Intercepting at a high-score moment systematically inflates the number

What it is

Placing a recommend or satisfaction survey right after payment succeeds, after a “done” animation, or after a good outcome is a peak-moment intercept. Affect is at a high, people on failing paths have already left, and both the sample and the mood lean toward high scores. The number rises because the measurement window was opened on the slice of the journey that is easiest to rate well, not because the whole experience improved. This is not the same problem as writing the metric into a bonus or coaching people to give nine or higher: the damage here is the sampling moment, not the incentive.

Why it happens

Evaluation leans on recency and current affect. Success copy, a confirmation page, and a celebration motion pin recency to the positive; people who did not finish never reach that page, so the survey never reaches them. Dissatisfaction in the population is structurally deleted, and those who remain sit in a brief high. Rules such as “orders completed only” or “sessions longer than n seconds” tighten the filter again. If the team then treats this lifted curve as baseline, moving the survey earlier or at random later looks like the product got worse. Choosing a high-score moment therefore both raises the level and locks in a measurement that cannot be compared with other touchpoints.

Studying it

Field the same item at different touchpoints with the same exclusion rules, and compare distributions rather than means only: success page, mid-task, exit, random session. Compute coverage at each point (sent, seen, completed) and the behavioral endings of people who were never sent the survey. If the success-page score is high, the exit intercept is low, and coverage leans toward success, the lift is from moment choice. Randomly assigning intercept time estimates the moment’s own effect so it can be taken out of a product effect. Freeze the trigger rule before any trend analysis; mark scores inside a rule-change window as their own series.

Where it stops holding

An immediate post-task rating in a usability session is a legitimate near-time measure if failures are asked as well as successes—not successes only. A transactional touchpoint after every delivery that covers completed and failed endings reduces the lift and can still miss people who never ordered. Parking the survey on a neutral page (settings, billing) lowers peak bias and trades it for “who walks into settings.” When a rule requires collecting feedback on a particular success page, collect it there, but do not treat that page’s score as the global experience.

Applying it

• Write the trigger as random-by-session or random-by-user; do not default it to payment success or a celebration motion.
• If the business must collect on the success page, keep a parallel random or exit intercept and report the two scores separately; do not average them into one official number.
• Log trigger changes; do not write the following weeks into the long trend.
• Check: among users who never saw the survey, how much higher are failure and churn than in the answered sample? If clearly higher, the success-page score does not stand for everyone.

Related

• Same group: Q3.17.1 Cultures occupy the same scale differently · Q3.17.2 An opaque industry benchmark is not a comparison · Q3.17.3 These scores are lagged attitude snapshots, not causes
• Adjacent: Q3.03 Net Promoter Score · Q1.04 Sampling and representativeness
• Search terms: intercept timing · peak-end · sampling bias