How to Test Startup Pricing Without Guessing

On this page

1. Start with value and alternatives, not costs
2. Choosing a pricing model
3. Learning about price in customer interviews
4. Pricing experiments you can run cheaply
5. The limits of stated willingness to pay
6. Where simulated testing fits, and where it stops
7. Your next step

It is common to pick a price by looking at a few similar offers, adjusting for gut feel and hoping for the best. Then, when sales are slow, they cannot tell whether the price was the problem or something else entirely.

You do not need a complex pricing model to do better. You need to treat your first price as a hypothesis, choose a pricing structure that fits how customers get value, and gather evidence from what people actually do rather than what they say. This guide covers the main pricing models, how to learn from interviews without being misled, small experiments you can run and the honest limits of any method that relies on stated or simulated preferences.

Start with value and alternatives, not costs

Your costs set the floor: below that price, every sale loses money. They do not set the price customers will accept. That depends on two things:

• The value of the outcome to the customer, in time saved, money earned, risk avoided or frustration removed.
• The cost of the alternative they would use instead, including doing it themselves or doing nothing.

Say you sell a done-for-you social media calendar for local bakeries. If the alternative is the owner spending a few evenings a month writing posts, your price competes with the value of those evenings, not with your hourly cost to produce it. Write down both numbers before you think about price points.

Choosing a pricing model

The structure of your price often matters as much as the number. Each model fits a different kind of value.

One-time purchases

A single payment for a product or package. This fits physical products, templates, courses and fixed-scope services where the value is delivered once.

• Works well when: the benefit is immediate and complete, and customers dislike ongoing commitments.
• Watch out for: needing a steady flow of new customers, since each one pays only once. Consider whether add-ons, refills or upgrades make sense.

Subscriptions

A recurring payment, usually monthly or yearly. This fits software, ongoing services, replenishable products and anything where value continues over time.

• Works well when: customers use it regularly and would notice if it stopped.
• Watch out for: subscription fatigue. Customers ask “will I use this every month?” A subscription for something used occasionally invites quick cancellations.
• Test separately: monthly versus annual billing, and what a discount for paying yearly does to commitment.

Freemium

A free tier with paid upgrades. This fits products where free users cost little to serve and where some of them naturally grow into needing more.

• Works well when: the free version is genuinely useful and the paid version solves a clear next problem, such as more volume, collaboration or advanced features.
• Watch out for: a free tier so generous that nobody upgrades, or so limited that nobody sticks around long enough to care. Decide in advance what behavior should trigger an upgrade, and check whether it happens.

A useful exercise is to describe, in one sentence each, what a customer gets under each model. If one sentence sounds obviously more natural for how your customer uses the product, start there.

Learning about price in customer interviews

Interviews are valuable for pricing, but only if you ask the right questions. “How much would you pay for this?” produces numbers that are easy to say and rarely reliable. People guess, try to be polite or try to negotiate.

Ask about what they already do and spend instead:

• “What do you use today to deal with this, and what does it cost you?”
• “When did you last pay for something to solve this? How did you decide it was worth it?”
• “Who would need to approve a purchase like this?”
• “What would make this feel too expensive to even consider?”

Then, late in the conversation, you can show a concrete offer with a real price and watch the reaction. Hesitation, follow-up questions about what is included, or comparisons to something they already buy tell you more than a number they invent on the spot. For help phrasing these questions neutrally, see the guide on customer discovery questions.

Pricing experiments you can run cheaply

Real pricing evidence comes from situations where people can actually say yes or no with money or a clear commitment. A few practical options:

• Price on the landing page. Show the actual price with a “buy,” “preorder” or “book” button. Compare how many visitors click through at different price points, using separate pages or periods, and keep everything else the same.
• Preorders and deposits. Ask early customers to pay a deposit at your intended price. Even a small number of strangers paying is meaningful.
• Tiered offers. Present two or three packages and see which one people choose. The option people pick shows you which level of value they consider worth paying for.
• Direct quotes for services. Quote different prospects slightly different prices for the same scope and track which ones accept.
• Raise the price for new customers. If almost everyone says yes immediately, that is a sign you may be underpricing. Raise it for the next group and watch what changes.

Keep the test fair: change only the price, not the headline or offer, and decide in advance what result would make you move up, move down or stay put. Small numbers mean noisy results, so look for clear differences rather than reading meaning into tiny gaps.

The limits of stated willingness to pay

Anything based on what people say they would pay has a built-in gap from what they will actually pay. Saying “I’d pay that” costs nothing. Typing in card details costs something. That gap can run in either direction: some people overstate to be supportive, others understate to seem careful.

This does not make stated preferences useless. They are good for spotting price ranges that feel absurd, understanding how people compare your offer to alternatives and learning which features make a higher price feel justified. They are not good for setting a final number. Use them to narrow your options, then let real purchases decide.

Where simulated testing fits, and where it stops

AI-simulated testing can be a fast way to explore how different groups might react to different price points before you run real experiments. On SellTo.us, you can enter several versions of the same idea in one test, each with its own price, and see how simulated participants respond to each. The larger test sizes allow up to three or up to five versions. All versions go to the same simulated participants, so the comparison inside one test is like-for-like.

It is important to be clear about what this does not do:

• Simulated preferences do not establish an optimal real-world price. The participants are AI personas who do not spend real money.
• A report’s “best price” is the AI’s judgment based on the simulated answers. It is not calculated from a demand or price curve, and it is not a scientifically verified optimum.
• Small differences should be treated as a tie. Reruns can give somewhat different numbers, and the order in which versions appear can influence the AI.

Used well, a simulated test helps you decide which two or three prices are worth putting in front of real customers, and what objections to address when you do. You can read more about the method on how it works.

Your next step

Pick the pricing model that matches how your customers get value, and write down two or three candidate prices with the reasoning for each. In your next few customer conversations, ask what they spend on the problem today. Then put one real price in front of real buyers with a way to pay or commit, and let their actions, not their opinions, tell you whether to move it. If your idea is a physical product, the guide on testing a product idea before buying inventory shows how to combine price tests with preorders.

Written by the SellTo.us Team. Examples in this article are hypothetical. SellTo.us results come from AI-simulated participants, not real people, and don’t predict sales or guarantee success.