How should I decide what our real activation moment is and redesign around it?

Ruben MangorrinhaGrowth & AI product design (SaaS) · Mermaid Chart, Ex-PandaDoc, Ex-Miro

Our project management tool gets about 2,000 signups a week, but only 18% create a second project within 7 days. Our onboarding is a 5-step setup wizard, and I suspect it's scaring people off.

Generic AI

Start by questioning the metric

"Second project within 7 days" may not be your activation moment. It's a plausible guess, but many teams get lasting value from one well-used project, especially if they've invited colleagues and are moving tasks through it. Before redesigning anything, confirm which early behavior actually predicts retention.

Define retention first, since activation only matters as a predictor of it. Pick something like "active in week 4" or "still active at day 60," based on your natural usage cadence.
Activation should be an early behavior, within the first few days, that strongly separates users who retain from those who don't.

Find the behavior that predicts retention

Pull 2–3 months of signup cohorts. That's roughly 15,000–25,000 users, plenty for this analysis. Then list candidate early actions:

Created a project (first, second, third)
Added N tasks
Invited a teammate, or had an invitee accept
Assigned a task to someone else
Completed a task, or moved one across a board
Set a due date, commented, or connected an integration
Returned on a second distinct day

For each candidate, compare retention among users who did it versus those who didn't. Then look for thresholds. "Invited 1 teammate" versus "2+ teammates" versus "3+ tasks assigned" often shows a sharp knee where retention jumps.

Judge candidates on more than lift:

Coverage: what share of retained users did this? A behavior with huge lift that only 5% of retained users perform is a weak target.
Precision: what share of users who did it retained? If many perform it and still churn, it isn't the moment.
Timing: it should happen early enough that onboarding can influence it.

Collaboration signals often win for project management tools, because a tool used alone is easy to abandon. Let your data decide, though.

Add qualitative evidence

Correlation can mislead. Motivated users do everything, so every action looks predictive. Balance the numbers with direct input:

Interview 8–10 retained users about the moment the product "clicked" for them.
Interview 8–10 churned users about what they expected and where they stalled.
Watch session recordings of the wizard, focusing on hesitation, back-button use, and abandonment.
Add a one-question signup survey, such as "What are you here to do?" or "Solo or team?" You'll likely find distinct segments with different activation moments.

Diagnose the wizard before blaming it

Instrument each of the five steps. Measure the share who start, complete each step, and finish, plus time spent per step. You might find:

One step causes most of the drop-off, such as importing data or inviting people too early. That calls for a targeted fix, not a rebuild.
Users finish the wizard but still don't activate. Then the wizard isn't scary, it's just not leading anywhere valuable.
Wizard finishers activate far more than abandoners. Check whether that's causal or just motivated users self-selecting.

Redesign around the activation moment

Once you know the moment, work backward from it. Every step should either move users toward it or be deferred.

Cut setup to the minimum needed to reach value. Defer workspace naming, settings, and integrations until after the first win.
Use templates or a pre-filled sample project so users see a working board immediately instead of an empty state.
Branch by intent, using the signup survey. Solo users and team leads need different paths.
If collaboration is the moment, make inviting feel natural, for example by prompting "Assign this task to someone" rather than a generic "Invite your team" step.
Replace the linear wizard with a short, dismissible checklist tied to activation actions, so users can explore without being blocked.
Follow up with triggered emails or in-app nudges for users who stall just short of activation.

Test it properly

At 2,000 signups a week, you can run clean A/B tests.

To detect a lift from 18% to 21%, you need roughly 2,700 users per arm, about 3 weeks at a 50/50 split. Smaller lifts take longer, so test bold changes first.
Your primary metric should be the validated activation behavior. Confirm with downstream retention at week 4 or later. It's easy to boost an activation metric without improving retention, especially by pushing hollow actions like auto-creating projects.
Set guardrails: invite spam rates, support tickets, and conversion to paid if applicable.
Keep a small long-term holdout to check whether gains persist.

Summary

Use your cohort data to find the early behavior with the best mix of lift, coverage, and timing. Validate it with interviews, measure where the wizard actually loses people, and then strip onboarding down to the shortest path to that moment. Prove the result with experiments judged on retention, not just activation.

Ruben’s mind

First, which metric is your team responsible for this quarter? If it's activation, this is the right place to spend effort. If it's conversion or revenue, I'd still look at it, but I'd tie the activation definition to whatever predicts paying, not just using.

I'd also question two assumptions before redesigning anything: that "second project" is your real activation moment, and that the wizard is what's scaring people off. Either could be true. I just wouldn't bet on them yet.

Finding the activation moment

"Second project within 7 days" is a reasonable proxy, but it might not be the aha moment. In project management tools, a lot of the value shows up inside the first project: tasks assigned, a teammate invited, something moved to done. Someone with one busy project shared with three people may be far more activated than someone who created two empty ones.

You might try this:

Pull a cohort of users who are still active (or paying) at 30–60 days. Compare what they did in week one with what churned users did. Look at a handful of candidate behaviours: tasks created, a teammate invited, a due date set, a status changed, a second project created.
Find the behaviour, or combination, that separates retained users most clearly, and that happens early enough to design toward.
Then talk to people on both sides. Interview a few retained users about when it clicked for them, and a few who dropped off about what they expected and didn't get. The data tells you what correlates. The interviews tell you why, and whether it's something you can actually lead people to.

My hunch is that something collaborative will show up, like inviting someone or assigning a task to another person. Use cases that genuinely need several people tend to drive retention and invites more than any invite button does. That's a hypothesis to check, though, not a finding.

The wizard itself

Five steps isn't automatically bad. A first-time user often has more patience than they'll ever have again, and a bit of investment can make them more likely to stay. The real question is whether each step gives them something back.

Look at drop-off per step. If most people leave at one particular step, that tells you more than "five is too many."
For each step, ask who the answer is for. If it's "what's your company size" and it only feeds your CRM, cut it or move it later.
If a step's answer could shape what they land on, keep it and make that payoff visible. For example, "What are you planning? A product launch, a client project, a sprint" could drop them into a prefilled project with sample tasks, columns and due dates, instead of a blank board.
Match it to the journey. Someone who arrives wanting to quickly set up one project may want a skip option and a lighter path.

The goal is that when the wizard ends, they're already partway to the aha moment, not just at the starting line.

What I'd test

Control: the current wizard.
Variant A: the same number of steps, with company-only questions swapped for ones that prefill a tailored template project.
Variant B: a shorter path (one or two questions) into a template, with the remaining setup offered contextually later.

At your volume you should be able to read this within a few weeks, but check the sample size for the effect you'd care about before you start.

Metrics to watch

Your main metric should be the activation behaviour you validated, not automatically "second project." Alongside it, watch:

wizard completion rate and drop-off per step
time to first meaningful action
teammate invites in week one
30-day retention, to make sure a lift in activation isn't just people clicking through faster

Next steps: confirm your quarter's metric, run the retention cohort analysis, do around eight to ten interviews split between retained and churned users, audit each wizard step for payback, then launch the A/B test and keep adjusting from what you learn.

Ruben Mangorrinha

Generated by an AI mind licensed from Ruben Mangorrinha, not by Ruben personally.

Ask Ruben's mind about your case

3 free answers.