How to Get Found in AI Search

When someone asks ChatGPT for the best plumber in Nashville, an answer comes back — and being in it is a different job from ranking on Google. This guide covers what the evidence actually supports, what's being oversold, and what a business owner can do about it. Also called AEO and GEO. Same work, three names.

The short version: content structure, named authors, original data, and third-party mentions. Not schema markup, and not llms.txt.

Who this guide is for

Any business whose buyers research before they call — legal, healthcare, B2B, remodeling, anything considered. If your demand is purely emergency and you aren't dominating the map pack yet, fix local SEO first. That's still a bigger lever.

Why does this matter now?

Because a meaningful share of research that used to start with a Google search now starts with a question to an assistant — and assistants return a shortlist rather than ten links.

If you're not on the shortlist, you're not considered. There's no page two to be on.

The useful part: almost nobody has started. Roughly 16% of businesses track their AI search visibility at all. That's an unusually open window, and it won't stay open.

What actually works

These are the tactics with evidence behind them.

Notice what these have in common. None of them are technical tricks. They're all about making content easy to extract and easy to trust — which is what these systems are actually doing.

What's being oversold

Two things, and you'll hear both pitched hard.

Schema markup. Google stated in May 2026 that structured data isn't required for AI Overviews or AI Mode, and a 2026 study found JSON-LD produced no measurable lift in AI citations. Implement it anyway — it drives rich results, breadcrumbs and map pack eligibility in regular search, and it's cheap. Just don't buy it as an AI play.

llms.txt files. Google has said machine-readable AI files aren't needed. Ship one if you like — it takes ten minutes and costs nothing — but assign it no expected return.

If an agency is charging you a premium for either of these as AI search work, you're paying for classic SEO with a new label.

The part nobody wants to hear

Roughly 85% of brand mentions in AI search originate from third-party pages, not your own website.

Which means your site alone cannot win this. What actually feeds the answers:

• Directory and review platforms — Google Business Profile, Yelp, industry directories, Clutch, G2 depending on your category
• Local and trade press — being covered, quoted, or mentioned
• Podcasts and interviews — where you're introduced by name and business
• Expert sourcing platforms — being quoted as a source in someone else's article
• Associations and professional bodies — member listings, contributor pages

This is unglamorous, slow relationship work. It's also most of the outcome, and it's why on-page-only "AI SEO" packages underdeliver.

How do you measure it?

There's no Search Console for AI search, and anyone claiming precise attribution is guessing. What works is a consistent manual method.

Share of Model:

1. Write 25 prompts across four types — discovery ("best HVAC company in Nashville"), comparison ("X vs Y"), evaluation ("how much does law firm SEO cost"), and implementation
2. Run them weekly in fresh sessions across ChatGPT, Perplexity, Google AI Mode, Claude and Gemini
3. Log whether you're mentioned, who else is, how prominently, and in what tone
4. Score appearances divided by total prompts
5. Allow 4 to 6 weeks before the trend line means anything

Fresh sessions matter — if you're logged in and have asked about your own business before, you'll get flattering results that mislead you.

Also watch branded search volume, direct traffic anomalies, and the "how did you hear about us" field, because those move when AI visibility moves.

What to actually do, in order

1. Build a real author page. A named person with credentials, linked from every article. Single highest-leverage fix on most sites
2. Restructure your commercial pages — answer-first opening, question headings, tables, FAQ section
3. Clean up your entity. Business name, address, phone and category identical everywhere. Ambiguity makes systems hedge instead of recommending you
4. Publish something only you can — your own cost data, close rates, or market benchmarks. This is what earns third-party citations
5. Start the off-site work. Directories, press, podcasts, expert sourcing
6. Set a quarterly refresh cycle. Put it on the calendar or it won't happen
7. Baseline your Share of Model before any of this, so you can tell whether it worked

What does this cost to do yourself?

The on-page work is genuinely DIY. Restructuring a page to lead with a direct answer and adding an FAQ costs nothing but attention.

When should you hire someone?

Do it yourself if you have a handful of important pages and time to restructure them. Steps 1 through 3 above are the bulk of the on-page benefit and none of them require an agency.

Consider hiring when:

• The off-site programme is the constraint. Directories, press, podcasts and expert sourcing take sustained relationship work, and it's where most of the outcome lives
• You want original research published and don't have the capacity to produce it
• You want measurement done properly and consistently, rather than sporadically
• You're in a competitive category where the window is closing faster

Be sceptical of anyone selling AI SEO who leads with schema markup or llms.txt. Ask them what percentage of AI citations come from third-party sources, and what their off-site programme actually involves.

If you'd rather hand it over, that's our AI SEO service at $1,500 a month, or a standalone visibility audit at $1,500.

Frequently asked questions

---

Written by Jared DeValk, founder of Nashville Digital. Last reviewed September 2026. AI search changes faster than any other channel — this guide is updated quarterly.

---