Search for how to make money with AI and you get income claims: monthly figures, hourly figures, ranges quoted in dollars by people who have not quoted a rate in three years. Almost none of it tells you how the number was built. Three models do, and you can run all three in an afternoon: cost-plus gives you a floor, market rate gives you a reference point, and value-based gives you the number you can charge without flinching.
A service pricing ai prompt cannot choose between them for you, because only you know your costs and only the client knows their problem. What it can do is hold the arithmetic in front of you while you decide, and catch the place where your enthusiasm has quietly replaced a number.
Why income claims crowd out pricing method
An income figure without its arithmetic is not information. It tells you that someone, somewhere, at some point, earned a number you cannot check, doing work you do not fully understand, under conditions you would not recognise. That is why these figures are so widely repeated and so rarely useful.
The same applies to rate claims. A blog stating that resume writing pays somewhere between 150 and 500 USD per document may well be accurate. But without knowing whether that is per draft or per revision round, whether it includes the hours spent sourcing a client, and whether the writer charged that in year one or year ten, it cannot price your work.
What you can use is method. Three models, applied to the same job, tell you more than any single quoted rate.
Model one: cost-plus, and the floor it produces
Cost-plus prices the hours and adds a margin. The useful output is not the quote. It is the floor: the lowest number at which the work is still worth doing.
Start with the honest numbers.
- Billable hours on this job: estimate at 20 percent more than your optimistic estimate.
- Unbilled hours: follow-up, revisions, invoicing, and the calls nobody logs.
- Direct costs: tools, stock, a paid brief, or a reference you must buy.
- Overhead share: a share of software and business costs you carry regardless.
- Your minimum hourly rate: the number at which an hour of your time is worth more than doing nothing.
Multiply, add, and you have the floor. Anything below it is a decision you must make deliberately, with the knowledge that you are subsidising the client with your own unpaid labour. That is sometimes worth it for a first client and a case study. It is never worth it by accident.
Model two: market rate
Market rate asks what comparable work sells for, then positions you within that range. This is where AI is genuinely useful, because the model can help you define the comparison set and hold the range you found.
The comparison set is the hard part. Compare like with like on three dimensions: the buyer's sophistication, the stakes of being wrong, and whether you are selling a defined deliverable or an ongoing partnership. A cheap rate for a template and a cheap rate for a bespoke system are the same number doing completely different work.
Most underpricing comes from comparing yourself against the low end of a range that does not contain your service. Collect at least three comparable offers, from real listings, then note what is included in each. A lower headline price that excludes revisions, rush delivery, and source files is not a cheaper service.
Model three: value-based
Value-based prices the outcome rather than the effort, which is the only model that lets a good job become a more expensive job. The test is simple and uncomfortable: what is this problem currently costing the client, and how much of that would this work remove?
You will rarely capture the whole value. Capture a fraction, and say so. Nobody expects a rebrand to be worth what a rebrand does for a business that has stalled, but they may accept a fraction of it without blinking.
The risk with value-based pricing is arithmetic dressed as confidence. Resist it. Anchor the value claim in something the client told you: a revenue figure, a lost contract, a cost per month they mentioned. If they never mentioned it, you have a hypothesis, not a value, and you should price with model one instead.
Comparing the three models on the same job
Same job, three defensible answers.
| Model | The number comes from | Strength | Where it fails |
|---|---|---|---|
| Cost-plus | Your hours, costs, and minimum rate | Never below your floor | Ignores what the job is worth |
| Market rate | Comparable offers in your market | Sanity check on the other two | Wrong comparison set misleads |
| Value-based | The cost of the problem to the client | Highest ceiling | Needs a value the client stated |
Run all three and you usually find the truth: the floor is lower than the market rate, and the market rate is lower than the value. Most sane prices sit between the market rate and the value, and never below the floor.
The prompt that builds all three
Help me price [service described in one line] for [client type]. My inputs: [hours I estimate the work takes], [unbilled hours as a percentage of billed], [direct costs for this job], [my monthly overhead and how many billable hours a month gives me], [my minimum hourly rate]. What the client told me about the problem: [paste their words about cost, time, or consequence, exactly as they said it]. Comparable offers I found: [three listings or quotes, with what each includes]. Produce three numbers with the arithmetic shown for each: (1) my cost-plus floor, (2) a market-rate estimate using my comparables, with the adjustments for scope and quality spelled out, (3) a value-based range tied only to what the client actually said. Then give me a recommended number and one sentence on the assumption it depends on. Do not invent a market rate for me if my comparables are too thin. Do not assume a conversion rate between currencies unless I give you one. If any input is missing, tell me which one before calculating.
Fill in these details
- Hours: the honest, slightly pessimistic number. Underestimation here silently becomes your price.
- Overhead: include it even if it feels like overhead. Software you pay for monthly is a cost of being able to deliver.
- Client's own words: paste them. Value-based pricing built on your assumptions is fiction with a decimal point.
- Comparables: three, with inclusions listed. Two is a coincidence and four is a research project.
Convert a rate without guessing
Foreign currency is where arithmetic goes to hide. If a market reference is quoted in dollars and your costs are in rupees, write down the rate you are using, the date you took it from, and a buffer you chose deliberately. Then decide which number the client sees.
The common mistake is quoting a dollar rate and hoping. It converts to less than you assumed, the margin disappears, and you discover it a month later. Decide your own base currency, convert once, and price the converted number. If the client pays in dollars, hold the exchange risk as a separate decision rather than letting it leak silently into your rate.
Frequently asked questions
Should I use a pricing ai prompt for my first three clients?
Use it to do arithmetic and to check your floor, not to choose the number. Early clients are priced with judgement about fit and opportunity, which no model can assess. What the prompt is good for is making sure the judgement rests on hours and costs rather than on nerves.
What if the market rate is below my floor?
Then you are not in that market, or that offer is not the same service. Find out which. Sometimes the gap is a positioning problem you can fix with a narrower offer. Sometimes the honest answer is that you charge more, or you do not take that work.
Is value-based pricing realistic for small jobs?
For small, well-defined jobs, value is hard to observe and easy to overclaim. Cost-plus or market rate is the safer model there. Value-based works when the consequence of the problem is visible and the client has already described it in money terms.
Write down your floor price today
Take the last job you finished, count the hours including the ones you did not log, add your costs, and divide by the fee you charged. That number is your real rate, and most people have never calculated it. Once you know it, run the prompt above for your next quote and choose deliberately. If the pricing side of running a business is new to you, our personal finance topic collection covers where the money goes afterwards.