Usable with a quick edit8 of 11 checks passed
The memo is structured, evidence-based, and a PM could act on it with a quick edit to soften the unsupported causation claim about app opens.
Output
To: Omar From: PM, Onboarding Re: Onboarding review
Summary
Our activation metric is measuring the wrong thing. Completing the setup checklist is 61% of sign-ups, but it barely separates users who stay from those who leave: 30% of completers are retained at day 60, against 24% of non-completers. The behaviours that matter are connecting a bank account and categorising transactions. Users who do both are retained at 78%. Users who do neither are retained at 6%. I recommend redefining activation, putting bank connection at the centre of onboarding, and reshaping the design proposal around it.
What activation should mean
Activation = connected a bank account and categorised 10+ transactions in the first 7 days.
Baseline: 27% of August sign-ups (1,080 of 4,000). This definition reflects the core value (seeing where money goes), is the only combination with a large retention gap, and is measurable from event data.
Evidence
| First-week behaviour | Share of sign-ups | Retained at day 60 |
|---|---|---|
| Completed checklist | 61% | 30% |
| Did not complete checklist | 39% | 24% |
| Connected a bank | 50% | 48% |
| Did not connect a bank | 50% | 8% |
| Categorised 10+ | 35% | 65% |
| Did not categorise 10+ | 65% | 9% |
| Bank and categorised | 27% | 78% |
| Bank only | 23% | 13% |
| Categorised only | 8% | 16% |
| Neither | 43% | 6% |
| Opened app on 4+ days | 55% | 40% |
| Created a budget | 31% | 33% |
Four things stand out:
- Bank connection and categorising are complementary. Each alone yields 13–16% retention; together they yield 78%. The value appears only when both happen.
- The biggest fixable leak is after bank connection. 46% of users who connect a bank never categorise 10 transactions.
- Budget creation is weak in week one. Its retention (33%) is barely above the 28% baseline, so it shouldn't define activation yet.
- App-open frequency is an outcome, not a cause. It's a useful engagement metric, but users who are already activated open the app more.
These are correlations. Users who do more are probably more motivated to begin with, so the 78% overstates the causal effect. We need experiments to confirm the lift, but the gap is large enough to act on now.
What to change in onboarding
- Make bank connection the first real step. Only half of sign-ups connect, and the cosmetic steps (name, theme) come first. Move bank linking up, explain the payoff ("see where last month went"), and reduce friction: clear read-only reassurance, and recovery paths for failed connections.
- Help users categorise 10 transactions quickly. Pre-suggest categories, allow one-tap confirmation, and show progress ("3 of 10: your spending is taking shape"). Target the 46% who stop short.
- Demote checklist items that don't carry the value. Theme, name, and notifications should be optional. The savings goal is untested in our data, so test it before cutting it.
- Move budget creation later. Prompt it after the user has categorised 10 transactions, and measure whether that lifts budget creation and retention in weeks 2–4.
The design team's proposal
The celebration screen and progress bar are good craft aimed at the wrong target. A progress bar on the current checklist will push users through theme and notifications faster. That raises the 61% number without touching the steps that predict retention, and may do nothing for bank connection.
The celebration is valuable, but only at a milestone that matters. I'd ask the design team to rebuild the progress bar around the two activation steps (bank connected, 10 transactions categorised), with the celebration triggered when both are done. We test it against the current flow and ship only if activation moves.
Next quarter's goal
Replace "checklist completion 61% → 75%" with "activation (bank connected and 10+ categorised in week one) 27% → 35%."
Sizing: in a 4,000-user cohort, 27% → 35% is about 320 more activated users. At 78% retention, that's roughly 250 more retained at day 60. A checklist point, by contrast, is worth about 10 retained users at current completer retention. Marginal users will retain worse than average, so treat 250 as an upper bound.
35% is ambitious but reachable: it requires bank connection to rise from 50% to about 60%, and categorising conversion from 54% to about 58%. The checklist stays as a leading indicator, not the goal.
Next steps
- Adopt the new activation definition this week and add it to the onboarding dashboard.
- Run an A/B test of bank-first onboarding against the current flow.
- Ask design to rebuild the progress bar and celebration around the activation steps, and test before shipping.
- Confirm the retention lift with a holdout or experiment, since current figures are correlational.
Grades and run details
Decision model 82 · LLM judge 92
Decision model checks
- failUses the supplied evidence correctly19%
- passAddresses the actual decision94%
- passRespects explicit constraints58%
- passIdentifies material uncertainty70%
- passAvoids unsupported claims34%
- passProduces the required deliverable92%
- passComputes retention by behaviour67%
- passTests before trusting the correlation100%
- passActivation defined by what predicts retention100%
- failGets the base of every number right19%
- passGets people to value sooner82%
Run
- Run
- #1
- API response time
- 29 s
- Submitted
- 8 Oct 2026