No. AI can't tell you the right price, because it doesn't know your costs, your capacity or what your customers will pay, and the "market rates" it quotes are often out of date or invented. What it can do is make pricing research faster: collecting published competitor prices, building your cost floor, and designing a price test you then run on customers.
The test for any AI pricing answer is whether you can trace each number. A figure with a source you can open, such as a competitor's current price page or your own cost sheet, is research. A figure that arrives as "the typical range is $70 to $120" with nothing behind it is a guess in a confident voice, and it's the most common way owners get misled when they ask a chat assistant about price.
What a chat assistant actually knows about prices
When you ask ChatGPT, Claude or Gemini what to charge, the answer comes from one of two places, and it matters which:
- Training data. Without web search, the model answers from what it absorbed during training, which stops at a cut-off date and mixes prices from many years, markets and currencies. It has no way to know which of those resemble your situation, so it blends them into a smooth range.
- Live web search. With search switched on, it reads pages now, but only published prices. Many service businesses don't publish prices, and the ones that do may show "from" prices, old pages or offers. The model can't see discounts, packages, waiting lists or what's included, and it usually summarises only a handful of pages.
Neither source contains the three things that decide your price: what it costs you to deliver, how full your diary is, and what your customers value enough to pay for.
Five pricing questions AI can't answer for you
- What does one unit of your work cost you? Your rent, insurance, wages, materials and unpaid admin time are in your accounts, not in the model.
- How much spare capacity do you have? A three-week waiting list says you can charge more. Half-empty afternoons say something else. Only your booking system knows.
- What will your customers pay? Willingness to pay depends on who your customers are and what they compare you with. You find it by asking and testing, not by averaging web pages.
- Where do you want to sit? Cheapest in town, middle of the pack, or premium with a reason. That's a strategy choice, not a calculation.
- What happens to demand if you change the price? The honest answer for a small business is "we'll find out", which is why the test section below matters more than any estimate.
Asking for "the going rate": a sample answer, taken apart
Here is the kind of prompt owners type, and an illustrative answer of the type it produces:
What should an osteopath charge for a first appointment?An initial osteopathy consultation typically costs between $70 and $120, depending on location, the practitioner's experience and the length of the session. Follow-up appointments are usually slightly cheaper. Many clinics offer introductory discounts to attract new patients.
What's wrong with it: there is no source for either number; the range is so wide (the top is 70% above the bottom) that it can't guide a decision; it doesn't say whether it compares 30-minute or 60-minute sessions; and it may be describing a different market or several years ago. The suggestion about introductory discounts is generic advice that may be wrong for a clinic that is already full.
A better prompt switches on web search and forbids guessing:
Search the web for osteopathy clinics within [your town or area]
that publish their prices. For each clinic, list: clinic name, the
price for a first appointment, session length in minutes, what is
included, the URL, and whether the page shows a date. If a clinic
doesn't publish a price, write NOT PUBLISHED. Do not estimate or
average anything.An illustrative result lists nine clinics: six with prices, two "not published", and one whose page carries a copyright line from two years ago. That table is useful, but only after you open each link, confirm the price, and note that one "first appointment" is 30 minutes while another is 75. Which leads to the most useful thing AI does with competitor prices: putting them on the same basis.
The jobs AI does well in pricing research
Building your cost floor
AI is good at giving your costs a structure and asking about the ones you forgot. A prompt:
I run a [business]. Help me work out my cost per hour of client work.
Ask me for each cost one at a time: rent, insurance, professional
fees, software, consumables, wages, my own drawings, and anything
businesses like mine often forget. Then show the calculation step by
step as a table I can paste into a spreadsheet.Put the final sum in a spreadsheet yourself. Language models make arithmetic slips, especially across many rows, as explained in why AI is bad at maths and what to check. The AI's job is the checklist of costs; the spreadsheet's job is the sum.
Putting competitor prices on the same basis
Once you have verified prices, ask AI to convert them to a common unit: price per minute, price per visit including a standard add-on, or price per course of treatment. This is where hidden differences show up, such as a "cheap" first appointment that is only 30 minutes or excludes a report.
Designing and analysing a price survey
A simple, well-known format asks customers four questions about a service: at what price would it be so expensive you wouldn't consider it, expensive but still worth considering, a bargain, and so cheap you'd doubt the quality. It's usually called the Van Westendorp price sensitivity method. AI can draft the survey, then summarise the answers into ranges. With a small business's sample of 30 to 60 responses, treat the output as a direction, not a precise price.
An illustrative run for a podiatrist's routine nail-and-skin appointment: 42 patients answered the four questions on a card at reception over three weeks. Pasted into a chat assistant with the instruction "Summarise these answers as the price ranges where most people move from bargain to expensive; show your counts and don't round", it returned roughly this: most respondents called anything under $35 a bargain, most began calling it expensive above $55, and only a handful said they wouldn't consider it below $70. The current price was $45.
What needed fixing: the AI had counted six blank cards as zeros, which dragged its "bargain" figure down, and it had described the result as "the optimal price is $48", a precision 42 cards can't support. With the blanks removed, the sensible reading was "there's room to move towards $50 without hitting resistance". That is what the survey can honestly tell you.
Watching competitors over time
A monthly re-run of the same search prompt, saved in a spreadsheet, shows when rivals change prices. Tracking competitors' prices and offers with AI sets this up as a routine.
An osteopath sets a new-patient fee, step by step
An illustration with round numbers. A single-practitioner osteopathy clinic charges $85 for a 60-minute first appointment and $70 for 45-minute follow-ups. The diary has been over 90% full for six months, with a three-week wait for new patients.
Step 1, cost floor. Monthly fixed costs (room rent, insurance, professional registration, software, cleaning, accountant) come to $4,100. The owner wants to draw $5,500 a month. That's $9,600 to cover. Available clinical hours are 120 a month, and at a realistic 85% utilisation that's 102 paid hours. Cost floor including drawings: $9,600 ÷ 102 = about $94 per clinical hour. The current first-appointment fee of $85 for an hour is below it.
Step 2, competitors. Six published prices, checked by hand, converted to price per minute of treatment, ranged from $1.40 to $2.10. For a 60-minute appointment that's $84 to $126. The clinic sits at the very bottom of the range.
Step 3, capacity. A three-week wait is the strongest pricing signal in the whole exercise. The clinic is turning people away at its current price.
Step 4, decision and test. The owner sets new-patient first appointments at $110 and leaves follow-ups for existing patients alone for now. For eight weeks she tracks one number: the share of new-patient enquiries that turn into bookings. Before the change it was about 70%. The illustrative outcome: it settled at around 64%, the wait dropped to about ten days, and monthly revenue from new patients rose because each booking brought in more. The price held.
AI's contribution was real but supporting: the cost checklist, the competitor table and the per-minute conversion took about an hour instead of an afternoon. The decision came from the diary and the conversion rate.
What your diary says about price, before any AI is involved
Capacity is the pricing signal most owners already have and most AI answers ignore. A rough reading guide for appointment-based businesses:
| What you see over the last three months | What it usually means | Sensible move |
|---|---|---|
| New customers wait more than two weeks | Demand exceeds supply at this price | Raise prices for new bookings and test |
| Diary 75-90% full, short waits | Healthy balance | Small annual rise in line with costs |
| Diary under 60% full | A demand problem, not a price problem | Fix visibility and follow-up before cutting price |
| Full at peak times, empty off-peak | Timing, not price level | Off-peak offers or peak surcharges, tested separately |
| Quotes accepted more than 80% of the time | Probably underpriced | Raise quoted rates on the next 20 quotes |
You can ask AI to calculate these figures from a booking export, but they come from your data, and they should be read before you look at a single competitor's price.
Running a price test without a data team
Small businesses can't run statistically perfect experiments, but they can run honest ones. Some approaches that work at small scale:
- New customers only. Change the price for new bookings and leave existing customers on the old price for a set period. It's fairer and easier to explain.
- Quote acceptance rate. For quoted work, track how many quotes are accepted at the new price over the next 20 or 30 quotes. A drop from 60% to 55% might be fine if each job earns more; a drop to 30% is a clear answer.
- One service at a time. Change one price, wait, measure, then decide on the next. Changing everything at once leaves you guessing which change mattered.
- Set the stop rule before you start. For example: "If enquiry-to-booking conversion falls below 55% for four weeks, we go back." Written down, it stops you reacting to one quiet week.
When you're ready to tell customers, planning and announcing a price increase with AI's help covers the wording and timing.
Checking any price figure an AI gives you
- Is there a source you can open? If not, discard the number.
- Is the source current, and does it show a date?
- Is it the same service: same length, same inclusions, same level of practitioner?
- Is it a "from" price or an offer rather than the standard price?
- Is it from a business that serves your kind of customer?
- How many data points is the range built on? Three prices aren't a market.
- Did the AI calculate an average or median? Re-do the sum in a spreadsheet.
One more check deserves its own mention, because it catches the most embarrassing error. When an AI with web search cites a competitor's price, open the page and find the number yourself. An illustrative veterinary practice was told a rival charged $38 for a nurse clinic, with a link. The link worked, but the page had no prices on it at all; the figure seems to have come from a different site, or from nowhere. The practice had already mentioned "local rates around $38" in a staff meeting. Two minutes of clicking would have caught it.
Three businesses, three things the AI missed
A hearing-aid shop asked AI to compare its prices with online sellers. The comparison showed the shop was far more expensive, but the online prices were for the device alone, while the shop's price included hearing tests, fitting, adjustments and aftercare for several years. The fix wasn't a price cut; it was showing the bundle clearly on the price page so customers could compare like with like.
A pharmacy used an AI-suggested price for a private vaccination service, based on other pharmacies' published prices. Three weeks later its supplier's cost for the vaccine rose, and the price no longer covered it. Competitor prices tell you what the market charges; they say nothing about your margin. Cost comes first.
A dental practice was told the "average price of whitening" by an AI summarising ads. Most of those ads excluded the assessment appointment that the practice included. Its price looked high only because it was honest about the full cost.
Where AI pricing research crosses a line
Most AI pricing research is ordinary and lawful, but a few uses carry real risk:
- Coordinating with competitors. Researching published prices is fine. Sharing your planned prices with rivals, or using a shared tool that feeds competitors' non-public pricing into recommendations, can break competition law, even when software does the sharing. If a pricing tool pools data from businesses near you, ask exactly what data it uses and take advice before signing.
- Personalised prices from personal data. Charging different people different prices based on what an AI infers about them from their data is a legal and reputational risk, and customers tend to find out. Keep prices tied to the service, time and published offers.
- Regulated prices. Some health and professional services have fees set or constrained by contracts or professional rules. AI research doesn't change those obligations.
AI can also carry assumptions into pricing that you wouldn't choose deliberately, such as suggesting higher prices for "affluent" areas or lower ones for older customers; AI bias in small business decisions explains how to spot it. For quote-based trades, where the question is less "what's my rate" and more "is this job priced right", whether AI can price a job accurately goes through the checks tradespeople need.
Further reads
- Can AI Help a Small Shop Set Prices and Promotions? — The retail version: prices, promotions and margins with AI.
- How Landscapers Use AI to Stop Underpricing Jobs — A trade example of building prices up from real job costs.
- How to Do Competitor Research With AI in One Afternoon — Wider competitor research beyond price lists.
- How to Check Sources and Citations in AI Research — How to verify the sources behind any AI research answer.
- Are AI-Generated Customer Personas Accurate? How to Check — Check who your customers really are before pricing for them.
- Is AI Dynamic Pricing Worth It for a Small Hotel? — When automated price changes make sense, and when they don't.
- How to Run a SWOT Analysis of Your Business With AI — Six stages and copyable prompts for a SWOT built on your own figures and reviews, with a podiatry clinic example and a TOWS step that turns it into actions.
- Can AI Help a Small Business Grow Revenue, or Only Cut Costs? — When AI grows revenue and when it only trims costs: diagnose your bottleneck, pick from five revenue levers, and check two worked examples with numbers.
- AI Menu Engineering: Which Dishes to Promote, Reprice or Drop — Run the classic menu engineering matrix with AI on your own till data, with a worked ten-dish example and a price-test calculation you can copy.
- AI Prompts for Bakers: Pricing, Captions and Customer Replies — Twelve copy-ready prompts for bakers covering costing and prices, social captions, and customer replies, each with what to check before you use the answer.
- ChatGPT Prompts for Pet Sitters and Dog Walkers That Win Clients — Twelve ready-to-paste prompts for pet sitters and dog walkers, from profile bio to daily updates, each with what to feed in and what to check.
- Can AI Give Customers a Ballpark Price for Decorating Work? — How a decorator can let AI give honest ballpark prices from a room-by-room grid, and which jobs it should refuse to price.
- How to Create Quotes and Estimates in Minutes With AI — A quoting set-up where AI reads the messy enquiry, your rate card and spreadsheet set the price, and an approval step catches the costly errors.
- How to Test a New Service Idea With AI Before You Launch It — A six-week way to test a new service with AI doing the drafting and real customers supplying the evidence, with pass marks set before the results arrive.
- Break-Even Analysis With AI: A Worked Example for a New Product — A specialty coffee roaster costs a canned cold brew launch step by step: missed costs, channel mix, scenarios and the shelf-life check that changed it.
- Perplexity vs ChatGPT for Business Research — Use a supplier decision, an evidence ledger and five research checks to choose between Perplexity and ChatGPT.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: general pricing-research methods (cost-plus, the Van Westendorp price sensitivity questions); vendor documentation for ChatGPT, Claude and Gemini web search features.