How to Price Your Product When You Have No Benchmark

The two anchors that mislead

Without comparable competitors, most people reach for one of two anchors, and both produce a price that is too low.

Your costs. Cost-plus works for manufacturing and misleads badly for software, where the marginal cost of another customer is close to nothing. Costs tell you the floor below which you lose money. They say nothing about what the outcome is worth.

What feels reasonable. Founder intuition about price is anchored to what the founder would personally pay, which is a poor proxy for what a business will pay to solve a problem that costs it real money.

Finding a first number

1. Ask ten customers what the problem costs them today

   Hours per week, a person's part-time salary, a tool they already pay for, deals lost. Ask during discovery rather than as a pricing survey.

   The questions in our guide to [running a client discovery meeting](/blog/client-discovery-meeting) surface this naturally, and the answers are far more reliable than anything a pricing survey produces.

   Ask during discovery

2. Establish the range, not the average

   You will get answers spread across an order of magnitude. That spread is the useful finding: it tells you there are different segments, not that your data is bad.

   Price for the segment you want, not for the midpoint of everyone you spoke to.

   The spread is the signal

3. Take a visible fraction of the value

   If the problem costs a customer the equivalent of two days a month, a price worth a few hours of that is obviously good value and still a real number.

   The point is a ratio the buyer can compute in their head. If they have to be convinced with a spreadsheet, the ratio is too tight.

   Make the maths obvious

4. Round up, and never price at a suspiciously low number

   Very low B2B prices signal a hobby product rather than a bargain, and they attract customers who churn quickly and ask the most of support.

   Buyers use price as a quality signal when they have nothing else to judge by, which is exactly the situation a new product is in.

   Cheap reads as risky

5. Set list price where you can discount from

   You will want to discount for early customers, references, and longer terms. All of that requires headroom.

   Starting low removes every lever you have, and raising later is far harder than the discount you would have given.

   Leave yourself room

Choosing what to charge for

The unit matters as much as the number. It should grow with the value the customer gets, be easy to predict, and be hard to game.

Per-user pricing has a specific trap worth naming. It charges the customer more for the thing you most want them to do, which is spread the product to a second and third person. Since widening adoption is the strongest predictor of retention, as covered in spotting a customer about to churn, a per-seat model can work directly against your own renewal rate in small accounts.

Whatever you choose, a buyer should be able to work out their bill without a calculator. Pricing that requires explanation slows every deal, and it makes your proposal longer than it should be.

Testing without a big sample

You will not get statistical significance at ten customers. You can still learn a lot.

Watch reactions, not just outcomes. Someone who accepts immediately and someone who pauses, asks a question and then accepts are two different data points at the same price.

If nobody negotiates, you are too cheap. Frictionless selling feels like success and is usually the clearest signal you have priced below what the market bears.

Track why you lose, from a fixed list. If price is genuinely the top loss reason across a dozen deals, that is a signal. If it is the reason everyone gives but the real cause varies, that is a different problem, and the interviews in win/loss analysis are how you tell the two apart.

Test on new customers only. Changing prices for existing customers while experimenting is how you damage trust in a way that is slow to repair.

Raising it later

Raising prices is normal and survivable. Doing it badly is not.

Grandfather early customers, at least for a period. The people who took a risk on you should feel like they got something for it, and the cost of that goodwill is small.

Give long notice, with a reason and a number. The mechanics are the same as any renewal change and are covered in asking for a renewal before it's due. Notice matters more than the size of the increase.

Raise the list price first, then existing customers later. New customers arriving at the new price validate it before you touch anyone's existing bill.

Expect to be too cheap for longer than is comfortable. Almost every founder who eventually raises prices reports they should have done it sooner, and very few report the opposite. That asymmetry is worth remembering when you are choosing the first number.