Skip to main content
Journal

We couldn't find our own studio in AI search

· 8 min read · By Max Muncy

Three weeks after launching the site I searched ChatGPT for “Mainsail.” I wanted to see what it would say about us.

It didn’t know we existed.

I tried “web design in Greenville NC” — nothing. “Local SEO Greenville” — nothing. “Mainsail Greenville” — a confident, wrong answer about a different company in a different state. Even when I explicitly said “I’m looking for a web design studio called Mainsail based in Greenville, North Carolina,” the response was a shrug dressed up as a paragraph.

I sell this service. I run it for clients. My site ranks for nothing.

This is embarrassing. It’s also useful — because what I’m feeling right now is the exact problem every small business owner I meet describes in their own voice. “I paid someone to build my site three years ago and I still can’t find it online.” That’s me. The local SEO guy. In real time.

So: I’m going to fix it in public. This post is week zero. I’ll update it monthly with screenshots, what moved the needle, and what didn’t. If you’re a small business owner watching your own site fail to show up in ChatGPT, you get to watch the exact fix play out on someone else’s domain before you pay anyone a cent to try it on yours.

What’s actually broken

After an hour of diagnostics, the problem isn’t one thing. It’s four, in order of severity:

1. The site is three weeks old, and AI search is lazy

ChatGPT’s web browsing uses Bing. Bing takes four to eight weeks to fully crawl, index, and start trusting a new domain. Even once you’re in the index, the AI won’t cite you until you’ve earned some signal. The cold start is real and it’s a little unfair, but it is what it is. A fresh domain without inbound links and without traffic history has zero confidence score.

This is the biggest single reason I’m not showing up. I can’t short-circuit the clock. I can shorten it by giving Bing + Google every signal I’ve got, which is what the rest of this list is about.

2. There are six other companies named “Mainsail”

This one I didn’t see coming.

  • Mainsail Partners — a San Francisco private equity firm investing in B2B software. They have a lot of PR, a lot of press releases, and a lot of backlinks. When ChatGPT searches “Mainsail,” it finds them first.
  • Mainsail Marketing — a Tampa-based marketing agency. Also not us.
  • Mainsail Management — multiple property management companies use this name.
  • A national web design and SEO agency also called Mainsail, which is the one that directly competes for the brand phrase I pay the most attention to.
  • Plus “mainsail” as a generic sailing term, which shows up in nautical content and pollutes the entity cluster further.

When an LLM searches for “Mainsail,” it fuses all of these into a single ambiguous entity. The only way to un-fuse is aggressive disambiguation: the brand name must always be anchored to a specific location and service (“Mainsail, a web design studio in Greenville NC”) in every public signal — my own site, my Google Business Profile, my Wikidata entry (more on that below), my social bios, my backlinks, everywhere.

If you share a name with a larger company in any industry, this applies to you too. Don’t fight for the single-word brand — fight for the qualified phrase.

3. Bing has no idea we exist

ChatGPT uses Bing. Bing Webmaster Tools is separate from Google Search Console. I had Google Search Console set up; I hadn’t even started the Bing side. Setting it up this week.

Most people assume “rank on Google, rank everywhere.” Not true. For AI search specifically — which is where my pipeline is heading — Bing is upstream. Miss Bing, miss ChatGPT, miss a big piece of the future.

4. Our entity had gaps at the structured-data level

Running our own audit tool against mainsail.design surfaced specific issues. Our existing Organization schema had no street address or postal code. It was missing geo coordinates, required by Google for LocalBusiness rich results. It didn’t declare knowsAbout — the structured topic signal that LLMs use to match “who does X in Y.” It wasn’t declared as both ProfessionalService and LocalBusiness, which means Google Knowledge Graph couldn’t fully connect us to a local-business entity profile.

All fixable. Most of it I already fixed this week. What I learned from doing it for myself will end up in the client playbook.

The plan, in two tracks

Everything above sorts into two buckets:

Track A: the off-site stuff — the identity signals that live outside my website. Nobody ranks without these; you can have the cleanest schema in the world and still not show up if there’s no off-site confidence signal pointing back at you.

Track B: the on-site stuff — the structured-data, disambiguation, and content signals that live on my own domain.

On-site is done in a week. Off-site takes a month because it depends on platforms actually crawling + indexing + trusting us.

On-site (done)

  • LocalBusiness schema alongside ProfessionalService, with street address (to the extent a remote studio has one), postal code, geo coordinates, and opening hours. These are all required for Google rich results to fully activate. Auto-generated from a single studio.ts config, so as the business grows the schema moves with it.
  • knowsAbout array — explicit topic signals on every page. Ten things we’re declaring competence in. LLMs use this to answer “who does X in Y” without having to infer.
  • alternateName entries for “Mainsail Greenville” and “Mainsail Design” — the location-qualified phrases that disambiguate us from the other Mainsails. These go in the Organization schema and the meta tags.
  • Title-tag sweep. Every page now has “Mainsail (Greenville, NC)” or similar as part of the title phrase, not just in the tagline. LLMs index title tags heavily for entity matching.
  • A new brand disambiguation page that explicitly says: we are this, they are that, here is the line. Ugly-simple on purpose — written for the crawler first, human second.
  • Meta-tag verification hooks for both Bing Webmaster Tools and Google Search Console so the verification flow is one paste.

Off-site (this week, me at a keyboard)

  • Bing Webmaster Tools — add the site, verify, submit the sitemap, request a scan. This is the single biggest lever for the “ChatGPT can’t find us” problem.
  • Bing Places for Business — the Bing equivalent of Google Business Profile. Most people skip this. We literally work in AI search, so we can’t.
  • Wikidata entry — one of the biggest AI-entity signals available, and free to create. A Wikidata item is a structured record that every major LLM trains on. Creating one for “Mainsail (Greenville design studio)” explicitly tells the training set “this is a distinct entity.”
  • Yelp + Apple Business Connect — citation sources Bing and Google cross-reference. Takes five minutes each.
  • Local directory listings — Greenville Chamber of Commerce, NC design community references.

I’ll publish each submission as it lands, with what I filled in and why.

The tracker

Starting today, I’ll update this post monthly with:

  1. A screenshot of what ChatGPT says when asked “Do you know about Mainsail, a web design studio in Greenville, NC?”
  2. Same for Perplexity and Google AI Overview.
  3. A count of how many AI-search queries return Mainsail as a citation.
  4. What I shipped that month.

Baseline (2026-04-23): 0 AI citations. ChatGPT doesn’t recognize the entity. Perplexity returns a generic “I don’t have information about a specific Mainsail company in Greenville” fallback.

I’ll come back and edit this post every month with the new numbers. If I’m not moving, that’s data too; I’ll write up what didn’t work.

What to take from this

If you’re a business owner reading this thinking “my site isn’t showing up either,” three honest things:

  1. You might be right on schedule. If your site is newer than a year old and doesn’t have 20+ inbound backlinks or meaningful search traffic, AI search doesn’t know you exist yet. That’s not broken; that’s how the system works. You can accelerate the timeline but you can’t skip it.

  2. The biggest lever is off-site. GBP, Bing Places, citation sources, Wikidata, directory listings. People hire agencies (us included) to “fix SEO” and spend three weeks optimizing meta descriptions. Meta descriptions matter. Off-site identity matters ten times more for local AI search.

  3. Name ambiguity compounds the problem. If there’s a bigger company in your industry with your name, or near-enough to yours, you’ll lose the AI-search entity fight by default until you explicitly disambiguate in both your on-site and off-site signals. “Mainsail” taught me this. If you’ve got a business with a common name, we can talk about this for hours.

The next update lands around 2026-05-23. I’ll paste screenshots, ChatGPT transcripts, and a real tally.

If you’re dealing with the same problem on your own site and want a look at what we’re doing, the free audit will tell you what’s missing. No pitch, no email gate beyond what we already had — just the score plus a written report. That tool runs on the same engine I used to audit our own site this week.

— Max

Book Email Call