पाठशाला Pathshala · ग्राहक Grāhak, The customer · Lesson 21 · Build
Willingness to pay: the research that sets your price
Customers are poor judges of what they would pay and good judges of what they will not. Use a price sensitivity survey to find the band, then a real offer to find the price inside it.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
Most Indian startups set their first price by looking at a competitor and subtracting something. That price then survives for years, because changing it feels risky and nobody owns the question.
This lesson sets out a two-stage method that replaces the guess: a short survey that finds the band of prices a segment will accept, and a set of real offers that finds the price inside it. A figure draws the survey’s four curves so you can see how the band moves when the answers do.
What willingness to pay is, and is not
Price matters more to profit than any other lever. McKinsey’s The Power of Pricing, working from the average income statement of an S&P 1500 company, found that a 1 per cent price rise with stable volume would raise operating profit by 8 per cent. For a startup with thin margins the multiplier is larger, because the price rise falls straight to contribution.
Yet customers are unreliable witnesses to their own willingness to pay. In Mind Your Pricing Cues, Eric Anderson and Duncan Simester report that for most items consumers have no accurate sense of what the price should be, and describe a catalogue test in which raising a dress from $34 to $39 increased demand by a third. Ask someone “would you pay ₹999?” and you measure politeness. Willingness to pay is better understood as a range of prices a segment will accept for a product they understand, compared with the alternatives they know. Research can find that range. Only a real offer finds the price.
Run the price sensitivity meter
The method most teams start with was proposed by the economist Peter van Westendorp in the 1970s. It asks four questions, as Conjointly’s guide sets them out. At what price would it be so cheap that you would doubt its quality? At what price would it be a bargain, a great buy for the money? At what price would it be starting to get expensive, so that you would have to think before buying? At what price would it be so expensive that you would not consider it?
Four rules make the answers worth having. Show the product first: a demo, a two-minute video or a working trial, so respondents price what you sell, not what they imagine. Ask one segment at a time: a Surat textile trader and a Pune clinic answer different questions even if the software is the same; 100 to 200 completed answers per segment is enough to draw stable curves. Ask in rupees per month or per year, whichever unit the customer pays in, in their language. Let them answer alone: the Pew Research Center notes that the tendency to agree with the questioner is more pronounced when an interviewer is present, so use a form or a WhatsApp flow rather than a salesperson reading the questions aloud.
Read the curves
Plot, for every price, the share of respondents who called it too cheap, a bargain, getting expensive or too expensive. The “too cheap” and “bargain” curves fall as price rises; the other two rise. The lower edge of the acceptable range is where “too cheap” crosses “getting expensive”; the upper edge is where “a bargain” crosses “too expensive”. The optimal price point is where the two outer curves cross, the price at which as many people find it too cheap as find it too expensive. Move the sliders below to see how the band responds.
Run the four questions separately for each segment and each plan, and resist pooling them. A pooled survey of shop owners and chains produces a band that fits neither: too high for the shop, too low for the chain. If two segments’ bands barely overlap, that is the finding. It says you need two plans, or two prices for the same plan with a fence between them, rather than one compromise price that loses both.
Two readings matter more than the optimal point. The width of the band tells you how much room you have: a narrow band means the segment has a settled view of the price and you will compete on it; a wide one means you can position. And the share who reject a given price, too cheap or too expensive, tells you what a price costs in lost buyers before any competitor is involved. A price below the band is not safe either: in many categories a price that looks too low reads as poor quality, and the buyers it loses never say so.
Where the survey misleads
The method has known weaknesses, and Conjointly lists them plainly. Respondents tend to underestimate the prices they name. The product is priced in isolation from its features and its competitors. It suits poorly a product so new that buyers have no settled sense of a fair price. And the optimal point does not necessarily maximise revenue or profit. Treat the output as a hypothesis about where to test.
Two Indian distortions add to these. Respondents who expect to bargain later name low prices as an opening position. And respondents who think the founder is listening name the price they think the founder wants. A self-administered form, an assurance that nobody will be sold to, and a question about what they pay today for the alternative reduce both.
Real-offer tests: make them pay
Choose two or three prices inside the band and offer them, for real, to comparable groups of new prospects. For a self-serve product, show each price to a different cohort of sign-ups for two to four weeks and compare paid conversion. For a sales-led product, quote different prices to alternate qualified leads and record the outcome. For a product not yet built, take a refundable pre-order deposit at each price. The [lesson on pricing experiments](/library/pricing-experiments-without-burning-customers) covers how to do this without treating existing customers unfairly: new prospects only, every price honoured, and no customer charged more than the price shown to them.
Decide on revenue per hundred prospects, not on conversion alone. A lower price that converts more can still earn less, and a higher one that converts slightly fewer can fund the support that keeps customers. Look at the second month too: a price that converts well and churns fast was a discount in disguise.
In sales-led Indian deals the quoted price is an opening position, and the buyer will expect to negotiate. Record the final price as well as the list price for every deal, and test list prices with the discount authority held constant. If one price closes at the same rate as another but with deeper discounts, the higher list price was not tested; the salesperson’s nerve was. A written rule on how much a salesperson may give away, and who approves more, keeps the test clean and protects the margin long after it ends.
A survey tells you the band the market will tolerate. Only a customer paying tells you the price.
A worked example: an invoicing app in Surat
A company sells GST invoicing and payment reminders to small traders. It surveys 160 traders in Surat after a short demo in Gujarati. The curves put the acceptable range at roughly ₹385 to ₹670 a month and the optimal point near ₹420; these are the figure’s starting values. The founder had planned ₹299 because a competitor charges ₹349. The survey says ₹299 sits below the range, at a price about a quarter of traders would think too cheap to trust.
For four weeks the company shows ₹399, ₹499 and ₹649 to three cohorts of new sign-ups from the same campaign. Paid conversion is 9, 8 and 5.5 per cent. Revenue per hundred sign-ups is about ₹3,600, ₹4,000 and ₹3,600 a month. It chooses ₹499, offers the year at ₹4,999, and books a test of ₹579 for the next quarter. Without the research it would have launched at ₹299, below the band its own customers drew, and never learned what the extra ₹200 a month was worth.
The quarterly price review
Each quarter, for each segment: re-run the four questions with 100 new respondents and redraw the band. Compare your price with the band; if it sits at an edge, plan a test. Run one real-offer test with new prospects, with a stated end date and a decision rule written before it starts. Check revenue per hundred prospects and second-month retention for every price tested. Write down the decision and the reason where the next person to own pricing will find it.
Price is a decision that compounds every month. Give it an owner and a date in the calendar, as you would any other number that decides whether the company survives.
The worked example is illustrative. Any price change for existing customers should follow the contract terms and the notice your customers were promised.
Sources
- Michael V. Marn, Eric V. Roegner and Craig C. Zawada, The Power of Pricing, McKinsey Quarterly, February 2003 — A 1% price rise with stable volume raises operating profit by 8% for the average S&P 1500 company.
- Eric T. Anderson and Duncan Simester, Mind Your Pricing Cues, Harvard Business Review, September 2003 — Consumers lack an accurate sense of most prices; a dress moved from $34 to $39 sold a third more.
- Conjointly, Van Westendorp Price Sensitivity Meter: the four questions, reading the range, and limitations
- Pew Research Center, Writing Survey Questions (acquiescence bias and the interviewer effect)