ESOMAR Survey-Based Techniques for Optimizing Prices and Products: Overview and Best Practices

Are you curious about what customers want and what they're willing to pay? So are we. Actually, that is the most frequently asked question here at boobook so it’s crucial we keep learning to stay on top of all the pricing techniques and methodologies.

This January, I participated in the ESOMAR Training on Survey-Based Techniques for Optimising Prices and Products. Organised in collaboration with MRII, a non-profit educational institute linked with the University of Georgia, teaching professionals of the industry conducting effective and robust market research, the training was the first of the 2023 series. Two speakers, Ed Keller, executive director of MRII (Market Research Institute International) and Brian Orme, CEO of Sawtooth Software, an expert on Conjoint Analysis, shared their insights on how to optimise the pricing of your product or service.

Conjoint analysis

Conjoint analysis is the gold standard survey-based technique for uncovering valuable insights. While Sawtooth Software is a celebrated industry leader in Conjoint & Max Diff, there are also other techniques we'll explore in this overview. Qualitative research plays a key role in laying the foundation for successful quantitative surveys. And with Sawtooth Software's consulting division and survey business, you can trust you're in good hands for all your research needs. It's no wonder they've thrived for 40 years!

If you use a five-point or a ten-point scale, you'll get a lot of high-level numbers, making it hard to distinguish what is essential and what is not. You'll get a lot of straight-lining. There's another problem: cultural scale use bias. Different people use scales differently. This bias is very different across other countries. For example, respondents in Germany score lower on average than respondents from India.

Let's imagine that there are three different groups of people. One group wants to use ratings 5-10, the second want to use 3-8 and the third group uses 1-6. When I plot them between one item on my rating grid, then you'll get an artificial correlation of 46.

This makes it look like everything is related to everything else or drives everything else. Suppose you try to drive multi-variant statistical techniques to tease out the separate effects of things, like through regression; this artificial correlation will give you a lot of problems. This is why in conjoint, we don't use a typical 5 or 10-point scale.

With Conjoint, we are not just testing one thing at a time, e.g., A/B testing. We can vary the features and prices and test hundreds or thousands of realistic-looking products or pricing combinations. It's not A/B testing; it's A to a gazillion testing.

Conjoint is a survey-based technique to optimise features or pricing. We will show people realistic buying scenarios and ask them what they would choose each time.

See the example below: 

We are studying different attributes and features, and we are varying them across each choice task. Each choice task has the same format (see 1 & 2 below), but the features and prices vary each time. When we vary the features independently such that each feature is shown an equal number of times and each feature is shown with another feature an equal number of times, we have a fair and balanced experiment which allows us to tease out what is driving people's preferences. No need to ask them how important is colour or price to them. We are replicating the real world to them and understanding what is going on in their heads. We don't let the respondent get lazy; they can't just respond 5 – 5 – 5, like on a 5-point scale. There is also a "no" option, so they can choose to walk away from all possibilities.

Another advantage is that conjoint analysis works really well on mobile phones. 

• We can statistically tease out what is driving peoples’ choices. We use a model to calculate the scores of the preference utilities that the respondent chooses. You will get high utility scores for features that you pick again and again, e.g. the colour red or low prices.
• For each respondent, you get a full set of utility scores for each attribute, each feature that you are trying to optimise and each price. Based on this, you can predict how each respondent would choose in thousands or millions of different combinations. This is a market simulator that you can build in Excel.

You can change the fields in this simulator, choosing if the car is blue or red and updating the prices. Every time you change a cell, the simulator updates, and you will see the share of respondents who would choose that product with those features. 

It’s a very powerful tool to optimise the products you are offering in a realistic competition with your competitors in order to understand how to capture the biggest share based on the ideal features and portfolio you need to offer. 

Another technique that was developed at the same time as the Conjoint is the Van Westerndorp Price Sensitivity Meter, which is a survey-based technique used to establish an acceptable price range for one single product.

• You show your respondents one single product concept and educate them about the concept
• Then you ask the respondents 4 questions:  

1. At what price is it so expensive that you would not want to buy the product  

2. At what price is it so cheap that you would not buy it because you would doubt the quality  

3. At what price is this product acceptably expensive 

4. At what price is this product acceptably cheap 

See the graph below: 

At 100 dollars, 10% of respondents thought that the product was too expensive for them. The intersection has meaning, according to the author of this method.  

• The intersection of the YELLOW & BLUE = point of marginal cheapness 
• The intersection of the RED & GREY line = point of marginal expensiveness. 

The gap between these two points is thought to be the acceptable price range. 

• The optimal price point is the intersection of ORANGE & YELLOW 
• The indifference price point = intersection between BLUE & GREY  

Weaknesses: 

• Focuses only on 1 product concept.
• If you want to test thousand or hundreds of modifications of your product, it will be very difficult
• Doesn’t put the respondent in a realistic context as the conjoint does

Strength: 

• A quick methodology that you can use in surveys is good when the product is truly new to the market and you have no idea how to price it.
• Open-ended questions mean less chance of bias  

Gabor Granger Stated Willingness To Pay  

Finally, Gabor Granger is another survey-based technique that is used for establishing the price sensitivity curve & optimum price point for one product. You create a series of prices in a list and ask if they would buy your product at each of the prices in the list.

Start out by randomising the prices and ask them if they would pay that price. If no, then they are shown a lower price until you get to the price they would pay. 

Let’s say for respondent 1, we randomly select 25 dollars. If they say yes, they would pay that price, we randomly show them a higher price, eg 40. Then they say no, so we randomly show them a price between the two price points. We do this until we get to the highest price they would be willing to pay. 

• We then can create a price sensitivity curve 

Weaknesses: 

• Only focuses on one product at a time in a vacuum, with no competitors, which is not a realistic market scenario for conjoint 
• It’s obviously a price game which may lead respondents to adopt a bargaining behaviour 

Strengths: 

• It’s a quick and dirty quantitative method
• Overall recommendation for pricing experiments
• Asks people realistic questions that mimic the real world
• Doesn’t ask people what they would pay
• Has them choose among different products or different prices like they would in real life

Best practices for conjoint studies

1. Getting your attributes right is crucial. This is why Qual is so useful. 
2. Recruit people who want to buy your product. 
3. There is a lot of bad data from respondents who do not answer realistically or carefully. Luckily conjoint analysis and max diff offer a fit score. We can compare the questions to see if respondents are answering consistently, and you can throw them away if they have a low fit score.  
4. Hire someone to show you how to do it and use good software. 
5. It’s typical to use samples of 200-800 but you can do more or less, this is just the norm. Sometimes we run models with 40-50-60 respondents because that’s all we can get. The models still work, it’s just that they are not as precise. But it’s still very worth it rather than not doing it at all. 
6. It’s typical to show 8-15 conjoint questions for each respondent. Each question takes about 10 to 15 seconds to answer, it’s a 2–4-minute survey once you have educated your respondents on your subject matter. 
7. In your educational feed up to your conjoint, make sure you aren’t overselling it and biasing your respondents to be positively predisposed to it 

Conclusion

To sum up, the ESOMAR Training on Survey-Based Techniques for Optimizing Prices and Products was an invaluable opportunity to deepen my understanding of pricing strategies. With expert advice from Ed Keller and Brian Orme, I was able to further develop my knowledge base in this area and come away with a better understanding of what it takes to set up an effective pricing study. Additionally, talking with other professionals at the event gave me insight into how they approach pricing, allowing me to adopt new strategies that are suitable for my market research goals.

If you find yourself in need of help setting up your own pricing strategy, get in touch. Together we can work towards creating a pricing structure that will optimize your company’s products or services for maximum profitability.

Never miss a story. Subscribe to our newsletter!