Build & Launch Checklist

How to use this

Start only after Design has handed over one approved version. Work top to bottom through build, QA, launch, technical monitoring, and the Evaluation handoff. Progress is saved in this browser on this device and is not shared.

Use this checklist to implement the approved experience, launch it safely, monitor technical health, and hand a complete launch record to Evaluation & Decision.

Reset for a new campaign

1. 1. Accept the handoff

   Review the approved design, hypothesis, surfaces, behavior, assets, tracking, acceptance criteria, audience, traffic split, and timing Confirm access, tools, theme, experiment platform, and dependencies Resolve gaps with Design or Data before implementation Confirm the control and rollback path

   Required output: A build-ready handoff with owners and dependencies confirmed.

2. 2. Build the approved experience

   Implement the approved mobile and desktop behavior Preserve the control experience for experiments Keep visitors mutually exclusive where variants or tests could overlap Implement the approved audience and eligibility rules Add the required analytics and experiment events Avoid unrelated production changes inside the campaign build Protect page stability, loading performance, and image dimensions

   Required output: A reviewable implementation that matches the approved design and experiment plan.

3. 3. Complete visual and content QA

   Compare the implementation with the approved design at relevant mobile and desktop widths Verify final copy, links, imagery, pricing, promotion language, and claims Verify responsive behavior and every required interface state Verify labels, contrast, focus, and keyboard path Confirm there is no unexpected layout shift or visual regression

   Required output: Visual and content acceptance criteria pass on the agreed devices and browsers.

4. 4. Complete functional commerce QA

   Verify product selection and variant behavior Verify add-to-cart behavior Verify cart and checkout paths Verify discounts, bundles, subscriptions, upsells, and relevant app integrations Verify sold-out, unavailable, validation, and error states Check for script conflicts and console errors Verify the rollback path

   Required output: The full affected customer journey works without blocking commerce.

5. 5. Verify tracking and experiment setup

   Confirm primary, secondary, and guardrail events fire once with the correct payload Confirm control and variant assignment persists correctly Confirm traffic split, audience rules, exclusions, and mutually exclusive bucketing Confirm experiment name, hypothesis, primary metric, start time, and planned end date match the approved plan Exclude internal or QA traffic where the platform allows it Record screenshots or evidence of the final configuration

   Required output: Data can trust exposure, allocation, and outcome measurement before traffic is admitted.

6. 6. Secure launch approval

   Obtain final Design approval for fidelity Obtain final Data approval for tracking and experiment configuration Obtain client or business approval where required Confirm timing does not conflict with an unplanned promotion, release, stock issue, or site event Confirm the person responsible for launch and rollback

   Required output: Every required owner has approved the same release candidate.

7. 7. Launch and verify live

   Record the actual launch date and time Verify the control and variant on the live site using eligible clean sessions Verify analytics and experiment exposure after launch Verify the affected customer journey again in production Confirm the live traffic allocation matches the plan Record the release, theme, experiment, and page links

   Required output: A verified live campaign and a reproducible launch record.

8. 8. Monitor days 1 to 3

   Monitor errors, broken journeys, allocation anomalies, tracking gaps, and severe guardrail issues Log promotions, traffic spikes, outages, campaign changes, stock issues, and other external events Fix or roll back critical implementation failures through the agreed rollback path Do not call a result early or stop for ordinary performance fluctuation Keep the fixed end date unless data quality or customer harm requires intervention

   Required output: Early technical health is confirmed and every anomaly is documented without premature winner language.

9. 9. Hand off to Evaluation

   Provide launch time, live URLs, experiment link, release reference, design version, tracking evidence, QA record, caveats, anomalies, and external events Record every implementation change made after launch Confirm Evaluation has the dates, configuration, exposure context, implementation history, and caveats needed for a valid read Confirm the Build & Launch handoff to Evaluation & Decision is complete

   Handoff gate: Evaluation & Decision can evaluate the campaign without reconstructing what launched or when.

Continue to Evaluation & Decision