Gabor-Granger Method Mechanics for Product Teams

Gabor-Granger gives product teams a structured path from purchase-intent surveys to a revenue-maximizing price candidate. Its value depends on getting three things right, namely how prices are presented, how the resulting curves are read, and where the method stops. It replaces pricing hunches with evidence, but it does not turn survey answers into guaranteed sales.

What the Gabor-Granger method is and what problem it solves

A cost-plus calculation tells a team what price covers its costs and adds a margin. It does not tell that team how many buyers would accept the price, or whether a different price could generate more revenue. A list price based on a hunch leaves the same gap. Gabor-Granger addresses it by measuring stated purchase intent at specific prices and using those answers to build a demand curve.

The method changes the price while keeping the offer the same. Researchers call this a univariate design, which simply means that only one variable changes. That narrow focus makes the output clear and actionable, because the team can see how stated demand changes as the price moves.

Where the method fits — the right conditions for running a Gabor-Granger study

Gabor-Granger works best when the offer is defined and the team already has a credible price range. The team still needs to choose a price in that range to consider more closely. A new product or line extension can meet those conditions even without sales data. So can a relatively simple subscription offer, an annual pricing review, or a decision about how low a promotion can go.

The same filter applies across markets and teams. A monetization lead adjusting subscription rates and a growth team reviewing campaign prices can both use it to test specific pricing hypotheses. For a multi-market launch, separate curves can show how price sensitivity differs by country before regional prices are set. Testing a fixed tier is different from asking buyers to choose which features belong in each tier.

An undefined concept fails that test because respondents cannot judge an offer that keeps changing. An unknown price range creates a different problem, since implausible price points produce noisy answers. To resolve those gaps, teams should establish the offer and explore the range before running the study. Categories that are very new may also need qualitative research and pilots because buyers have little to compare them with. When complex bundles or feature trade-offs dominate the decision, conjoint or discrete-choice experiments fit better.

How the sequential monadic design works in practice

Sequential monadic means showing one price at a time. In the classic ladder design, which price comes next depends on the respondent’s answer. Acceptance sends the ladder upward, while rejection sends it downward until the respondent accepts or reaches the lowest tested price. The aim is to locate each person’s highest accepted price within the tested range using a selective sequence of questions.

Consider a simplified ladder with three price points, which are low, middle, and high. One respondent starts at the middle price and accepts it, then rejects the high price. A second respondent says no to the middle price, moves down, and says yes to the low price. These two paths demonstrate the branching logic without pretending to establish an exact willingness to pay between the tested points.

A full ladder usually has five to seven prices, from conservative to expected and higher options, all within a range the business could realistically charge. The starting price should be randomized to avoid anchoring everyone to the same figure. The wording of the question should stay the same as prices change. AMC Insights specifies the wording as “At [Price X], how likely would you be to purchase [Product Name]?” Survey tools such as Qualtrics can automate the branching.

The response scale changes how much detail the study captures. Binary yes/no answers are quick but coarse, while a five-point Likert scale runs from “definitely would buy” to “definitely would not buy.” Treating “definitely” and “probably” as acceptance creates a top-2-box demand measure, but practitioners should sensitivity-test that threshold. Design variants also trade sample size for less anchoring. A randomized subset of prices, typically three to five, requires a larger overall sample, while single-point monadic shows each person only one price and needs separate price cells to reconstruct the curve.

Reading the output — demand curve, revenue curve, and price cliffs

The demand curve starts with the share of respondents willing to buy at each price. With a Likert scale, the analysis can instead use weighted purchase probabilities. Picture price moving from left to right and demand sloping downward. If raw responses produce anomalies, monotonic smoothing keeps the curve from rising as prices increase. Separate curves for heavy and light users, or by geography or channel, can show differences that a curve for the whole audience can hide.

The revenue curve adds the price to that picture. At each tested point, multiplying the demand percentage by the price produces a directional revenue index. A higher price can make up for losing some buyers, so revenue can hold up even when demand falls. Eventually, lost demand may outweigh the gain from charging more. Picture the revenue curve rising to a peak and then falling, with that peak identifying the revenue-maximizing candidate among the tested prices.

A price cliff deserves especially close attention. It is a small price increase followed by a disproportionate drop in stated demand. If a small reduction roughly doubles demand, buyers are highly price-sensitive around that threshold. If large increases barely reduce demand, buyers are less sensitive within that range.

Before using the peak, teams should account for hypothetical bias through corrections to stated intent. Those corrections address respondents’ tendency to overstate what they would actually buy. Uncertainty also remains around the peak, especially with a small sample, so bootstrapping can help quantify a confidence interval rather than present one price as unquestionably correct. Revenue is not profit, either. Adding variable costs or contribution margin can identify a different profit-maximizing candidate.

Where Gabor-Granger stops and other methods begin

Gabor-Granger leaves out competitive choice, including how demand would shift if a competitor cut its price or introduced a comparable lower-tier offer. That omission becomes important when brand equity, switching costs, or substitution drive the purchase. The resulting curve describes acceptance of the presented product.

Nor does correcting stated intent turn it into revealed behavior. A respondent still has not paid. The revenue curve thus gives a directional index, and using it as a sales forecast risks basing plans on demand that is too high. This boundary of the evidence persists even with a smoother curve. Real purchasing requires a separate validation step.

The other boundary is the fixed offer itself. When buyers must weigh features against price or choose between basic and premium tiers, conjoint or discrete-choice experiments can test how they make those trade-offs. When the unknown is the acceptable price range, Van Westendorp is the more useful starting point. It asks four open-ended questions covering too cheap, cheap/bargain, expensive, and too expensive, without supplying preset figures. The handoff rule is straightforward, with range exploration going to Van Westendorp, specific prices for one offer going to Gabor-Granger, and feature-price choices going to conjoint.

How teams combine Gabor-Granger with other methods and validation steps

The recommended workflow begins with what buyers pay today. Current solutions establish the commercial context, especially in B2B, where buyers rarely accept prices at multiples of current spend even for a demonstrably better product. Van Westendorp can then map an acceptable range before the team chooses its test prices. Gabor-Granger narrows that range using a ladder of conservative, expected, and stretch options. A live A/B pricing test then checks the candidate against actual behavior before full commitment.

Segmentation can change the pricing decision. Separate demand and revenue curves for freelancers and SMB admins, or for different geographies, may peak at meaningfully different prices. A blended result can obscure those differences. In that situation, the evidence can support considering a segmented pricing architecture rather than treating one audience-wide peak as the only option. The segments still need to be analyzed as distinct buyers of the defined offer.

Commercial costs add another layer to that decision. Subtracting variable costs or applying contribution margin shifts the focus from maximum revenue to maximum profit. Where distributors are involved, channel terms can be incorporated through a price waterfall that accounts for the deductions affecting the realized price. The survey supplies the demand evidence that the team combines with those economics before validating a price in the market.

What AI moderation and synthetic audiences add to a Gabor-Granger study

A conventional survey records acceptance or rejection but does not necessarily reveal why someone hesitated. Conveo describes AI-moderated video interviews that probe in the moment when a participant hesitates at a price. That follow-up can distinguish budget constraints from a perceived value gap or an attractive competing offer. At a price cliff, those explanations help a team understand what the drop represents.

Synthetic audiences can help before human fieldwork starts. They can screen price acceptance, identify possible demand drop-offs, and explore directional revenue trade-offs under synthetic evaluation conditions. Those outputs are not market transactions. Their practical use is to eliminate weak price points, sharpen questionnaire wording, and focus the human study on the transitions that need closer examination. They can also generate targeted validation questions where the simulated demand shifts most sharply.

The validity gap limits how far those results can go. Synthetic panels can overestimate willingness to pay, and subgroup error can be substantially larger than topline error. That makes segment-level pricing conclusions especially uncertain. The recommended approach is a hybrid sequence, with synthetic audiences screening hypotheses and shaping study design before human respondents validate them. Its value is a more focused, less costly human study.

Running Gabor-Granger continuously rather than as a one-time study

Subscription rates, tier structures, and promotional floors return for review. Annual pricing reviews already allow for changes in price sensitivity as competition, market context, and the mix of customers change. Multi-market launches bring choices about prices in each region, and campaign pricing needs to be reviewed again over time. A launch-only study cannot answer all of those later questions. Rebuilding the workflow each time also makes it harder to track how demand has shifted.

Keeping audience definitions and category qualifications consistent makes successive curves comparable, so a team can track the movement of a price cliff. Saving price-ladder logic and the wording of questions also means less setup work next time. Synthetic pre-screening can cut the added cost of another study, making it easier to run reviews each quarter or when a campaign calls for one without hiring an agency each time.

Speed helps only when it improves the decision. AI-native platforms can compress fielding and analysis from a quarter-long schedule to 24–72 hours, making research within a sprint cycle feasible. Pricing decisions close on a date regardless of whether current evidence is ready. A continuous capability gives teams an alternative to defending changes from memory or outdated studies. Over time, comparable demand curves and reusable study components become a pricing research asset whose value compounds beyond launch.