why would a cloud engineer do this?
Because the most common failures are infrastructure failures. A firewall rule blocking a crawler, a CDN default, a page that renders only in the browser: these are found by someone who reads request logs and network configuration, not by someone writing blog posts.
Next step
Send me your website and the three questions you most want to be the answer to. I will tell you, briefly and for free, whether there is anything worth fixing.