Willingness to Pay Research: How to Price Before You Build

Three methods measure willingness to pay before you build anything: the Van Westendorp Price Sensitivity Meter, which maps a defensible price range from four questions; Gabor-Granger, which tests specific price points against a group of respondents to find a demand curve; and direct interview asks, which check whether a stated price survives contact with a real buying decision instead of a hypothetical one. None of the three gives you a single "correct" number — together they narrow a guess into a range you can defend.

Quick Answer: Run interviews first to sanity-check that a price is even in the right neighborhood, use the Van Westendorp Price Sensitivity Meter to bound an acceptable range from a broader sample, then use Gabor-Granger to sharpen that range into specific price points and a rough demand curve. Skip straight to "would you pay $X?" and you'll get an answer — just not one you can trust, because a hypothetical price question measures politeness, not willingness.

Prerequisites: A Concrete Offer and Access to Real Buyers

Willingness to pay research only produces a usable number if two things are already in place: an offer specific enough that a stranger can price it in their head, and a way to reach people who actually have the problem you're solving. Skip either one and the research still runs — it just measures something other than willingness to pay.

A concrete offer, not a category description. "A tool for managing freelance invoices" is a category, not an offer, and nobody can price a category — every respondent will silently substitute their own assumptions about what's included, and your price data will average across offers that don't actually match. Write the one-sentence version first: what the buyer gets, how often, and what problem it removes. If two people in your target list would describe the offer differently after you explain it, the offer isn't tight enough to price yet.

Access to people who have the problem, not people who are willing to talk to you. Friends, family, and your own social network will give you polite, encouraging numbers that have nothing to do with what a stranger would actually pay — the same bias Rob Fitzpatrick spends most of The Mom Test warning against, just applied to a dollar figure instead of a feature opinion. Recruit from the same pool you'd use for market research as a technical founder: people who currently experience the problem, ideally people who currently pay for some alternative, even a spreadsheet or a competitor's product.

A rough sense of your cost floor. You don't need a full unit-economics model before you start, but you should know roughly what it costs you to deliver the offer, so a price range that comes back from research below your floor is a signal to change the offer, not a number you quietly ignore because it's inconvenient.

Method 1: The Van Westendorp Price Sensitivity Meter

Van Westendorp asks four fixed price questions and uses where the answers cross to define an acceptable price range — it's the fastest way to bound a range from people who've never seen your product before.

Ask each respondent, in this order, about the same concrete offer:

1. "At what price would this be so expensive you would not consider buying it?" (too expensive)
2. "At what price would this be priced so low that you'd question the quality?" (too cheap)
3. "At what price does this start to feel expensive, but you'd still consider it?" (getting expensive)
4. "At what price does this feel like a bargain — a great deal for what you get?" (getting a bargain)

Plot the four answers as cumulative curves across your respondent group. Where "too expensive" crosses "getting a bargain" marks the upper bound of an acceptable range; where "too cheap" crosses "getting expensive" marks the lower bound. The range between those two crossing points — not a single number — is the output, and that range is the honest answer to "what should I charge": a defensible band, not a point estimate.

Van Westendorp works because none of the four questions asks "would you pay this?" directly — each one asks about a threshold, which is easier to answer honestly and harder to game toward a socially desirable number. Its weakness is the same shape as its strength: because respondents never see a single number and commit to it, the method tells you about a range of reasonable prices, not about actual purchase intent at any specific point inside that range.

Method 2: Gabor-Granger

Gabor-Granger shows each respondent a specific price for the concrete offer and asks a direct yes-or-no purchase question, then moves the price up or down depending on the answer — the output is a demand curve, not a range.

The script is simple and repeats:

• "Would you buy this at $[price]?" If yes, ask again at a higher price. If no, ask again at a lower price.
• Continue for a small number of steps per respondent (three or four is typical) until you've bracketed the price where their answer flips from yes to no.
• Randomize the starting price across respondents so early anchoring doesn't bias the whole sample toward whatever price you happened to lead with.

Aggregate the flip points across your full respondent group and you get a demand curve: the percentage of respondents who'd say yes at each price point you tested. Multiply that percentage by the price at each point and you can estimate a revenue-maximizing price — the point on the curve where price times acceptance produces the highest number, not necessarily the highest acceptance rate on its own.

Gabor-Granger's strength is that it produces something closer to an actual purchase decision than Van Westendorp's abstract thresholds do — a respondent has to commit to yes or no at a real number. Its weakness is the same one every stated-preference method shares: a hypothetical yes still costs the respondent nothing, so the yes rate at any given price will usually run higher than what a real checkout page would show.

Method 3: Direct Interview Asks

Direct interview asks skip the structured questionnaire and instead ask about past spending behavior and current budget reality — the goal is to catch a price that sounds right in the abstract but falls apart against what someone actually spends today.

Useful questions, asked conversationally rather than as a script read verbatim:

• "What do you currently use to solve this, and what does it cost you — in money or in time?"
• "What's the most you've paid for a tool like this in the past year?"
• "If this existed today at $[price], what would have to be true for you to move your budget toward it?"

Notice the shape: every question anchors to something that already happened, not to a hypothetical future decision — the same discipline The Mom Test applies to product feedback generally. "Would you pay $49/month for this?" invites a polite, low-cost yes. "What did you pay for the last tool you switched to?" forces the respondent to reference a real number they can't easily fudge, and it often surfaces a budget ceiling Van Westendorp and Gabor-Granger both miss because neither method asks about existing spend at all.

The strongest version of a direct ask isn't a question at all — it's an actual pre-sale. If your interview subject says a price sounds reasonable, the next test is asking them to pay it. Pre-selling your product turns "sounds reasonable" into a checked or failed prediction, which is a category of evidence none of the three methods above can produce on their own.

Method Comparison: Sample Size, Effort, and Precision

The three methods trade sample size, effort, and output precision against each other, which is why the strongest studies run more than one rather than picking a single "best" approach.

Read the table by what stage of validation you're actually in. Early on, when you're still not sure the problem is real, direct interview asks are the right investment because they're cheap to run in small numbers and catch the worst pricing mistakes — a price that's absurd relative to what people currently spend. Once the problem is validated and you need a defensible number for a pricing page, Van Westendorp and Gabor-Granger are worth the larger sample because their output is a range or a curve you can actually put in a spreadsheet.

Worked Example: Pricing a Sample SaaS Tool

Priya was building a scheduling tool for independent massage therapists — automated rebooking reminders, no-show deposits, and a public booking page — and had validated the problem through a dozen interviews before she ever mentioned a price.

Step 1 — direct asks first. In the same interviews where she validated the problem, Priya asked what therapists currently paid for scheduling software, if anything. Most named a specific existing tool and a specific monthly cost, which gave her a real anchor instead of a guess: whatever she charged needed to make sense next to what people were already paying, not against an abstract idea of "reasonable SaaS pricing."

Step 2 — Van Westendorp for the range. Priya sent a short four-question Van Westendorp survey to a wider list of therapists sourced from a local trade group, well beyond the dozen she'd already interviewed. The crossing points on the resulting curves gave her a band — a monthly price low enough that "too cheap" didn't start signaling low quality, high enough that "too expensive" didn't cut off most of the list. The band was wide enough that she still didn't have a launch price, only a fence around where the launch price needed to sit.

Step 3 — Gabor-Granger to sharpen it. Using price points spread across the Van Westendorp band, Priya ran a short Gabor-Granger sequence with a subset of the same list, asking a direct yes/no at each point. The resulting curve showed acceptance dropping off sharply above one particular point in the band — not at the very top of the Van Westendorp range, but noticeably below it — which told her the revenue-optimal price was closer to the middle of the band than the top.

Step 4 — the real test. Priya didn't treat the Gabor-Granger number as final. She built a one-page pre-sale offer at that price and asked the therapists who'd shown the strongest interest across all three rounds of research to actually pay for early access. The conversion from "said yes to a hypothetical price" to "paid a real deposit" was her actual answer — lower than the survey data alone would have predicted, which is normal, and exactly why a pre-sale sits above every survey method in the hierarchy of evidence.

Mistakes: Hypothetical Bias and Leading Price Anchors

Trusting a hypothetical "yes" as if it were a purchase. Every stated-preference method — Van Westendorp, Gabor-Granger, and a plain "would you pay this" question alike — suffers from hypothetical bias: saying yes to a hypothetical price costs the respondent nothing, so stated willingness to pay reliably runs ahead of what people do at an actual checkout. Treat every survey number as an upper bound to be tested, never as a final price, and weight a real pre-sale far above any hypothetical answer when the two disagree.

Anchoring respondents with your own price before you ask. Mentioning a specific number before the pricing questions — even in passing, even as "just for reference" — pulls every subsequent answer toward that anchor, whether the respondent means to be influenced or not. Run pricing questions before you show a pricing page, and if you're testing Gabor-Granger price points, randomize the starting price across respondents so your own default doesn't become the sample's average.

Asking your existing network instead of your target buyer. A price question answered by friends, existing users who already like you, or your own team measures social pressure, not market willingness to pay — the same trap covered in the prerequisites above, worth repeating here because it's the single most common way a pricing study gets quietly invalidated before it starts.

Running only one method and treating its output as final. Van Westendorp alone gives you a range with no purchase commitment behind it. Gabor-Granger alone gives you a curve built on hypothetical yeses. Interviews alone give you a small, deep sample that may not generalize. Madhavan Ramanujam's Monetizing Innovation makes the broader case that pricing research fails most often not because the wrong method was chosen, but because a single method's output got treated as a finished answer instead of one input among several — the studies that hold up combine methods and then confirm the result against real payment behavior.

Skipping the competitive anchor entirely. A willingness-to-pay number that ignores what buyers already pay for adjacent tools is incomplete — pair this research with a competitive pricing analysis so your final number accounts for both what buyers say they'd pay and what the market has already trained them to expect.

Key Takeaways

• No single method produces a final price. Van Westendorp gives a range, Gabor-Granger gives a curve, and direct interviews give a spending-behavior sanity check — combine them rather than picking one and stopping.
• Hypothetical willingness to pay always runs ahead of real willingness to pay. Every stated-preference answer is an upper bound to be tested, not a number to build a pricing page on directly.
• A real pre-sale outranks every survey method. When a pre-sale conversion disagrees with your survey data, believe the pre-sale.
• Anchor interview questions to past spending, not future hypotheticals. "What did you pay last time" beats "would you pay this" for the same reason The Mom Test prefers past behavior to future opinion.
• Recruit from real target buyers, not your existing network. A price question answered by people who already like you measures goodwill, not market willingness to pay.
• Pair willingness-to-pay research with competitive pricing data. What buyers say they'd pay and what the market has already trained them to expect are two different inputs to the same number.
• Treat pricing research as ongoing, not a one-time gate before launch. Monetizing Innovation frames pricing as a discipline that keeps running after launch, not a question you answer once and file away.

Frequently Asked Questions

How many people do I need to survey for willingness to pay research?

It depends on which method you're running and how precise you need the output to be: direct interview asks can produce useful directional signal from a small, well-targeted group, while Van Westendorp and Gabor-Granger need a broader respondent pool before their curves stop looking noisy. Start with a handful of interviews to catch obvious pricing mistakes, then scale up sample size for whichever structured method you run next.

Is Van Westendorp or Gabor-Granger more accurate?

Neither is strictly more accurate — they answer different questions. Van Westendorp maps an acceptable range without asking anyone to commit to a specific number, while Gabor-Granger forces a yes/no commitment at specific price points and produces a demand curve. Most practitioners use Van Westendorp to find the range and Gabor-Granger to sharpen a price inside that range, the same sequence covered in the worked example above.

Can I skip willingness to pay research and just match competitor pricing?

You can, but you'll be pricing against what the market already expects rather than what your specific buyer would actually pay for your specific offer — the two often overlap but aren't the same number. A competitive pricing analysis is a useful input alongside willingness-to-pay research, not a substitute for it, especially if your offer is differentiated enough that competitor pricing isn't a reliable anchor.

What's the difference between willingness to pay and a pre-sale?

Willingness to pay research (Van Westendorp, Gabor-Granger, interviews) measures what people say they'd pay, using hypothetical or semi-hypothetical questions that cost the respondent nothing to answer. A pre-sale asks for a real payment, which is why it sits above every survey method as evidence — see pre-selling your product for how to run one ethically once your survey research has narrowed the price down to a testable number.