The worst pricing decision a founder building a category of one can make is to wait. Wait for a competitor to publish a price, so there is finally something to react to. There is no competitor. There may never be one, or one may show up eighteen months from now running a completely different business model. Founders who land a durable price early do it by asking the customer directly what a fair number looks like, then checking that number against the value of the next-best alternative, not against a market that does not exist yet.
Competitor triangulation was never the real method
Most SaaS pricing advice assumes a market with five competitors already publishing tiers, so the job is just picking a lane above or below them. That shortcut collapses the moment the product is genuinely new. OpenAI had this problem in a pure form: ChatGPT Plus was a category-creating product with no direct precedent to price against, and the company still had to pick a number. It launched the subscription at $20 per month on February 1, 2023. No competitor set that number. A judgment about willingness to pay did.
Ask four questions before you guess one number
The tool for that judgment is nearly fifty years old. Dutch economist Peter van Westendorp introduced what is now called the Price Sensitivity Meter at the ESOMAR Congress in Venice in 1976. The method asks four questions of a target buyer: at what price does this become too expensive, at what price does it start feeling expensive, at what price is it a good deal, and at what price is it so cheap you would not trust it. Patrick Campbell's pricing consultancy, Price Intelligently, still runs versions of these same four open-ended questions to determine customer willingness to pay when there is nothing else to benchmark against.
Superhuman is the clean case study. Before it had a pricing page, the company ran Van Westendorp analysis across three buyer personas, and the answers did not converge. Founders landed near $30, VCs indicated $45, and operators leaned closer to $20. Rather than build three tiers to chase three answers, Superhuman split the difference and shipped one flat price with no pricing page at all. The number that came out of the model was $29. An intern, Cam Wiese, made the call to round it up to $30, and that number stuck.
Then anchor the number to the alternative, not the competitor
A survey gives a range. It does not tell you where inside that range to land, and it says nothing about whether the range itself is defensible. For that, use the older sibling of Van Westendorp: Economic Value to the Customer. John Forbis and Nitin Mehta published the framework in Business Horizons in 1981, and its logic still holds up. Price against the customer's next-best alternative (often a manual process, a spreadsheet, or a headcount hire) plus the monetized value of whatever your product does that the alternative cannot. That framework is precisely what you need when there is no competitor list price to hide behind, because it forces the number to come from the customer's cost structure instead of the market's.
A Forbes Business Council piece from earlier this year makes the same case directly: value-based pricing is the right model specifically when there is scarcity in the market, meaning the product is one-of-a-kind or only a small handful of companies do anything close to it. Scarcity is not a reason to price cautiously. It is the reason you get to price on value instead of on a competitor's spreadsheet.
The range comes from the customer. The anchor comes from the alternative they were using before you existed.
What to do this week
- Run the four Van Westendorp questions on ten to fifteen target buyers across the personas that actually pay, not just the ones that use the product.
- Write down what those buyers do today without your product, and put a dollar figure on that alternative.
- Pick one number inside the range the survey gives you, weighted toward the alternative's cost, and ship it without a public pricing page if the range is wide.
- Revisit the number at the next renewal cohort, using actual conversion and churn data, not a second round of guessing.
A category of one does not get a shortcut. It gets two forty-year-old methods, run back to back, and the discipline to pick a number and ship it instead of waiting for a market that has not shown up yet.