To find the optimal price for a product, ask buyers directly through a structured pricing research method matched to your question, rather than marking up cost or copying a competitor. Three survey methods cover almost every case. Van Westendorp finds the acceptable price range, Gabor-Granger picks the revenue-maximising point inside that range, and conjoint analysis models price against features and rivals.
Key takeaways
- There is no single optimal price. Pick the method that answers your question, then read “optimal” as the price that maximises the goal you set, whether that is revenue, volume, or margin.
- Use Van Westendorp to map an acceptable price range for a new or repositioned product with no reference price.
- Use Gabor-Granger to optimise a single price for an established product.
- Use conjoint analysis when price competes with features and competitors.
- Triangulate your survey data. A stated willingness to pay is not a purchase, so validate the answer against a sample large enough to segment and, where you can, against real buying behaviour.
This guide is for product managers, pricing leads, insights teams, and founders setting a price in a real situation, whether that is a new product with no price reference, a launch into an unfamiliar market, or a repricing under cost or competitive pressure.
What are the hidden costs of guessing a product’s price?
The hidden cost of guessing your price is that you permanently bleed profit and lose potential customers without even realising it.
A price set by cost-plus or competitor-matching leaves money on the table in both directions, and you rarely see which way. Set it too low and you give away margin you could have kept. Set it too high and you lose sales you never measure.
The effect compounds, because a price change moves profit faster than it moves revenue. Extra revenue from a higher price carries no extra production cost, so a 3% price increase can lift profit by 5% or more. That same 3% increase can also cut volume by 10% or more, depending on how price-sensitive your customers are.
Structured pricing research replaces the guess with evidence of how buyers actually trade money against value. The rest of this guide covers which method produces the right evidence for your question.
What does “optimal” actually mean for a product price?
Define the objective the price must serve before you pick a method, because “optimal” is meaningless until you name the goal.
The revenue-maximising price, the volume-maximising price, and the margin-maximising price are usually three different numbers, and a method tuned for one will mislead you on another. A subscription product chasing market share optimises for adoption; a premium product defending positioning optimises for margin per unit.
Write the objective in one sentence, name the metric you will move, and every later choice gets easier. Skip this and you will run a clean study that answers the wrong question.
Why can’t you just ask customers what they will pay?
Measure price through a structured method, not a direct “what would you pay for this” question, because the direct question produces a number no one acts on.
Asked outright, respondents lowball to look shrewd or anchor on whatever figure is in front of them, and stated willingness to pay drifts well above what they part with at the till.
The structured methods below get around the problem by inferring price sensitivity from choices and thresholds rather than asking for a price outright. This is the core reason a willingness to pay figure from research beats a number pulled from a sales meeting: it comes from a designed task, not an opinion.
Which pricing research method should you use?
Pick the pricing method from the question you need answered, because each method answers a different one and none answers all three. A new product with no reference price needs a range; an established product needs a single optimised point; a product whose value lives in its features needs the trade-offs modelled.
The three questions and their methods are:
- What range of prices will the market accept? Use the Van Westendorp Price Sensitivity Meter. Respondents name the price at which a product is too cheap, a bargain, getting expensive, and too expensive, and the intersections map an acceptable band and an indicative optimal price point. Van Westendorp suits new or repositioned products where no reference price exists.
- What single price maximises revenue or demand? Use the Gabor-Granger method. Respondents state purchase intent at a series of rising prices, which builds a demand curve, a price-elasticity estimate, and the revenue-maximising point. Gabor-Granger suits established products with a known price you want to optimise.
- How does price trade off against features and competitors? Use conjoint analysis. Respondents choose between competing product configurations with price as one attribute, and the model isolates the value of price against every feature. Conjoint analysis is the most realistic method because it mirrors a real buying decision, and the one to use when price is not the only thing that moves the sale.
These three cover most pricing questions, which the next section compares head to head.
Should you test price against competitors or on its own?
Put a competitive frame around the price wherever the product has rivals, because a price tested in isolation overstates what buyers will pay. A respondent shown only your product has no alternative to walk to; the same respondent shown your product beside two competitors prices it against the market, which is the decision they actually face. Conjoint analysis builds the competitive set into the task by design, and a brand-price trade-off study does the same when brand is the lever you are pricing against. Van Westendorp and Gabor-Granger test a product more or less alone, which is fine for a range or a first read but a known reason their numbers can run optimistic.
Can you trust a single pricing study?
Treat any one pricing study as a strong signal, not a verdict, and confirm it against a second source before you commit. A common and robust design runs Van Westendorp first to bound the range, then Gabor-Granger or conjoint inside that range to find the point, so each method covers the other’s blind spot. Where you can, validate the stated number against real behaviour: an A/B price test in market, historical sales at different price points, or a soft launch. Stated willingness to pay sits above revealed willingness to pay almost every time, so the in-market check tells you how far to discount the survey optimism before it reaches a price tag.
Van Westendorp vs Gabor-Granger vs conjoint: which should you use?
The four methods below answer different questions, so compare them on what they tell you rather than ranking them. The right one is the method that matches your product stage and the decision in front of you.
| Method | The question it answers | Best for | Main limitation |
|---|---|---|---|
| Van Westendorp Price Sensitivity Meter | What price range will the market accept? | New or repositioned products with no reference price | Measures perceived acceptability, not purchase intent; gives a range, not a single price |
| Gabor-Granger | What single price maximises revenue or volume? | Established products optimising a known price | Tests prices in isolation; can anchor on the prices shown |
| Conjoint analysis | How does price trade off against features and competitors? | Products where features and rivals drive the choice | More design and sample effort; needs a clear attribute set |
| In-market A/B price test | What price do buyers actually transact at? | Validating a researched price with real money | Slow, operationally costly, and exposes real customers to the test |
No method is the winner in the abstract. Van Westendorp and Gabor-Granger are faster and cheaper and answer narrower questions; conjoint is more work and answers the richest question; an A/B test is the only one that uses real money but is the slowest and riskiest to run.
The honest position is that the best method depends on your product’s stage and what you can afford to field, which the next section turns into fixes for the usual mistakes.
What goes wrong most often in pricing research, and how do you fix it?
Across the thousands of pricing studies run on the Conjointly platform, the same few patterns explain why a study fails to deliver its intended impact, and they come down to the design and the use of the results:
Treating willingness to pay as a fixed truth
A single survey number gets written into a business case as if it were a measured fact, when it is a stated intention under hypothetical conditions.
Solution: To avoid treating willingness to pay as fixed, read the research figure as the top of a range, triangulate it with a second method, and discount it toward revealed behaviour before it becomes a price.
Running the wrong method for the question
A new product gets a Gabor-Granger study that anchors respondents on prices the analyst invented, or an established product gets a Van Westendorp range when it needed a single optimised point.
Solution: To avoid running the wrong method, match the method to the question first, using Van Westendorp for an unknown range, Gabor-Granger for a known price to optimise, and conjoint when features compete with price.
Fielding too small a sample
A pricing study fielded on too few respondents produces a noisy curve and, worse, cannot show that two segments will pay very different prices.
Solution: To avoid an underpowered sample, size the study to the segments you need to price for, and confirm the number against a sample-size calculator before fielding.
Frequently asked questions
Can I find the optimal price using only historical sales and competitor data?
Partly, but not reliably. Historical sales at different price points and competitor benchmarking give a starting range, while a structured method such as Van Westendorp, Gabor-Granger, or conjoint is what turns that range into a defensible price.
How many respondents do I need for a pricing study?
Enough to estimate a stable curve and to split the segments you intend to price for, which for most pricing studies means a few hundred respondents, with the exact number driven by your design and segment count rather than a fixed rule.
Is conjoint analysis always better than Van Westendorp or Gabor-Granger?
No. Conjoint is the most realistic method and the right one when features and competitors drive the decision, but for a quick acceptable-range read or a single-price optimisation on an established product, Van Westendorp or Gabor-Granger answer the question with far less effort.
How often should I re-test the price?
Re-test when the market moves, namely a new competitor, a cost shock, a repositioning, or a major feature change, rather than on a fixed calendar, because a price is only optimal against the conditions it was measured in.
Conclusion and next steps
The best way to find the optimal price for a product is to choose the method that fits your question, field it properly, and confirm it against a second read before you commit. Work the problem in order and the method chooses itself.
- Write down what “optimal” means for this product: revenue, volume, or margin.
- Identify the product’s stage and pick the matching method, namely Van Westendorp for an unknown range, Gabor-Granger for a known price to optimise, or conjoint analysis when features and competitors drive the choice.
- Size the sample to the segments you need to price for and confirm it with a sample-size calculator.
- Field the study, then triangulate the result with a second method or, where possible, an in-market check.
- Set the price, and diarise a re-test for the next time the market moves.
For a fuller view of the methods and where each fits, start with Conjointly’s pricing research overview and the comparison of Gabor-Granger or Van Westendorp.
