METHODS FOR PRICING RESEARCH
Several different research methods are commonly used in pricing research—each with their own strengths and weaknesses. This document discusses four techniques that are commonly used by survey researchers. The four techniques are: • • • • van Westendorp Price Sensitivity Meter Concept Test Conjoint Analysis Discrete Choice Modeling
Price Sensitivity Meter (van Westendorp)
Introduced in the 1970s by a Dutch economist, Peter van Westendorp, the Price Sensitivity Meter (PSM) is used fervently by some researchers. The premise of the PSM is to ask respondents four price-related questions and then evaluate the cumulative distributions for each question. Specifically, respondents are asked: 1. At what price would you consider the product to be so expensive that you would not consider buying it? (Too expensive) 2. At what price would you consider the product to be priced so low that you would feel the quality couldn’t be very good? (Too cheap) 3. At what price would you consider the product starting to get expensive, so that it is not out of the question, but you would have to give some thought to buying it? (Expensive) 4. At what price would you consider the product to be a bargain—a great buy for the money? (Cheap)
The cumulative frequencies are plotted, and the four key intersections are interpreted. The point at which an equal number of respondents believe the test product is expensive as believe it is too cheap is referred to as the point of marginal cheapness – PMC. The point at which an equal number of respondents believe the test product is too expensive as believe it is cheap is referred to as the point of marginal expensiveness – PME. The point at which an equal number of respondents believe the test product is expensive as believe it is cheap is referred to as the indifference price point – IPP. The point at which an equal number of respondents believe the test product is too expensive as believe it is too cheap is referred to as the optimal price point – OPP. These distributions are usually displayed in a chart, as shown below.
© 2002, 1997 MarketVision Research, All Rights Reserved www.mv-research.com
respondents will not have a good reference price.2
Optimal Price Point Point of Marginal Cheapness Indifference Price Point
Point of Marginal Expensiveness
Inexpensive Too Inexpensive Too Expensive Expensive
To illustrate. we would suggest that an optimal price could only be determined after setting an objective. Moreover. van Westendorp himself made the following statement regarding PSM: A word of caution is in order: price-consciousness of this nature should never be equated with propensity to buy. For almost any product that is not a direct line extension. such as revenue maximization. any such analysis must include an understanding of the cost structure of the product or service. Many pricing researchers question that this is the definitive optimal price for a product.mv-research. “Very early in the development of survey techniques for marketing. The lack of a good reference price. offers the following. or respondents’ use of an inappropriate reference price. share maximization or profit maximization. often causes the underestimation of a product’s ability to command a premium price.com 2
.100% 90% 80% Percent of Respondents 70% 60% 50% 40% 30% 20% 10% 0% $1 $2 $3 $4 $5 $6 $7 $8 Price
In this method. In a large sense.1 Another limitation of this approach to pricing research is that respondents’ ability to answer these questions is dependent upon their having a good reference price. researchers learned that it was futile to ask consumers outright what they would be willing to pay for a product…”
Methods for Pricing Research • MarketVision Research www. PSM becomes a test of price awareness rather than a measure of price sensitivity. the optimal price point for a product is the point at which the same number of respondents indicate that the price is too expensive as those who indicate that the price is too cheap. and not a price. Tom Nagle. 2 Several researchers believe that consumers should be asked to respond to questions with a behavior or the likelihood of a behavior. Another concern among several researchers lies in asking a respondent to provide a specific price in response to a question rather than in providing a likely behavior. in The Strategy and Tactics of Pricing.
While we do not recommend the analysis or interpretation as outlined in PSM. a sample of respondents evaluates this concept at $200.
Each limitation is discussed below. a different sample of respondents evaluates the same concept at $100. the respondent must compare the price presented in the concept to an internal reference price to determine if the price is fair or not. most researchers would suggest that each respondent only evaluate one concept. A demand curve is constructed by evaluating purchase intent at each price. When shopping. Relies on price awareness As in van Westendorp. (After introducing the product concept) How likely. Respondents are presented with a product concept and asked how likely they would be to purchase this product at a specific price. there might be an interest in the market’s willingness to pay for a specific feature or how the inclusion or exclusion of a product characteristic influences purchase likelihood. and another sample of respondents evaluates the concept at $300. However. Therefore to evaluate various product specifications. The standard purchase intent question is shown below. Despite the ubiquitous nature of these questions. When presented with a set of products to select from. The concept test can be used to evaluate these various specifications. respondents may have difficulty answering reliably.mv-research. we have found the questions used in PSM can provide useful diagnostics. researchers commonly encounter four limitations when using this approach for pricing research: • • • • It provides no competitive information. would you be to purchase this product in the next 12 months if it costs $200? Definitely would purchase Probably would purchase Might or might not purchase Probably would not purchase Definitely would not purchase To evaluate price sensitivity using this example. Inefficient to evaluate various product specifications Often. For instance. the total sample size
Methods for Pricing Research • MarketVision Research www. a researcher would like to evaluate a small number of specific product variations at the same time price is being evaluated. Provides no competitive information A concept test asks respondents to evaluate how likely they would be to purchase a specific product without any information about other products that might be available in the market.
Concept Test/Concept Evaluation
The standard purchase intent question from a concept test is also commonly used for pricing research. In the absence of this comparative task. Typically the researcher will expose independent samples of respondents to different prices.com 3
. consumers can make trade-offs between features and price to determine their preferred product. It is inefficient when evaluating numerous product specifications. consumers generally have the chance to see a set of competing products and pick one from the set. It relies on aggregate-level analysis. This determination is based on a respondent’s awareness of the current pricing in the category. It relies on price awareness.
. it is a concern—especially when the
Methods for Pricing Research • MarketVision Research www. much as buyers are forced to do when actually shopping. as follows: WHICH WOULD YOU PREFER? Extremely Clear Picture Quality $300 Strongly Prefer Product on Left 1 2 3 4 5 6 7 or Clear Picture Quality $200 Strongly Prefer Product on Right 8 9
In conjoint analysis. To illustrate. respondents are asked to make hypothetical trade-offs between configured products. The traditional concept test can be effectively used in pricing research when the product features are already determined. Each respondent answers a series of trade-off questions. Each respondent provides enough information through his or her trade-offs that the utility of each product characteristic (including price) can be estimated for each respondent. Relies on aggregate-level analysis Also like van Westendorp.must grow. conjoint analysis presents concepts to respondents. These preferences can be modeled in a market simulator. in each question the combination of features shown together changes. if we wished 200 observations per cell. in conjoint analysis each respondent is exposed to many concepts. respondents are forced to make trade-offs between products and product features. In most research. we now have nine cells and would require 1800 respondents. These utilities also allow prediction of preference for any product that can be defined using the product characteristics in the study. This precludes the measurement or incorporation of interaction terms into the market model without substantial pre-specification. if we have three alternative product variations. Two common drawbacks exist with conjoint analysis. instead of exposing each respondent to a single concept. In this way. This individuallevel estimation allows for the identification of individual differences that can lead to a market segmentation scheme and can be used to help predict acceptance of products by different individuals in a heterogeneous market. this assumption is not likely to be a problem. A demand curve can be produced from these simulations. this approach will make respondent heterogeneity difficult to detect and measure. However. we would require 600 respondents.mv-research. However. a large number of product features can be evaluated.
Like concept tests. a concept test will rely on aggregate. For example. a respondent might be asked to express his preference between two VCR alternatives. with each variation at three prices. or at most subgroup-level analysis. Nearly all conjoint techniques assume a main-effects only design. For each treatment. and we are only testing three prices (three cells). That is. A market simulator allows “what-if” analysis for any configuration of products in any competitive environment. the level of price awareness is high. In pricing research. and the competitive context is such that evaluating a single product is not too limiting.
the respondent is presented with a set of products and the respondent is asked to pick one. In practice. Discrete choice modeling has been used with great success in pricing research. A ratings-based conjoint can be used to develop the utilities for the product
Methods for Pricing Research • MarketVision Research www. such as brand equity research. While conjoint provides a level of realism in that consumers are asked to make trade-offs between product alternatives. Conjoint analysis requires the respondent to state a preference among a set of configured products. One other concern is sometimes raised about conjoint analysis. but it does not ask whether the respondent would like a configured product well enough to buy it.mv-research. simulation capability). but it also includes a more realistic respondent task. If the focus of the research is product design and price/brand. Since discrete choice modeling is generally conducted with full profiles. we suggest that ratings-based conjoint is the best approach for measuring many product features at the product design stage. we recommend using both conjoint and discrete choice. enjoys many of the benefits of conjoint analysis (e. Put another way.
Discrete choice modeling. we recommend a concept test approach. the number of product features studied is limited. and market decline as products become less attractive.design involves price in addition to brand or model since different brands might have different price sensitivities.
MarketVision recommends that discrete choice modeling be used for “price only” or price and brand studies. Also. conjoint analysis will fail to measure market growth as products become more attractive. For instance. competitive products. conjoint analysis is oriented toward share rather than volume. Further. In some cases. neither of these drawbacks is limiting. referred to by some as choice-based conjoint. ability to include a large number of features. and both discrete choice and conjoint allow what-if simulations.g. both approaches are able to produce utilities at the individual level.com 5
. like consumers actually do when shopping. as illustrated below: WHICH VCR WOULD YOU PURCHASE? 1 Brand A 75 Channels Extremely clear picture quality $300 2 Brand C 250 Channels Clear picture quality $200 3 Brand B 150 Channels Somewhat fuzzy picture quality $100 4 NONE If these were my only alternatives I would not purchase anything
The results from discrete choice modeling are very similar to those from conjoint. In discrete choice. the respondent task of providing a rating is still not as realistic as choosing a product..
In instances such as these.com 513. Jon (1994). Beaver Creek.com 6
.mv-research. it is possible to combine the two tasks using either a bridging design or re-weighting.” Paper presented at American Marketing Association Advanced Research Techniques Forum. Methods for Pricing Research • MarketVision Research www.com
The interested reader is referred to Pinnell.features. CO.791.3100 • or visit our website at: www.3
For more information please contact MarketVision Research at: info@mv-research. Discrete choice modeling can then be used to determine how best to price the features with different brands. “Multistage Conjoint Methods to Measure Price Sensitivity.mv-research.