Gabor Granger method for pricing surveys (how it works, setup, when it's wrong)
Clive Granger won a Nobel Prize in 2003 (NobelPrize.org). It wasn't for this. Before the econometrics that got it, he was a young statistician at the University of Nottingham working with an economist called André Gabor on a smaller question: do shoppers know what things cost? The method they built to answer it is still the fastest way to find the price that earns you the most revenue.
This guide covers what Gabor Granger is, what it measures, the chart it produces, how to set up a Gabor Granger survey, and when it's the wrong research tool to use.
What is the Gabor Granger method?
The Gabor Granger method is a pricing survey that finds the highest price each participant will pay for a product or service, by showing them one price at a time with a yes/no question: would you buy this at this price?
Say yes and the next price goes up, say no and it goes down. The survey narrows the range of possible prices after each vote until it finds the highest price that person is willing to pay.
It works because it never asks anyone to do maths about their own preferences. "How much would you pay for this?" asks a person to come up with a number under no pressure, and people are generally bad at that. They lowball their price to nudge you cheaper, or reach for a round figure that sounds reasonable in theory. "Would you buy this for $24?" is a question people answer accurately in shops every week, so they answer it accurately in a survey too.
Gabor and Granger published this across the 1960s, starting with a Nottingham study of fifteen grocery products (On the Price Consciousness of Consumers, Applied Statistics, 1961), then extended it into price as a quality signal (Economica, 1966).
How the survey platform picks the next price to show after each vote is a critical factor, and tools handle this very differently. Some use inefficient randomisation or simple sequential ordering (yes to $10 leads to $15, then $20, etc.), while others like OpinionX use more specialised methods like binary search to half the remaining range every vote, meaning you can test 100 prices in just 8 votes per person. Our guide ‘How a Gabor Granger survey picks the next price’ covers why this selection method is so important and how it determines the number of prices you can afford to test in your survey.
Try an interactive Gabor Granger example
The best way to understand how Gabor Granger works is to vote in a Gabor Granger survey yourself. The example below takes just thirty seconds to complete, requires no login, and shows you the interactive results at the end:
What does a Gabor Granger survey produce?
A chart with two curves and one key number.
Count how many people's maximum price sits at or above each price point and you get a demand prediction. Multiply demand by each price point and you get average revenue per target customer.
For example, here's a monthly coffee delivery subscription tested through Gabor Granger with a testing range of $10 to $38:
| Price | Demand | Average revenue per target customer |
|---|---|---|
| $10 | 86% | $8.60 |
| $18 | 66% | $11.88 |
| $26 | 45% | $11.70 |
| $30 | 34% | $10.20 |
| $38 | 14% | $5.32 |
One multiplication per row delivers the key insight. At $26, 45% of the market buys, so $26 × 0.45 = $11.70 from every target customer, buyers and non-buyers alike. Averaging across everyone rather than just buyers is what makes the rows comparable.
Plot these numbers on a line chart and you’ll see that the revenue line has a clear peak in the middle. The top of that peak is your optimal price, ie. the price that generates the maximum total revenue for your population of target customers.
Look at the column again though: $11.88 is the average revenue per customer when the product costs $18, while the average revenue remains almost the same at $11.70 for a $26 price tag. This shows an 18-cent difference across a price range where you'd be charging 44% more. Gabor Granger often hands you a revenue-neutral band instead of a single number, and choosing inside that band is a strategy call about margin against market share. How to read a Gabor Granger chart covers the four key numbers that a Gabor Granger survey produces, how to understand them, and the common mistakes people make when interpreting their results.
How do you set up a Gabor Granger survey?
When you create a new survey on OpinionX, even on the free tier, you’ll see a question type called Gabor Granger listed under the ‘Advanced’ section of the question menu. Once you click that, you’ll have five quick decisions to make before you’re ready to launch your survey:
1. Describe what's being priced. A clear product image and a description naming what's included. Vague context is where pricing data goes noisy: half your sample ends up pricing something different from the other half.
2. Set the price ladder. Minimum price, number of prices you want to test, and either the interval (how big the gaps should be between each price) or the maximum price. Ten prices is a reasonable default, but the more you can test, the better your results will be. The price range you choose decides the answer, for example picking a max price that’s too low will hide the true optimal price. Our guide on how to choose your price points explains this in detail.
3. Set the prefix and suffix. The suffix decides what you're measuring, for example ‘per month’ prices a subscription, ‘/seat’ prices a variable purchase that charges per seat, or ‘per 100 credits’ explains a usage-based model. The prefix is commonly used to set the currency symbol that comes before the number. This helps clarify the pricing details to survey participants so everyone shares the same understanding when voting.
4. Keep the interest prescreener on. Before people start the yes/no voting, it asks whether they would buy at any reasonable price, and routes the never-buyers out beforehand. Without it you can't tell a ‘no’ about your price from a ‘no’ about your product overall. Turn it off only when your list is already qualified.
5. Keep the above-max question on. If someone says yes to your top price, you don't know their ceiling, only that it's higher than you looked. This asks them.
Picking a floor and a ceiling for something nobody has priced before is where people get stuck. A research expert will walk through your ladder and your audience with you before you launch.
How many participants you need depends on what you want from the results. Around 100 participants shows you roughly where the optimal price sits. Around 300 gives you an optimal price you can be confident in. If you plan to filter the results to look at one customer segment on its own, aim for 500 or more, because each segment contains only part of your participants. A segment that makes up a quarter of 500 participants gives you results from just 125 people.
To help you understand how many participants your pricing survey needs, take a look at our Gabor Granger sample size calculator. If you’re sourcing your own participants without a paid recruitment panel, check out our guide on finding participants for free.
What are the disadvantages of the Gabor Granger method?
The Gabor Granger method is good at one job: telling you which price in your range produces the most revenue per target customer. Its weaknesses show up when people ask it for more.
The main one is that people overstate what they'd pay when no money changes hands. Researchers call this hypothetical bias, and it's large. The largest meta-analysis on consumer products found survey answers ran about 21% above what people paid (Schmidt and Bijmolt, Journal of the Academy of Marketing Science, 2020), and an earlier one across 28 studies found stated values running at 1.35 times real ones (Murphy et al., Environmental and Resource Economics, 2005). A few participants lowball on purpose once they work out you're setting a price, but not enough to cancel the overstatement. Gabor Granger is strong when it comes to identifying which price will generate the maximum overall revenue, but the exact estimate it shows for average revenue per target customer isn’t a predictive number you can rely on. Keep the chart’s revenue figures out of any forecast.
That's no reason to switch to a fancier method, though. The same 2020 study found indirect methods, conjoint analysis among them, overstated willingness to pay by roughly 10 percentage points more than direct questions like Gabor Granger's.
The second weakness is that the answer can only be as good as the range you tested. If most people's real ceiling sits above your top price, the peak lands on your top price because there was nowhere higher to go. The No Ceiling and All Rejected boxes on OpinionX’s results dashboard for Gabor Granger warn you when this has happened, and our guide on how to choose your price points covers setting a range that avoids this issue.
Third, the chart knows nothing about your costs. It finds the revenue-maximising price, and unless extra customers cost nothing to serve, the most profitable price sits somewhat above it.
Finally, one price for a whole market hides the decision you probably need to make. The peak averages together segments who'd pay very differently, and the average of a $15 segment and a $40 segment is a price neither of them wants. Segment the result before anyone writes a number on a slide.
Each of these has a fix, and our guide to the common mistakes in Gabor Granger studies walks through these fixes one at a time.
When is the Gabor Granger method the wrong tool?
You have no price anchor. You can't build a ladder around a number nobody can guess within a factor of three. Run Van Westendorp first to get a band, then set the ladder inside it. Van Westendorp and Gabor Granger in one survey covers running them together.
Price trades off against features. Gabor Granger prices one fixed offer. If the question is what should be in the box at that price, you want conjoint analysis, or marginal willingness to pay for the value of one specific upgrade. When not to use conjoint analysis is the counterweight.
Competitors are in the decision. Neither Gabor Granger nor Van Westendorp puts an alternative in front of the participant. If your price only makes sense next to what the alternative costs, you need a discrete choice experiment or a market simulator.
You don't know what to price yet. Rank the features first with MaxDiff, then price the winners.