How many prompts should I track per industry?

Start with fifteen to twenty-five prompts. That is enough to cover the four intent types without making your list unreadable. Grow from there based on what you see: if you lose structurally on one type of question, add variants. More than sixty prompts rarely gets read in practice.

Should I track my brand name too?

Yes, but keep brand prompts separate from the rest. Being named for your own name mostly says whether AI knows you, not whether AI recommends you. The prompts your customer uses without knowing you are the interesting ones: that is where you win or lose new customers.

Do prompts differ per AI platform?

The questions do not, the answers do. The same prompt often returns different sources in ChatGPT, Gemini and Perplexity, because the platforms select differently. So track the same set on several platforms instead of a separate list per platform. The difference between platforms is a finding in itself.

How often should I review my prompt list?

Once a quarter is enough for most businesses, plus right after a change in your offer. Add what is new, remove what never yields anything. A prompt that gives the same picture for six months uses measurement space you can better spend on a new question.

What if my industry is not listed?

The four intent types apply to every industry: the problem question, the comparison question, the local question and the selection question. Fill them in with the words your customer uses and you have a usable set within half an hour. The eight sets below mainly illustrate that pattern.

By Industry

September 5, 2026updated October 16, 2026GreatCompany

More customers from Google and ChatGPT?

Join the waitlist. Your free report shows what to improve on your site and who ChatGPT mentions instead of you.

Free reportNo payment details for your reportOpening in the US soon

Which prompts to track by industry? Eight example sets

Knowing how to choose prompts is not the same as knowing which ones to enter. Which prompts should you track lays out the pattern: five kinds of prompts, how many you need, what to watch for. This article makes it concrete per industry, with sets you can copy and adjust.

The pitfall when setting up a prompt tracker is almost always the same: people enter their own brand name, see they get named, and conclude all is well. But someone who already types your name knows you. The question that counts is what AI tells someone who does not know you yet.

Four questions that come back in every industry

Before the examples: the sets below are all built from the same four types. Know this pattern and you can build a set for any industry yourself.

The problem question. The customer describes their problem, not the solution. They do not yet know which kind of business they need. This is the earliest moment you can be named, and usually the moment with the least competition.
The comparison question. The customer knows which solution they want and wants to know who provides it. Here your competitor literally sits next to you in the answer.
The local question. The customer adds a city or region. For anyone with a service area this is the most important type, and the one that moves fastest.
The selection question. The customer asks what to look for. They are not looking for a business but a criterion. Whoever supplies the answer here is often named as a provider in the same breath.

A usable set has all four. A set made only of comparison questions measures the end of the journey and misses where your customer starts.

Healthcare

In GEO for healthcare the problem question dominates: people describe a complaint before they know which provider they need.

I have had lower back pain for weeks, who should I see?
What is the difference between a physical therapist and a chiropractor?
Physical therapist in [city] with short wait times
What should I look for when choosing a physical therapy practice?
Does my insurance cover physical therapy without a referral?

Note the last one. Coverage and wait time questions are often the most asked in healthcare, and the answer is a fact AI likes to cite. If that fact is not on your site, AI cites someone else.

Online stores

In GEO for e-commerce it is about product selection. AI assistants increasingly build a shortlist, and you want to be on it before a brand name comes up.

Which [product category] is best for beginners?
Best [product] under $100
What is the difference between [product type A] and [product type B]?
Where can I buy [product] with fast shipping in the US?
What should I look for when buying [product]?

The price limit in the second prompt is not a detail. Answers to price questions are often more concrete and change faster, so there is gain for whoever puts their prices readably on the page.

SaaS

In GEO for SaaS the comparison question is king. See also how the comparison pages on this site are set up: they are built for exactly this moment.

Best [category] software for small business
Alternative to [competitor]
How much does [competitor] cost per month?
[Competitor A] or [competitor B], what is the difference?
Which [category] tool integrates with QuickBooks?

Track your competitor’s name, not just your own. Appear in the answer to “alternative to [competitor]” and you are there exactly when someone wants to switch.

Restaurants

In GEO for restaurants almost everything is local, and almost everything has an occasion.

Where can I eat well in [city]?
Restaurant in [city] for a business lunch
Family-friendly restaurant [city] with a patio
Where can I eat gluten-free in [city]?
Fun restaurant for a birthday in [area]

The occasion is the distinguishing word, not the cuisine. “Business lunch”, “birthday” and “family-friendly” return very different answers than just “restaurant”, and those are exactly the questions people ask an AI instead of a map.

Real estate

In GEO for real estate the gain is in process and cost questions, not in the listings themselves.

How much does a realtor charge to sell my house?
Listing agent or for sale by owner, what is smart?
Best realtor in [city]
How long does it take to sell a house in [area]?
What should I look for when choosing a buyer’s agent?

Commission questions are the equivalent of the coverage question in healthcare: frequently asked, answerable with facts, and rarely answered well on the agent’s own site.

Home services

In GEO for home services the problem question often comes in as a breakdown, with urgency behind it.

My furnace is not working, what now?
How much does it cost to replace a furnace?
Heat pump or furnace, what is smarter for an older home?
HVAC company in [city] that can come today
Which tax credits or rebates can I get on a heat pump?

Rebate and cost questions change every year. That is a downside for your archive and an upside for your visibility: a current answer is more likely to be cited than an outdated one. See also content that AI cites.

Law firms

In GEO for law firms the client describes a situation, not a practice area.

My employer wants to fire me, what are my rights?
How much does an employment lawyer cost?
Do I need a lawyer for a divorce?
Employment lawyer in [city]
What should I look for when choosing a lawyer?

The third prompt is the most interesting. “Do I need a lawyer” is a question where an honest answer is sometimes no, and exactly that kind of answer gets cited well. See E-E-A-T for AI authority.

Accountants

In GEO for accountants the business owner asks a question about their own situation.

Should I form an LLC or stay a sole proprietor?
How much does an accountant cost for a small business?
Can I do my own bookkeeping or do I need an accountant?
CPA in [city] for freelancers
Which expenses can I deduct as a business owner?

The last one is a classic citation question: factual, frequently asked and changing every year. Whoever gives the clearest answer gets named, also by people not yet looking for an accountant.

What you do with the results

Entering a set is step one. The set only pays off when you get three things out of it.

Where you are missing. Not being named for a problem question almost always means the answer to that question is not on your site in usable form. That is a content task, not a technical one.

Who is there instead. The names that appear instead of you are your real competitors in AI, and that is regularly a different list than your competitors in Google.

What is said about you. Being named is not the same as being named well. The wording you appear in shows what AI thinks it knows about you.

How to turn those three into a monthly report is in measuring AI visibility. Want to understand the difference with classic rankings first? Read prompt tracker vs rank tracker.

Start with fifteen prompts

Do not wait for the perfect list. Take your industry’s set above, replace the city and product terms with yours, and add five questions your customers really ask you. That is fifteen to twenty, and that is enough for a first measurement.

What you see after that adjusts your list better than any scheme you devise in advance.

The prompts above are a start. The details per trade are on GEO for accountants, GEO for law firms, GEO for restaurants, GEO for home services, GEO for real estate, GEO for SaaS, GEO for e-commerce, GEO for healthcare and GEO for salons. Longer pieces: B2B, financial services, education and travel. The hub is GEO for local businesses.

Frequently asked questions

How many prompts should I track per industry?: Start with fifteen to twenty-five prompts. That is enough to cover the four intent types without making your list unreadable. Grow from there based on what you see: if you lose structurally on one type of question, add variants. More than sixty prompts rarely gets read in practice.
Should I track my brand name too?: Yes, but keep brand prompts separate from the rest. Being named for your own name mostly says whether AI knows you, not whether AI recommends you. The prompts your customer uses without knowing you are the interesting ones: that is where you win or lose new customers.
Do prompts differ per AI platform?: The questions do not, the answers do. The same prompt often returns different sources in ChatGPT, Gemini and Perplexity, because the platforms select differently. So track the same set on several platforms instead of a separate list per platform. The difference between platforms is a finding in itself.
How often should I review my prompt list?: Once a quarter is enough for most businesses, plus right after a change in your offer. Add what is new, remove what never yields anything. A prompt that gives the same picture for six months uses measurement space you can better spend on a new question.
What if my industry is not listed?: The four intent types apply to every industry: the problem question, the comparison question, the local question and the selection question. Fill them in with the words your customer uses and you have a usable set within half an hour. The eight sets below mainly illustrate that pattern.

See whether ChatGPT mentions your business.

Join the waitlist. When GreatCompany opens in the US, you get your first report right away: who ChatGPT recommends in your market and what to improve on your site.

Free to join. When we open in the US, your free AI visibility report is first in line.

Curious what AI crawlers see on your site? About the technical scan Pro $49, you do it yourself · Or automate your SEO for $99, built for WordPress