More customers now ask an AI assistant "who is the best bakery near me?" before they ever open a search results page. If the answer names your competitor and not you, you never get the chance to win that customer. This guide shows how to track AI mentions week by week, so you can see whether AI is starting to recommend your business, and what moved the numbers.
Why one AI check is not enough
AI answers change. Ask the same question twice and you may get a different list of businesses. Ask again tomorrow and the order may shift again. That is normal. AI assistants build each answer fresh, often with a live web search, so small differences in the sources they find lead to different results.
This means a single check is a snapshot. It tells you where you stand today. It does not tell you which way you are heading.
Think of it like weighing yourself. One reading on one morning says little. The same scale, at the same time, every week, shows the real trend. AI visibility works the same way. You need the same questions, asked the same way, on a steady schedule.
If you have not run a first check yet, start with our guide on whether ChatGPT recommends your business. That first check becomes week one of your history.
What "tracking AI mentions" really means
Tracking AI mentions is not the same as rank tracking. A rank tracker records your position on a page of search results: position 3, position 7, page two. An AI answer has no fixed positions and no second page. You are either named in the answer or you are not.
So instead of a position, AI tracking reads the answer itself. For each question it records:
- Whether your business was named
- Where in the answer you appeared, if you were named
- Which other businesses were named instead
- What the answer said about you
- Which websites the answer relied on
Over several weeks, those records turn into a trend you can trust.
Recommended vs known
Two numbers matter most, and they measure different things.
- Recommended: AI names you when a customer asks for a business like yours without using your name. For example, "What are the best bakeries in Raleigh?"
- Known: AI can describe your business when asked about it by name. For example, "What does Golden Crust Bakery in Raleigh offer?"
Most businesses become known before they become recommended. If AI knows you but never recommends you, the gap is usually trust and sources. If AI does not know you at all, the first job is giving it clear facts to work with.
How IndexPedia AI Monitoring works
AI Monitoring asks the same customer questions every week to four AI assistants: ChatGPT, Perplexity, Gemini and Claude. The answers are collected live, with web search, and the results are saved. It costs $19 a month and you can cancel any time. If you cancel, your history stays in your account.
The questions are the same ones from your first AI Visibility Check. That free check asks one AI model with live web search (ChatGPT search). Monitoring then asks the same questions across all four assistants each week. There are two kinds of question:
- Discovery questions. Questions a new customer asks when looking for a business like yours in your city. These show whether you get recommended.
- Name questions. Questions about your business by name, such as what you offer, your prices and your hours. These show whether AI knows you and what it says.
Keeping the questions fixed is the whole point. If the questions changed every week, you could not tell whether a rise came from your work or from a different question.
- 1Run the free check
Enter your business name, city and what you do, and the first report becomes week one.
- 2Turn on monitoring
Switch on AI Monitoring for $19 a month, with no setup.
- 3Watch the weekly checks
The same questions are asked every week and each result is saved.
- 4Read the trend
Compare recommended, known and competitors side by side over the weeks.
What you see each week
Each weekly check adds a new row to your history. Here is what it covers.
Your weekly score
Each week's score combines how often you were recommended in discovery answers and how often AI knew you by name. Being recommended carries most of the weight, because that is what brings new customers. The score history shows every week side by side, so a slow climb is easy to spot.
Competitor movement
Every check records the businesses AI named instead of you. Over time you see who leads in your area, who is rising and who dropped out. If a competitor suddenly appears in every answer, that is worth a look. They may have earned a new review site listing or a press mention.
New sources
AI answers lean on websites: directories, review sites, local news, lists. AI Monitoring notes the new sources AI starts to rely on in your area. If a "best bakeries in Raleigh" list keeps showing up as a source, getting on that list is a practical next step.
What AI says about you
Wrong facts spread. If an AI answer says Golden Crust Bakery closes at 5 PM when you close at 7 PM, customers may not come. Reading what AI says each week helps you catch errors early, before they become the "truth" in every answer.
See if AI starts recommending you
A sample month for Golden Crust Bakery
Let's walk through a made-up example. Golden Crust Bakery is a small bakery in Raleigh, NC, with a website at goldencrust.example.
In week one, the owner runs the free check. AI knows the bakery by name and gives its address, but gets the hours wrong. In the discovery questions, AI names three other bakeries and not Golden Crust.
In week two, the owner publishes an AI Profile, a clear page of business facts made for AI tools. Nothing else changes.
In weeks three and four, the name answers start using the correct hours. Golden Crust appears in one discovery answer, near the end of the list.
These numbers are made up to show the shape of a trend, not a promise. Real results depend on your market, your competitors and the sources AI finds. The point is the method: one change, then several weeks of the same questions, so you can see whether it helped.
How to use the trend to grow
Data only helps if you act on it. Here is a simple routine that works for small businesses.
Make one change at a time
If you publish your AI Profile, update your Google Business Profile, get listed on two directories and redesign your homepage all in the same week, you will not know which one helped. Make one change, note the date, and watch the following weeks. Because the questions never change, a rise after a change is a strong sign that it worked.
Fix the facts first
If the "known" answers contain wrong details, fix those before chasing recommendations. Make sure your name, address, phone, hours and services match everywhere: your website, your Google listing, your social profiles and your AI Profile. AI tools trust facts that agree across many sources.
Follow the sources
Look at the sources AI uses for your discovery questions. If the same review site or local list shows up again and again, that is where AI is looking. Getting listed there, with accurate details, is often worth more than another blog post.
Make sure AI can read your site
None of this works if AI crawlers cannot reach your pages. Run AI Reach to test whether crawlers like OAI-SearchBot and PerplexityBot can read your site. Our guide on whether AI crawlers can read your website explains the common blocks.
Keep a simple change log next to your monitoring history: the date and one line on what you changed. After two months it becomes the most useful page you have.
Before and after: guessing vs tracking
Most owners check AI once, feel good or bad, and move on. Weekly tracking replaces that guesswork with a record.
You ask ChatGPT once, get a different answer than a friend got, and cannot tell whether anything you did made a difference.
The same questions every week, saved side by side, show whether you are named more often, who leads, and what changed after each update.
Daily, weekly or monthly?
People often ask why the checks run weekly and not every day. AI answers vary from one ask to the next. Checking more often mostly measures that noise. A daily chart would jump up and down without telling you much.
Monthly checks have the opposite problem. They are too slow. If a wrong fact appears, you would wait weeks to see it.
| How often | What you learn | Main problem |
|---|---|---|
| Daily | Mostly random variation | Noise hides the real trend |
| Weekly | Real direction over a few weeks | Needs a few weeks to settle |
| Monthly | Big changes only | Too slow to catch errors early |
Weekly is frequent enough to catch real changes and steady enough to show a reliable trend.
What AI Monitoring does not do
It is worth being clear about limits. No tool can promise that AI will recommend you. AI assistants decide for themselves, and their answers depend on many sources outside your control.
What monitoring does is show you the effect of your work. A clear, consistent profile, listings on trusted sites and a site that AI crawlers can read are what AI assistants rely on. Monitoring shows you, week by week, whether those efforts are moving the numbers.
You also do not need an AI Profile to use monitoring. It works on its own. But in practice, the profile is often what moves the numbers, because it gives AI clear facts to use. Read more in our guide on getting your business into AI answers with an AI Profile.
If your score drops one week, do not panic. Look at the next two weeks before you act. A single dip is often normal variation.
See if AI starts recommending you
FAQ
What is AI Monitoring?
AI Monitoring checks every week whether AI assistants recommend your business. It asks the same customer questions each week to ChatGPT, Perplexity, Gemini and Claude, live and with web search, and records whether you were named, how well AI knows you, and which competitors it named instead. Over time you see a clear trend instead of a single snapshot.
How much does AI Monitoring cost?
AI Monitoring is $19 a month, and you can cancel any time. If you cancel, your history stays in your account. The first AI Visibility Check, which becomes your week one, is free.
Which questions are asked each week?
The same ones as your first check: questions a new customer asks for a business like yours in your city, and questions about your business by name. Keeping the questions fixed is what makes week-to-week changes meaningful.
How is this different from rank tracking?
Rank trackers record your position on a page of search results. An AI answer has no fixed positions and no second page; you are either named or not. AI Monitoring reads the answer itself, including who was named and what was said.
How long until I see a change?
It depends on your market and on what you change. Because AI answers vary, give any change a few weeks of checks before you judge it. A steady rise over several weeks means more than one good week.
Can I see which competitors AI recommends?
Yes. Every check records the businesses AI named instead of you. You can see who leads in your area and whether that changes over the weeks.