Research Note: Product Launch Postmortem

Research Note: Product Launch Postmortem

Question

What should a launch team preserve, compare, and change after the event?

What the sources establish

Google SRE's postmortem culture guidance defines a postmortem as a written record of an incident, its impact, response, causes, and follow-up actions. It emphasizes blameless analysis and effective preventive work. The SRE Workbook adds examples and stresses that a postmortem creates value when it drives action and shared learning.

A product launch is not necessarily an incident. The reusable principle is a factual record, separation of contributing conditions, blameless language, assigned actions, owners, and follow-up.

E079 demonstrates why one metric is insufficient. A launch can attract attention while frustrating a segment. A viewer can dislike the show and still want the product. A strong product can survive weak communication. A clear launch can precede poor product experience.

Venture Step framework

The postmortem should preserve the pre-launch objective, audience, claims, message hierarchy, run of show, demonstration modes, speaker assignments, distribution plan, and expected next actions before reviewing results.

The result should be separated into reach, attention, comprehension, product evaluation, trust, action, product experience, and commercial outcome. Each layer needs its own evidence and limitation.

Views show exposure, subject to platform definitions. Completion shows viewing behavior, not comprehension. Surveys can assess recall or reaction but inherit sampling and question-design limits. Clicks and preorders show action but not whether the launch caused it. Returns, activation, support, retention, and customer research can test whether the product later delivered the promise.

Disagreement and uncertainty

Attribution between a launch and later sales is rarely complete. Product availability, price, reviews, retail placement, competitor actions, promotions, seasonality, and prior brand demand can affect the same outcome.

Sentiment is sensitive to platform, sample, timing, and coding. A self-selected poll should remain an audience reaction, not a market estimate.

Editorial use

The operating guide should produce a launch postmortem with evidence, decisions, owners, dates, and a later effectiveness review. It should not use "blameless" to avoid accountability for repeated or intentional decisions.