Problem This Solves
Pricing decisions require quantitative estimates of how customers respond to price changes, but firms often rely on gut instinct or flawed methods (e.g., asking customers directly what they would pay). Chapter 8 provides a systematic framework for selecting, executing, and interpreting price sensitivity measurement techniques -- matching method to product stage, balancing cost against accuracy, and integrating managerial judgment with empirical data.
Key Principle
Measurement techniques are classified along two dimensions: (1) what is measured -- actual purchases vs. preferences/intentions, and (2) conditions of measurement -- uncontrolled observation vs. experimentally controlled. No single technique is universally best. Technique selection depends on the product's development stage, and even precise numerical estimates are not necessarily accurate. Managerial judgment and empirical measurement are complements, not substitutes.
Exhibit 8-1 Classification Framework:
| Variable Measured | Uncontrolled | Experimentally Controlled |
|---|---|---|
| Actual Purchases | Historical sales data; Panel data; Store scanner data | In-store experiments; Laboratory purchase experiments |
| Preferences/Intentions | Direct questioning; Buy-response survey; In-depth interview | Simulated purchase experiments; Trade-off (conjoint) analysis |
Technique Selection by Product Stage:
| Product Stage | Recommended Technique |
|---|---|
| Concept / Prototype | Conjoint analysis; preference/intention surveys |
| Fully developed, frequently purchased, low-cost | In-store experiments or sophisticated laboratory purchase experiments |
| Fully developed, high-cost durable | Simple lab experiment or simulated purchase survey; buy-response for acceptable price range |
| Established on market | Historical sales data analysis |
Good Examples
- E-book pricing experiment: An online laboratory experiment with 2,000 respondents was designed and launched in 1.5 weeks at one-tenth the cost of physical experiments. It revealed demand was inelastic below $9.99 but very elastic above $9.99 for widely available titles, while new titles not available elsewhere supported prices above $9.99. The retailer launched a segmented pricing model based on these results.
- Ski technology conjoint case (Blue Sky): Aggregate conjoint data showed revenue maximized at $450 with an unviable margin. Slicing by segment revealed an "innovator" segment (35-50-year-old men, former skilled skiers with aging knees) that profitably supported $800. Extending the warranty from 90 days to 1 year more than doubled the take rate.
- Tractor manufacturer competitive database: A B2B company built a competitive bid database for approximately $50,000 by having sales forces log competitor bid information -- a small sum for a multi-billion dollar company. This enabled segment-level price sensitivity estimation.
- Motel 6 natural experiment: Electronic billboard pricing changed by the hour at nearly no cost, enabling study of price responsiveness by location, day of week, and time of day.
Bad Examples
- Direct questioning: Asking "What is the most you would be willing to pay?" elicits bargaining behavior (understating) or desire to please (overstating). The authors state this "should never be accepted as a valid methodology."
- Treating buy-response data as directly predictive: Buy-response surveys yield plausible answers but cannot be treated as directly comparable to actual in-store sales because consumers' answers depend on recalled competitor prices.
- Ignoring aggregation problems: Data aggregated across stores and time periods conceals individual price differences and produces elasticity estimates that "may, on average, be correct, but do not really apply to any single store setting."
- Blind reliance on conjoint without judgment: Conjoint studies of physician prescribing "invariably predict much higher price sensitivity among physicians than, in fact, is revealed by prescribing behavior." Side-by-side comparisons overstate price focus.
- Running regressions with insufficient price variation: "If there has been little historical variation in a product's price, then no statistical technique applied to its sales data can reveal the effect of price changes."
Key Quotes
- "Accuracy is a virtue in formulating pricing strategy; precision is only a convenience."
- "The measurement of price sensitivity is not an end result but a catalyst to learn more about one's buyers."
- "No estimation technique can capture the full richness of the factors that enter a purchase decision. In fact, measurements of price sensitivity are precise specifically because they exclude all the factors that are not conveniently measurable."
- "Numerical estimation of price sensitivity is no shortcut to knowing a product's buyers -- who they are, how they buy, and why they make their purchase decisions."
- "Uncontrolled direct questioning as a research technique to estimate price sensitivity should never be accepted as a valid methodology."
- "Of all the methods used to estimate price sensitivity from preferences or intentions, trade-off analysis promises the most useful information for strategy formulation."
- "Integrating soft managerial judgments about buyers and purchase behavior with numerical estimates based on hard data is fundamental to successful pricing."
Rules of Thumb
- Never use direct questioning ("What would you pay?") as a standalone method -- it produces useless or misleading results.
- Match technique to product stage: conjoint for concepts, experiments for established low-cost goods, historical data for mature products with price variation history.
- 10-12 in-depth interviews per segment uncover approximately 90% of key customer needs; 40-60 interviews cover a full market of 4-6 segments.
- Always segment conjoint data before concluding a product is unviable -- aggregate results can mask profitable niches.
- Calibrate surveys against experiments: Run a few in-store experiments alongside simulated purchase surveys; if bias is stable, use the cheaper survey method adjusted by the measured bias.
- Design sales data for research value: Make some price changes independently of other marketing variables so researchers can isolate price effects. Log unusual events that distort data.
- Begin with qualitative research (ethnographic interviews, focus groups, in-depth interviews) before quantitative measurement to understand who buys, why, and how.
- In B2B, use evocative anchoring -- ask respondents what budget items they would trade off to obtain the value your product provides, rather than asking willingness-to-pay directly.
- Panel data bias: Fewer than 5% of invited households accept traditional panels, and recording purchases makes members more price-aware. Prefer scanner-based panels (70%+ participation) to reduce both biases.
- Online lab experiments can be completed in as little as one week at one-tenth the cost of physical experiments, with much larger sample sizes.
- Always ask "why" after any measurement result -- compare against prior expectations and investigate inconsistencies from both the judgment side and the measurement side.