Growth Loops · Session 02 · Dates TBD
Find activation and remove one barrier to first value
Ship AIPhoenix, ArizonaPress → to begin
Completing onboarding is not the same as receiving value.
Activation is an early behavior worth caring about because it is plausibly connected to a later customer outcome.
The journey is instrumented. Tonight uses that trustworthy baseline to find first value; session three asks whether customers have a reason to repeat it.
Act one
An activation event should happen early enough to change and connect to an outcome far enough away that the signal matters.
Easy to count
Worth testing
The right column is still a set of hypotheses. Test whether each behavior has coverage, happens early, and separates later outcomes.
| Test | Question | |
|---|---|---|
| Outcome link | Compare later retention | Do performers return more often? |
| Coverage | Share of eligible new accounts | Can most good-fit customers reach it? |
| Timing | Median and distribution | Does it happen early enough to influence? |
| Specificity | Compare non-performers | Does it separate, or merely reflect heavy users? |
| Actionability | Map product levers | Can the team help without gaming it? |
Association is a filter for candidates, not proof that forcing the event will create retention.
Exercise: test candidate moments
compare first-7-day behaviors → week-4 return
candidate A: common, weak separationcandidate B: strong separation, low coveragecandidate C: useful share, plausible + reachableactivation v1 selected; caveat recorded
Create cohorts for each early behavior and compare the same later outcome. Inspect sample size, eligibility, and segment mix before choosing the first working definition.
Use enough history to cover the natural return interval. If the product is young, keep the definition explicitly provisional.
Act two
The funnel shows where customers stop. Elapsed time and session boundaries show how difficult the journey actually feels.
Illustrative values only. Build the baseline from eligible customers and show the elapsed time between stages, not just completion percentages.
A two-day activation path can be healthy when value naturally requires collaboration. Diagnose which clock owns the delay before trying to remove it.
Remove or repair
May create value
Reducing clicks can reduce activation when the removed step supplied the context or input that made the outcome useful.
Act three
Change the highest-leverage step you can plausibly improve without rebuilding the entire onboarding experience.
| Product move | Use when | |
|---|---|---|
| Better default | Preselect a common good path | Choice cost is blocking progress |
| Example state | Show a useful completed result | The blank product hides the payoff |
| Progressive setup | Ask for context when it becomes useful | Front-loaded setup delays all feedback |
| Recovery | Explain and preserve work after failure | Errors restart the journey |
| Human handoff | Route the right teammate into the task | Activation requires another role |
Ship the smallest change that tests the diagnosis. Do not replace the journey before learning which part mattered.
| Definition | Illustrative contract | |
|---|---|---|
| Hypothesis | Cause and expected effect | Example state helps eligible teams reach first share |
| Primary | One decision metric | Activation within 7 days |
| Guardrail | A cost we will not trade away | Error rate and permission reversals |
| Audience | Who can be affected | New eligible workspaces only |
| Window | When the result can be read | One full activation + return cycle |
| Decision | Ship, revise, or stop rule | Written before the first comparison |
Do not change the activation definition after seeing which version won. Version the definition separately from the experiment.
Live build: change one step and verify the measurement
baseline → ship change → replay events
eligible audience rule verifiedactivation event unchangedguardrail events firingwrote 02-activation/experiment-readout.md
Ship the intervention, complete the journey in both expected and failure states, and confirm the exact event stream still matches the activation contract.
The first result tonight is not statistical certainty. It is a clean baseline, a bounded change, and a readout that will not move the goalposts later.
Activation is not finished when the early metric rises. The later retention outcome decides whether the signal remained meaningful.
First value is a customer outcome, not an onboarding step.
Next
Next session defines healthy engagement at the product's natural rhythm, finds what retained customers repeat, and builds the next useful action into the experience.
shipai.club · Free, always