What do you do if you are launching a brand new product or service, and there is nothing on the market to give you a guide to what a reasonable price should be? How do you establish a price?
One method is the Gabor-Granger pricing model.
The model is used to determine the price elasticity of demand for a product or service. It’s named after the economists Andrew Gabor and Clive Granger, who developed it in the 1960s. It attempts to identify the point at which the perceived benefits of a product or service are equal to its price.
The model is relatively simple to use. It involves asking potential customers a series of questions (such as about features or benefits) about your product or service, including how much they would be willing to pay for it. The answers to these questions are used to plot a demand curve, which shows the quantity of the product or service that customers are willing to buy at different price points.
The pricing question can be a simple ‘what would you pay for this’, but is more often a propensity to buy question. To do this, you first need to have an idea about the range of prices inside which you think the actual price is likely to lie - let’s say (from some initial research) that you think the optimum price is likely to be between £30 to £40.
You then split this range into a number of individual prices, e.g. £30, £32, £34, £36, £38 and £40. For the customer you are talking to, pick a price at random (let’s say £34), and ask a propensity question such as ‘please rate your likelihood that you would buy at this price’. The scale would be ‘definitely no’, ‘probably no’, ‘unsure’, ‘probably yes’ and ‘definitely yes’.
If the respondent chooses one of the top two (probably or definitely ‘yes’), then they are tested with a random higher price; if they don’t choose one of the top two, they are tested with a random lower price.
This continues until you have discovered the highest price at which they choose one of the top two - the price above that will have been an ‘unsure’ or one of the ‘no’s.
Ideally, you want to have between 8 to 16 price options for customers to choose between.
The process above is very labour intensive, as it assumes you individually talk to each customer. Of course, it’s much easier if you automate the survey using software, and ask a large number of potential customers all at once. If you do this then you need to tweak the process. Instead of asking about one price and then selecting a random number above or below depending on the answer, you use a survey platform that repeats the same question for every price but allows the order of those questions to be randomised for each participant.
How many customers do you need to survey? Whichever approach you use you should probably aim for more than 100, but the actual number depends on how confident you want to be in the answer and the size of your potential customer base.
There are online calculators - such as https://www.surveysystem.com/sscalc.htm - to help you determine how many customers to survey. You have to decide on the level of confidence you want (95% confidence in the result is a common objective) that the % of customers who would buy at each price is within a confidence interval (such as 5%).
Ok, what does that mean? Let’s say that you want to be 95% confident that the result you get, such as 52% of customers would buy at a certain price, is going to reflect reality within +/- 3%. This would mean that, if every single possible customer had been surveyed, you can be 95% confident that between 49% and 55% (which is 52% +/- 3%) would buy at that price. For this, if you have 10,000 potential customers, you need to survey 964 of them.
If you want to be 99% confident within +/- 2% then, for 10,000 potential customers, you need to survey 2,938 of them.
What does the output look like? You plot a count of the maximum price the customers were prepared to pay, and it looks like this:
The orange line shows the % of customers prepared to buy at each price (right hand axis); the blue line shows the revenue at each price (left hand axis). In this example, the optimum price to maximise revenue would be £33.
The Gabor-Granger model has several strengths.
First, it is relatively simple and easy to use, especially when automation (such as using survey software) is employed.
Second, it can be used to estimate the price elasticity of demand for a wide range of different products and services, and is especially useful where the product or service is new and there is no existing market expectation regarding price.
Third, because you get an estimate of demand at each price, you can model the optimum price to maximise revenue; or, if you calculate the net margin at each volume, you can model the optimum price for maximum profit, which is even better.
However, the Gabor-Granger model also has some weaknesses.
First, it assumes that customers have perfect information about the product or service in question. In reality, customers may have incomplete or inaccurate information, which can affect their willingness to pay.
Secondly, it also assumes that customers are behaving in the survey the same way they would behave in real life. This is a major issue with all pricing models that try to establish future intent. When customers are asked ‘what would you do if…’ then two things happen: first, they imagine an ideal logical version of themselves; and second, they engage their conscious brain to assess what that ideal version of themselves would do.
In reality, when we actually make a pricing decision, it is taken subconsciously using the typical heuristics we use to make approximately right decisions, and is not taken using logic or reason. For example, in reality a customer could be massively swayed by an online influencer or by knowing the product or service is really popular with their friends.
To get around this, you need to find ways to put the customer into as realistic a scenario as possible, and then ask them the questions.
Third, it assumes that customers will answer the buying price question honestly and accurately. However, customers may have a tendency to overstate or understate their willingness to pay, depending on a variety of factors such as social desirability bias or budget constraints.
Fourth, it ignores other psychological factors such as those where the price itself communicates something about the product. For example, in the days when all vacuum cleaners cost around £60-£70, this model (and to be fair, no pricing model) could not have helped Dyson establish his launch price of £399 for his bagless vacuum cleaner. But in the consumer market for vacuum cleaners that price communicated that there was something very special (and desirable) about the product. In a B2B market it communicates lower risk.
Fifth, your customers are not stupid - they will clearly understand that the purpose of the survey is to establish a price, and might deliberately understate their willingness to pay to ‘game’ the company into setting a lower price.
Furthermore, it ignores external and competitive elements. In an ideal world, every possible customer might actually buy the product or service; in the real world, competitors will introduce competing products, economies go up or down, or many other factors could depress the actual demand.
Finally, the model assumes that the demand curve is linear. However, in reality, demand curves are often non-linear, with elasticity changing at different price points. This is especially true when the first digit of a price changes - there might be a smooth curve from £36 to £37 to £38 to £39, and then a significant drop at £40 after which the curve becomes smooth again - this is left-digit bias. Of course, if that full range of prices are captured in the survey, then the demand and profitability at each price point can still be modelled.
Despite its limitations, the Gabor-Granger model remains a useful tool for estimating the price elasticity of demand for your product or service.