01 / Declare the objective
Measure the result you actually value
“Optimize revenue” is incomplete without a definition of revenue and the constraints surrounding it. Gross ticket value, retained receipts after refunds and fees, contribution after incremental costs, and total event income can move differently. Choose a primary outcome that the finance team can reconcile. State whether the analysis covers only the event or also considers membership value and future purchasing behaviour.
Write secondary outcomes before the test starts. Attendance, affordability, group access, complaints, refunds and operational workload may matter even if net receipts rise. Some belong as monitoring measures; others should be firm stopping conditions. Do not combine every concern into an opaque score that makes it impossible to see which customer or operational outcome has deteriorated.
02 / Separate price from selection
Do not read elasticity from a simple correlation
Popular fixtures often carry higher prices and sell more tickets. That does not mean increasing a price causes more demand. Staff may have chosen the price because they already expected a strong event. Likewise, late discounts can appear associated with weak sales because the discount was introduced in response to weak sales. Both patterns confound a naive historical comparison.
A demand model can help organize relevant context, but a counterfactual still needs justification. Elasticity estimates depend on meaningful price variation, product comparability and assumptions about unobserved influences. Report those assumptions alongside the estimate. If a model is asked to evaluate prices well outside the range seen in training, label the exercise as extrapolation rather than presenting an unsupported optimum.
03 / Choose a credible comparison
Design around substitution and fairness
A randomized experiment can offer stronger evidence when it is lawful, fair and operationally appropriate. However, ticket products are connected: customers can switch sections, dates or channels. Randomizing adjacent seats may create a confusing experience and contaminate the comparison. Decide whether an event, sales window or product group is a defensible unit, and check whether the available sample can answer the question.
Where randomization is unsuitable, a matched-event or other quasi-experimental design may be useful, but it relies on additional assumptions. A change in opponent quality, weather, marketing or available inventory can undermine the comparison. Record concurrent interventions, inspect pre-test patterns and explain what uncertainty remains. A before-and-after chart alone should not be described as proof of a pricing effect.
| Decision | Required statement |
|---|---|
| Offer under test | Eligible products, visible total price and selling window |
| Comparison | What would otherwise happen, and why it is comparable |
| Primary outcome | Reconciled net measure and observation window |
| Protections | Approved bounds, access commitments and complaint escalation |
| Stop / rollback | Trigger, authorized owner and reversal procedure |
04 / Bound the commercial policy
Approve the rules before the recommendation
Document permitted prices, frequency of change, eligible products and exclusions. Distinguish general dynamic pricing from personalized offers based on customer characteristics. The latter can raise different privacy, fairness and consumer-law questions. Avoid using sensitive traits or unreliable proxies to infer what an individual can be charged. Specialist legal review should consider the actual implementation and jurisdiction.
Price transparency is part of the test, not an afterthought. Define how fees, eligibility and offer duration are communicated. Review obligations under relevant consumer rules, including the EU Unfair Commercial Practices Directive where applicable, without assuming one rulebook covers every event or customer. The venue should also protect existing entitlements and explain material differences between otherwise similar products.
05 / Read uncertainty honestly
Do not stop at the first favourable result
Agree the observation window and analysis method in advance. Repeatedly checking results and stopping when a positive difference appears can exaggerate evidence unless the method explicitly accounts for sequential monitoring. Small samples may leave a wide range of plausible effects. “Not statistically distinguishable” does not mean two policies are commercially equivalent, and a positive point estimate is not a guarantee.
Inspect displacement as well as additional sales. A discounted product may attract buyers who would otherwise have purchased a higher-priced seat. A premium offer may shift bookings rather than increase total receipts. Include refunds, chargebacks and customer-service consequences where relevant, and avoid attributing unrelated ancillary spending to the price test without a defensible measurement design.
06 / Make a bounded decision
Record what the test supports
Close with a recommendation that matches the evidence: continue, revise, stop or investigate further. State the event types and offer conditions to which the result may reasonably apply. A successful trial for one competition does not automatically justify the same policy for a different audience, venue configuration or season.
Save the protocol, results, approvals and customer-facing wording. Keep rollback practical rather than theoretical: identify who can suspend changes, how inventory is reconciled and how affected customers are supported. The purpose of testing is to reduce uncertainty around an accountable decision, not to supply a scientific-looking rationale for a change already chosen.
Limitation: this is not a pricing recommendation
No prices or commercial returns are prescribed here. A venue needs its own evidence, legal review, financial definitions and customer protections before implementing a policy.