Pricing evidence: design a test that measures profitable demand
Produced: 2026-09-20
Language: en
IDER in-house training materials: practical cases, illustrations and guided exercises.
All cases and figures are hypothetical teaching examples, not current market data.

[00:00:00.000 – 00:00:49.235]
Test profitable demand, not a favourite price
ON SCREEN
Hypothetical two-price experiment
Customer behaviour plus economics
A decision under uncertainty

TEACHING COMMENTARY
A pricing study often begins with the question, what will customers pay? For a commercial decision, that is incomplete. We need to know how a feasible price changes purchases, costs and the contribution generated from the customers we can reach. This clinic uses a hypothetical experiment with two prices and the same product. Every number is illustrative, not a claim about any current market. You will learn why a clear reduction in demand can coexist with uncertain evidence about profit. Keep a calculator or spreadsheet available and work through both pauses. Our objective is a pricing recommendation that explains its estimand, its experimental design and its limits, so another practitioner can reproduce and challenge the reasoning.

[00:00:49.235 – 00:01:42.571]
Define the outcome before collecting data
ON SCREEN
Unit: eligible customer offer
Primary outcome: contribution per offer
Fixed eligibility and observation window

TEACHING COMMENTARY
Choose the quantity you want to estimate before choosing a questionnaire. Here it is average contribution per eligible offer, including zero for customers who do not purchase. That denominator matters: looking only at buyers ignores the customers a higher price drives away. Define eligibility, the product, delivery terms and the time allowed to purchase. Contribution is realised revenue less costs that change because the purchase occurs, with refunds and payment fees treated consistently. If each customer can buy multiple units, measure the full customer outcome rather than forcing the data into a one-purchase model. State whether acquisition costs differ by price and whether they belong in the decision. An explicit outcome definition prevents attractive but irrelevant statistics from replacing the commercial objective.

[00:01:42.571 – 00:02:35.769]
Use interviews to explain behaviour
ON SCREEN
Purchase context and alternatives
Budget, approval and switching friction
Hypotheses to test in observed behaviour

TEACHING COMMENTARY
Before randomising prices, investigate how customers make the decision. Ask them to describe the most recent comparable purchase, who approved it, which alternative they considered and what delayed payment. Concrete episodes reveal constraints more reliably than asking whether a product sounds useful. A firm may value the service but lack procurement approval; a household may prefer it but have cash only after payday. These mechanisms guide the offer design and the timing of observation. Interview findings also suggest prespecified segments, such as existing users versus new customers. They do not establish each segment's population size or causal price response. Use qualitative evidence to generate and interpret hypotheses, while preserving a clear distinction between explanation and measurement of effect.

[00:02:35.769 – 00:03:32.357]
Separate stated willingness from purchase evidence
ON SCREEN
Survey intention: hypothetical response
Binding offer: observed choice
Repeat use: durability of demand

TEACHING COMMENTARY
A willingness-to-pay question can help explore a plausible price range, but the answer carries no necessary financial consequence. Customers may please the interviewer, anchor on the first suggested number or ignore budget constraints. An actual offer creates a different decision. Make the product, delivery conditions, payment obligations and any refund policy clear, and follow applicable ethical and commercial requirements. Record the offer even when the customer declines, so the denominator remains complete. A purchase still does not establish long-term value: discounts, novelty or temporary urgency may drive one transaction. Where repeat business matters, define a follow-up window and track retention and returns. The research design should match the decision horizon, rather than treating initial conversion as lifetime demand.

[00:03:32.357 – 00:04:31.591]
Randomise at the unit that receives the price
ON SCREEN
Comparable offers apart from price
Stable assignment and complete records
Account for repeat users and spillovers

TEACHING COMMENTARY
Suppose one thousand eligible customers are randomly assigned to two prices, with five hundred in each arm. Keep product quality, marketing copy and fulfilment conditions comparable. Save the allocation procedure and assigned price before outcomes arrive. If one customer visits repeatedly, preserve that customer's assignment; treating each visit as independent can mix exposure and understate uncertainty. If customers share prices or purchase for each other, consider a suitable group assignment and analyse at that level. Random assignment supports causal interpretation because the offer groups are comparable in expectation, not because every characteristic will match exactly. Monitor missing records, price overrides and stockouts. Report assignment-based results so excluding inconvenient customers after randomisation does not undo the protection provided by the design.

[00:04:31.591 – 00:05:23.338]
Calculate contribution for every offer
ON SCREEN
GHS50: 200 purchases from 500 offers
GHS60: 150 purchases from 500 offers
Unit variable cost: GHS30 in both arms
(50 − 30) × 0.40 = 8 ; (60 − 30) × 0.30 = 9
TEACHING COMMENTARY
At fifty cedis, two hundred of five hundred customers purchase, giving forty percent conversion. At sixty cedis, one hundred and fifty purchase, giving thirty percent. With a variable cost of thirty cedis, the lower price contributes twenty per purchase, while the higher contributes thirty. Multiply each margin by its conversion rate: average contribution per offer is eight cedis at fifty and nine at sixty. The higher price has fewer buyers but a larger estimated contribution per reachable customer. Across five hundred offers, those contributions total four thousand and four thousand five hundred cedis. These are sample estimates before any differing fixed or acquisition costs. We have not yet established that the higher price produces a reliable improvement beyond this experiment.

[00:05:23.338 – 00:06:13.969]
Which price wins on which outcome?
ON SCREEN
Compute revenue per eligible offer
Compute contribution per eligible offer
Explain why the rankings differ

TEACHING COMMENTARY
Pause the video and calculate two outcomes for each price. First divide total sales revenue by the five hundred eligible offers. Then calculate contribution per offer using the thirty-cedi variable cost. Identify which price wins under each measure and write a sentence explaining why. Do not compare the thirty-cedi margin at the higher price with the twenty-cedi margin alone; customers who decline still consume an opportunity to sell. Equally, do not choose the lower price solely because it converts more people. The intended decision outcome determines the ranking. Assume one unit per buyer, no returns and equal offer costs for this exercise. Keep those assumptions beside your calculation, because changing them could alter the commercial recommendation.

[00:06:13.969 – 00:07:05.820]
Revenue and contribution can rank prices differently
ON SCREEN
Revenue per offer: GHS20 versus GHS18
Contribution per offer: GHS8 versus GHS9
Estimated gain: GHS1 per offer

TEACHING COMMENTARY
The fifty-cedi price generates ten thousand cedis revenue from five hundred offers, or twenty per offer. The sixty-cedi price generates nine thousand, or eighteen per offer. Revenue therefore favours the lower price. Contribution favours the higher price: eight versus nine cedis per offer, a difference of one cedi. Lower revenue can coexist with higher contribution because the business avoids variable costs on the purchases it loses. This distinction is essential when managers reward sales volume while the organisation needs cash generation. However, a contribution estimate is not a complete policy decision. We must examine sampling uncertainty, repeat behaviour, operational effects and any strategic constraints. The result tells us what happened in this sample under these stated conditions.

[00:07:05.820 – 00:08:02.651]
Estimate local elasticity with a clear convention
ON SCREEN
Quantity: 200 → 150; price: 50 → 60
Midpoint changes: −28.57% and +18.18%
Arc elasticity: approximately −1.57
(−50 ÷ 175) ÷ (10 ÷ 55) = −1.5714
TEACHING COMMENTARY
Elasticity describes proportional demand response relative to proportional price change. Using the midpoint convention, divide the quantity change of minus fifty by the average quantity of one hundred and seventy-five. That is minus twenty-eight point five seven percent. Divide the price increase of ten by the average price of fifty-five, giving eighteen point one eight percent. Their ratio is approximately minus one point five seven. The negative sign indicates declining demand as price rises. This is an arc across the tested prices, not proof that the same elasticity applies at every price. It also assumes comparable numbers of offers; otherwise use purchase rates. Do not extrapolate a straight demand curve far beyond the experiment without evidence about customer substitution and market conditions.

[00:08:02.651 – 00:08:54.293]
Demand certainty is not profit certainty
ON SCREEN
Conversion difference: −10 percentage points
Approximate 95% interval: −15.88 to −4.12
Independent customers; normal approximation

TEACHING COMMENTARY
We first quantify uncertainty about conversion. The higher-price purchase rate minus the lower-price rate is minus ten percentage points. For two independent samples of five hundred, add the estimated binomial variances: thirty percent times seventy percent over five hundred, plus forty percent times sixty percent over five hundred. The square root is three percentage points. Multiplying by one point nine six gives an approximate ninety-five percent margin of five point eight eight points. The interval excludes zero, so the data provide evidence of lower conversion at the higher price under this design. This does not yet establish a contribution gain. That outcome weights purchases by different margins and has a different variance, which we must calculate separately.

[00:08:54.293 – 00:09:51.875]
Power the test for the decision outcome
ON SCREEN
Set the minimum worthwhile contribution gain
Estimate variation in contribution per offer
Plan sample size and stopping in advance

TEACHING COMMENTARY
Planning sample size around conversion can leave a profitability question unresolved. Decide the smallest contribution improvement that would justify changing prices, then estimate the variability of contribution per offer. In our simplified binary-purchase case, the lower-price outcome is either twenty cedis or zero, while the higher-price outcome is thirty or zero. Their variability differs even with similar conversion uncertainty. Use pilot data or conservative assumptions to calculate power for the chosen outcome, including clustering when assignment occurs by group. Set an observation horizon and analysis plan before looking repeatedly at significance. Repeated informal checking can create misleading confidence. The practical question is whether the proposed test can distinguish a decision-relevant improvement within the available customer flow and budget.

[00:09:51.875 – 00:10:45.491]
Protect validity beyond the primary average
ON SCREEN
Prespecified segments and complete denominators
Refunds, repeat use and service performance
Stockouts, interference and changing competition

TEACHING COMMENTARY
A primary outcome does not remove the need for guardrails. A higher price could improve immediate contribution while increasing complaints, refunds or churn. Define those measures and acceptable limits before the test. Segment analysis should follow a mechanism identified in advance, not a search through dozens of groups until one looks favourable. Report uncertainty for each subgroup and consider the extra risk of false discoveries. Check whether stockouts, delivery delays or competing promotions affected one arm differently. If the market changes during the experiment, preserve concurrent comparisons rather than comparing one month's low price with another month's high price. Finally, distinguish a causal effect in the tested customer population from a claim about every channel, season or geography.

[00:10:45.491 – 00:11:38.131]
Does the evidence establish a contribution gain?
ON SCREEN
Contribution estimate: +GHS1 per offer
Estimated variance of difference: 0.57
Calculate a 95% interval and recommendation

TEACHING COMMENTARY
Pause and evaluate the contribution result directly. In this simplified experiment, the estimated variance of the difference in mean contribution is zero point five seven cedis squared. Take its square root to obtain the standard error. Multiply by one point nine six and place that margin around the estimated gain of one cedi per offer. Does the interval exclude zero? Then write a recommendation that acknowledges both the clear conversion decline and the uncertainty about contribution. Assume independent customers, a single prespecified comparison, equal acquisition costs and complete purchase records. Do not treat a statistically clear demand reduction as proof of commercial success or failure; the contribution question has its own outcome definition and uncertainty calculation.

[00:11:38.131 – 00:12:29.378]
A plausible gain remains uncertain
ON SCREEN
Standard error: approximately GHS0.755
95% interval: −GHS0.48 to +GHS2.48
Profit improvement is not established
1 ± 1.96 × √0.57 ≈ [−0.48, 2.48]
TEACHING COMMENTARY
The square root of zero point five seven is approximately zero point seven five five cedis. The ninety-five percent margin is about one point four eight, giving an interval from minus zero point four eight to plus two point four eight cedis per offer. It includes zero, so this experiment does not establish a positive contribution effect at that confidence level. It also does not establish that the prices are equivalent. The data remain compatible with a modest loss or a meaningful gain. A defensible response is to assess whether further testing has value, considering customer flow, switching costs and risk tolerance. If management acts now, describe the decision as one made under uncertainty rather than declaring a proven profit improvement.

[00:12:29.378 – 00:13:24.944]
Connect evidence to an implementable price decision
ON SCREEN
Outcome, effect and uncertainty
Commercial constraints and follow-up
Owner, review date and reversal trigger

TEACHING COMMENTARY
Prepare a pricing memorandum that begins with the decision, the population and the primary outcome. Show conversion, revenue and contribution together so the reader can see why their rankings differ. Include uncertainty around the contribution effect and disclose the assumptions behind it. Then address repeat use, customer fairness, channel agreements, capacity and any cost that changes at scale. Name the person responsible for implementation and define a review date with a reversal trigger. If prices are personalised or regulated, obtain the relevant commercial and legal review before deployment. A well-designed experiment answers a defined causal question; it does not automatically resolve every policy constraint. Preserve the allocation data and reproducible calculations so later analysts can distinguish market change from measurement error.

[00:13:24.944 – 00:14:22.858]
Build a test your organisation can learn from
ON SCREEN
Audit the denominator in an existing report
Recompute contribution and uncertainty
Specify one decision-ready next experiment

TEACHING COMMENTARY
Choose an existing pricing report and check its denominator. Does it describe buyers only, visits, unique eligible customers, or all assigned offers? Recompute contribution using a consistent unit and include customers who bought nothing. Check whether the uncertainty calculation respects repeat customers, grouped assignment and the actual outcome distribution. Then propose one feasible experiment with a primary metric, minimum worthwhile effect, sample plan and follow-up window. Junior practitioners should be able to rebuild each calculation from raw records. Senior practitioners should be able to explain why the design answers the commercial question and what uncertainty remains. Complete the scored questions and applied tasks below. Your aim is to make pricing evidence actionable without confusing a favourable point estimate with a reliable improvement.

METHOD REFERENCES
J-PAL — Randomization
https://www.povertyactionlab.org/resource/randomization
Methodological basis for random assignment and choosing an assignment unit. Original IDER hypothetical price experiment and arithmetic. / Fondements de la randomisation ; expérience et calculs hypothétiques originaux IDER.

NIST — Confidence limits for a proportion
https://www.itl.nist.gov/div898/handbook/prc/section2/prc241.htm
Background on uncertainty for proportions. The two independent-arm normal approximations and contribution calculations are explicitly derived in this clinic, not presented as NIST Wilson intervals. / Repères sur les proportions ; calculs à deux groupes explicités dans la formation.

Audio credits
https://iderafrica.com/downloads/briefings/audio-credits.txt