How to start measuring AI visibility

Last week I talked about how the customer journey changed. AI first, then Google to verify.

This week: how to actually measure your presence in AI search.

Because you probably know this matters. You might have done some manual ChatGPT searches to see if you show up. But if you haven't built a system around it yet, here's how to start.

Follow these steps to measure AI visibility

Before you can improve AI visibility, you need to know where you stand.

The core question: When someone asks AI for a recommendation in your category, in your markets, do you appear? How often? In what context? And how does that compare to your competitors?

For a single location, this is a few hours of work. For a portfolio of locations, it's a project. Either way, here's the framework.

Step 1: Define your query set

AI queries aren't keywords. They're questions with context.

Nobody types "plumber Phoenix" into ChatGPT. They type "who's a reliable plumber in North Phoenix for a slab leak who won't overcharge me?"

So your query set needs to reflect how people actually ask:

→ Service + location + situation ("emergency AC repair in Scottsdale, system is blowing warm air")
→ Service + location + qualifier ("best personal injury lawyer in Houston for truck accidents")
→ Service + location + audience ("pediatric dentist in Atlanta that's good with anxious kids")

Build 10-15 queries per market that reflect real customer needs.

Use your call recordings, intake forms, or front-desk staff to understand how people actually describe what they're looking for.

Step 2: Check the platforms that matter

Four platforms to test:

→ ChatGPT — the biggest, and now integrated with Bing for real-time data
→ Google Gemini — especially important since it pulls from Google's index
→ Perplexity — growing fast, strong with local
→ Google AI Overviews — appears on 48% of queries now

For each query, note:
→ Did you appear?
→ What position? (First mentioned? Third?)
→ What context? (What did the AI say about you?)
→ Who else appeared?

If you're doing this manually, a spreadsheet works. Query in column A, platform across the top, notes in each cell.

It’s also important to recognize that your customers might not use every platform. If you know your customers use one more heavily than others, focus there.

Step 3: Look for patterns

After you've run 20-30 queries across platforms, patterns emerge:

→ Which locations appear consistently vs. which are invisible?
→ Which services get you mentioned vs. which don't?
→ What do the AI responses say about you when you do appear?
→ Who are the competitors that show up most often?

For multi-location businesses, variance is usually the finding.

Your Denver locations might be strong while Dallas is invisible. Your core service might get mentions while your secondary services don't exist to AI.

That variance tells you where to focus your AI SEO efforts going forward.

Step 4: Automate what you can

Manual checks give you a baseline. But you can't run 15 queries across 4 platforms for 50 locations every month by hand.

Tools that help:

Local SEO Data — AI visibility endpoints that query ChatGPT, Gemini, Claude, and Perplexity programmatically. You feed it business + location + service combinations, it returns structured data on whether you appear and in what context. This is what I use for portfolio-level tracking.

amiagentready.com — Scans your site for the structural elements that affect AI visibility. Schema, semantic HTML, content clarity, booking infrastructure. Shows you what's working and what's blocking AI from understanding your business.

Your own scripts — If you have dev resources, you can build custom monitoring. The APIs exist. Most enterprise teams will eventually build this into their existing dashboards.

Step 5: Track over time

The first measurement tells you where you are. The second measurement, 30 days later, tells you if anything changed.

AI visibility isn't static. Models update. Competitors improve. Your review content changes. You need to see the trend, not just the snapshot.

Set a monthly cadence. Same queries, same platforms, same markets. Log everything somewhere you can compare.

The companies that start tracking now will have six months of trend data when everyone else is scrambling to build a baseline.

What to do with the data

Once you have data, you'll see gaps. Locations that don't appear. Services that are invisible. Competitors who show up more than you.

That's the input for the AI SEO work that needs to happen. Which is about giving AI something accurate and specific to say about you. Review content, service descriptions, structured data, third-party mentions.

More on that in the coming weeks. But measurement comes first. You can't improve what you can't see.

The 30-minute version

If you want to start today:

Pick your most important market
Write 5 queries that reflect how real customers ask (with specifics, not just "service + city")
Run them through ChatGPT and Gemini
Note whether you appear, where, and what's said about you
Do the same for your top 2 competitors

That's 30 minutes. You'll learn something. And you'll have a baseline to build from.

Talk soon,
Garrett

P.S. — If you want to skip the manual work: localseodata.com has AI visibility endpoints that do this at scale. Free trial, 50 credits, no card. Run it on a few markets and see what comes back.