A launch is not a result. It is the moment you start collecting one.
Go back to the success metrics you wrote down before anyone opened a design tool. Answer the question they exist to answer: did this release move them, by how much, against what? Decide the comparison before you look — the same metric over the same span, the weeks before launch against the weeks after — or you will find a story in the noise.
Teams that skip this keep shipping into the dark, and the roadmap becomes a list of opinions sorted by seniority. Post-launch measurement is not a report you file and forget — it is the input to the next decision: what to fix, what to push further, what to take out. If a number cannot change what you do next, you are not measuring. You are decorating.
Product Analytics
Analytics tells you what people did, at scale, without you having to ask them. It is a record of behaviour — the route someone took through your product, and the exact point where they abandoned it — for everyone who ever used it, not the handful you managed to interview.
Instrument events that map to real intent — started checkout, invited a colleague, saved a draft — not every button press. A dashboard with four hundred events and no owner is a landfill with a login screen.
What analytics cannot tell you is why. A funnel showing sixty per cent drop-off at the address form is a fact and a mystery at the same time. Use the numbers to find where the problem lives, then put five people through that screen and watch them.
Quantitative data points at the wound. Qualitative research tells you what caused it.
- Visit the Dedicated Data Analyst Roadmap — roadmap
- What every product manager needs to know about product analytics — article
- Amplitude Guide to Product Analytics — article
- One-Hour Product Analytics Tutorial — video
HEART Framework
HEART gives you five things worth counting, and a name for each: Happiness, Engagement, Adoption, Retention and Task success. Happiness is attitude — ratings, survey scores. Engagement is depth and frequency of use per active user. Adoption is people picking the feature up for the first time. Retention is how many of them are still there weeks later. Task success is completion rate, time on task, error rate.
The point is not to fill in all five. Choose the ones that matter for what you are evaluating and say which you are dropping. A new onboarding flow is judged on task success — do people get through it — and on retention, which tells you whether getting through it meant anything. Engagement measures depth of use over time, which a one-off flow was never going to produce.
For each dimension you keep, write three things: the goal in plain words, the signal — the behaviour that would show it happening or failing — and the metric that counts that signal. Goal: people finish setup without help. Signal: they stop opening the help panel halfway through. Metric: help-panel opens per completed setup. A dimension with no metric is a mood.