Van Westendorp in B2B: Where the Price Sensitivity Meter Breaks

TL;DR The Van Westendorp Price Sensitivity Meter asks one respondent four price-perception questions and intersects the cumulative curves into an acceptable price range. In B2B SaaS it breaks at four structural points: the respondent is not the buying group, the single price it asks about does not exist on any B2B contract, stated willingness to pay runs well above what buyers actually pay, and the questions assume a reference price enterprise buyers do not carry. It remains a legitimate cheap directional read; treating its output as a B2B price decision is the failure. The replacement is evidence from transactions.

The Price Sensitivity Meter may be the most frequently recommended pricing survey in software. Four questions, a spreadsheet, two intersecting curves, and out comes a range that looks like measured demand. Survey platforms ship it as a template. Product managers reach for it because it is fast, cheap, and produces a chart the room accepts.

In consumer markets, that trade can be reasonable. In B2B software, the method fails before the first response comes back. The failures are structural, baked into what the four questions assume about who answers, what is being priced, and how answers relate to money actually changing hands. Even when the survey is fielded flawlessly, the team inherits every one of them.

What the Van Westendorp Price Sensitivity Meter Actually Does

The method asks each respondent four questions about one product. At what price would it be so cheap you would question its quality? At what price would it be a bargain? At what price does it start getting expensive, though you would still consider it? At what price is it too expensive to consider?

Plot the cumulative responses and the curve intersections define the output: a point of marginal cheapness, a point of marginal expensiveness, the acceptable price range between them, an indifference price read as the perceived going rate, and an optimal pricing point where the fewest respondents object in either direction. The method dates to 1970s consumer research, built for products with posted prices that one person decides to buy.

The output is a map of price perception: what individuals can imagine paying for a described product, collected with nothing at stake. Every structural break below follows from that.

What the 1976 Paper Actually Claims

The paper that introduced the method, presented at a market research congress in 1976, is more careful about it than most of the people selling it today. Its author calls the approach “essentially psychometric”: a measure of how people perceive prices. There is no purchase question in the instrument, because “we don’t use buying-intention questions as a standard part of the technique.”

On the indifference price, the number most teams treat as the market’s answer, the paper says it “generally represents either the median price actually paid by consumers of the product, or the price of the product of an important market-leader,” and that it “is based upon people’s experience with price-levels in the market.” The meter hands back the price the market already charges.

The warnings are the author’s own. “Price-consciousness of this nature should never be equated with propensity to buy. One can be fully conscious that a product is ‘expensive’ and yet prefer it over a cheaper alternative.” On the optimal pricing point: “No relation to the ‘optimum-theory’ in economy is meant.” It is the price with the fewest objections, which is a different thing from the price that makes the most money. On the foundations: “there is certainly not yet a well developed theory behind PSM,” and “proof is needed, that PSM perceptual data has also a behavioural meaning.”

Everything the paper tested was a consumer product with a posted price: coffee, margarine, electric razors, a newspaper subscription. The author placed the meter beside panel data on what people actually bought and asked his own profession to study “the conditions under which it works and the conditions under which it most certainly does not.” The behavioural proof he asked for is what the peer-reviewed work since then went looking for, and it found the gap running the other way: stated price perceptions sit above what the same people pay. The four breaks below are where that gap opens in B2B software.

Break One: The Survey Respondent Is Not the B2B Buying Group

A Van Westendorp survey lands on whoever answers it: a user, a team lead, a director who saw an invoice once. A B2B software purchase is decided by a buying group: economic buyer, users, IT, security, finance, procurement. Procurement in particular brings discipline the survey cannot simulate: reference net prices from other vendors, a negotiation calendar, budget authority the respondent does not hold.

The four questions collect one person’s price perception. There is no slot in the exercise for any of the machinery that produces the actual purchase decision. Every individual elicitation instrument measures one person, and almost no B2B purchase is one person’s call; we work through the full argument in why willingness-to-pay surveys fail in B2B software. Worse still, the people most likely to answer your survey often have no idea what their employer pays for the software they use all day, a pattern we covered in our earlier look at willingness to pay.

Break Two: B2B SaaS Pricing Is a Surface, Not a Point

Every one of the four questions presumes the product has a price, singular. A consumer app with one posted plan satisfies that assumption. A B2B software offer is a schedule: a value metric that counts seats or API calls or tickets resolved, editions that gate capabilities, commitment terms, and volume bands. The number a specific customer pays is produced by where they land on that surface, and it differs by design from customer to customer.

Three distinct numbers exist where the survey assumes one. List price is the published figure. Scheduled net price is what the pricing structure produces for a given commitment. Landed net price is what the deal actually closes at after concessions. Which of the three is the respondent rating when they say “too expensive”? The exercise cannot tell you; the respondent does not know either. Our piece on price elasticity in B2B software works through why the gap between those layers swallows naive demand measurement.

The deeper problem is architectural. A software company’s monetization rests on three decisions in order: the licensing model, where the value metric lives; the packaging, which capabilities sit in which editions; and the pricing itself, the surface of scheduled net prices across commitments. The four questions compress all three decisions into one imagined sticker. An acceptable range for a number that never appears on a contract is not an input to any of them.

Break Three: Van Westendorp Overstates What Buyers Actually Pay

Peer-reviewed pricing research has run the direct comparison many times: ask one group to state what they would pay, require another to back their stated number with real money. Stated numbers come in high. Across published consumer studies the overstatement averages about a fifth, it runs larger for complex, high-consideration products, and the sharpest single field experiment found stated numbers nearly twice what the same buyers paid when the purchase was real. Enterprise software sits at the complex, high-consideration end of that spectrum.

The validation most often cited for the Price Sensitivity Meter was run on an inexpensive chocolate. The method’s optimal pricing point did land near the benchmark from an incentive-aligned mechanism, and the researchers’ own explanation was a coincidence: hypothetical inflation pushed stated answers up while the method’s focus on minimum customer resistance pulled its recommendation down, and the two biases happened to cancel. They called for validation on expensive industrial products. Decades later, it has not materialized. B2B use of the method rides on a cancellation observed once, on a confection.

About that “optimal” point: it is the intersection that minimizes stated objections. Nothing in the mathematics connects it to revenue, margin, or value capture. A number that annoys the fewest survey respondents is a strange target to steer a software company by.

There is a further problem with that point: it does not hold still. Re-derive the four intersections across resampled sets of the same respondents and they move, widely enough that the acceptable range covers several editions’ worth of price. Peer-reviewed work argues the method deserves better mathematics than raw intersections, and demonstrates that statistical modelling of the same answers narrows the band considerably. It does. But modelling the responses more carefully improves the precision of the estimate without changing what was estimated, which is a set of imagined purchases.

The peer-reviewed record on business software is two applications. One used the meter to decide which of two add-on modules a vendor should build first. The other put confidence intervals on the crossing points for a software launch. Both ranked perceptions, and neither compared the output with what those customers then paid. The papers that set out to supply the method’s theory reached the same place from the other side: the crossing points have no economic justification and are not tied to revenue, and the authors proposed other instruments in their place. In the newest comparison, the meter’s acceptable range sat well below what the same buyers gave to a direct willingness-to-pay question, so two survey methods disagree with each other before either is tested against a purchase.

Break Four: The Four Questions Assume an Anchor Enterprise Buyers Do Not Have

The method works, where it works, because the respondent already carries a price frame for the category. When a shopper rates a coffee maker, the shopper has already bought coffee makers, seen the shelf, and knows the going rate.

An enterprise buyer evaluating a novel B2B capability has no such frame. There is no shelf. What fills the vacuum is whatever is nearby: the budget line the purchase would draw on, the incumbent contract it would replace, the last renewal that crossed their desk. Peer-reviewed measurement work finds stated price answers track whatever reference prices respondents were recently exposed to, so the survey output moves with the anchors in the room rather than with the value of the product. A published field application of the method to a B2B software product showed the same mechanism: respondents reasoned from an existing reference frame, supporting quoted levels inside it and objecting to levels outside it, in both directions.

The “so cheap you would question quality” question fails on the same grounds. It leans on a price-as-quality heuristic that belongs to sticker-price shopping. Enterprise buyers do not infer quality from list price. They run demos, call references, pilot the product, and put it through security review.

What the Price Sensitivity Meter Is Still Good For

The method is defensible in the contexts it was built for: consumer and prosumer products with a posted price and a single decider. There, it is a legitimately cheap directional read for bounding early conversations.

It also has a real use inside a B2B company as a conversation starter. Fielding it against your own team, or a friendly user community, surfaces how differently people frame your product’s value, and the argument that follows is often more useful than the chart.

The failure mode is specific: treating the output as a price decision. Setting list price at the optimal pricing point. Presenting the acceptable range to the board as measured demand. Building an edition’s scheduled net prices from the indifference price. At that moment a directional sketch becomes the foundation of the monetization architecture, and real deals will find every structural break above.

What Replaces Van Westendorp in B2B Software: Evidence From Transactions

The alternative is not a better survey; it is the evidence you already generate. Won and lost deals at real landed net prices answer, under real stakes, what the four questions only gesture at. The record holds the buying group, procurement pressure, the discount pattern by customer group and commitment level, and the configuration choices: which editions buyers picked, which commitments they accepted, what they walked away from, all decided with real budget.

Across decades of engagements in our pattern library, the deal record consistently tells a different story than the survey deck sitting in the same company’s files: resistance shows up at levels and in customer groups the stated-preference exercise never predicted. A software company with a few years of closed deals is holding more pricing evidence than any survey will produce, and most have never mined it.

One tempting detour to close off: conjoint analysis is not the rigorous upgrade. It is a more elaborate instrument with the same defect, respondents choosing with nothing at stake, and we do not use it in any form. Real value-based pricing runs on evidence of value received and prices paid, not on imagination sampled at scale. If you want a fast read on whether your pricing architecture is built to produce and use that evidence, the Pricing Architecture Assessment scores it in a few minutes.

A Van Westendorp range may already be sitting in your pricing deck. The fastest test costs nothing: put the range next to your closed-deal record and see if they agree. If you would like help reading the comparison, talk to an expert. Describe what you are seeing, the range the survey gave you and what your deals actually closed at, and a pricing expert will reply.

FAQs