# Event registration landing-page experiment — editable run card

**Event / edition:** [ ]  **Page URL and owner:** [ ]  **Experiment ID:** [ ]  **Decision owner:** [ ]

**Attendee task:** [For example, find the right ticket and finish registration.]

**Observed friction and evidence:** [Support questions, usability observation, analytics path or other evidence.]

**Hypothesis:** If we change [one defined page element] from [control] to [variant], then [defined attendee action] may change because [reason].

| Design | Record before launch |
|---|---|
| Control and variant | Exact copy/design/version link for each; what else stays the same |
| Assignment and audience | How visitors are assigned; who is excluded; how repeat visitors stay consistent |
| Primary outcome | A visitor-level event such as completed registration, if observable across the checkout boundary |
| Diagnostic events | CTA click, checkout start, form error or support request; do not confuse these with completion |
| Guardrails | Accessibility, price and policy accuracy, page performance, error rate, attendee support |
| Data boundary | What the organizer can observe and what the hosted ticketing provider reports separately |
| Feasibility and decision rule | Traffic forecast, planned sample/evaluation method, end condition and inconclusive outcome |
| SEO handling | If variants have separate URLs: canonical/temporary redirect plan and removal after test |
| Launch QA | Both versions: mobile, keyboard, labels, errors, language, date/time, ticket price, link and checkout handoff |

**Launch time and time zone:** [ ]  **Pause/rollback owner:** [ ]  **Evidence links:** [ ]

## Decision record

| Date | Visitors by variant | Completed registrations by variant | Guardrail observations | Interpretation | Decision and owner |
|---|---|---|---|---|---|
| [ ] | [ ] | [ ] | [ ] | [winner / inconclusive / stop for harm, using predeclared method] | [ ] |

Do not invent a winner from a handful of purchases or stop merely because a dashboard briefly shows a lead. If expected traffic is too small for a useful experiment before the event date, run moderated attendee task checks and fix verified barriers instead. A click on an external checkout is not a completed purchase; confirm the provider's reporting boundary. Do not show search crawlers a special version hidden from people. Use your own analytics, consent and accessibility requirements.