How a SaaS team would turn onboarding into a journey that finishes itself

The starting point

In this scenario an analytics SaaS has healthy signups and weak activation: plenty of people create an account, connect nothing, and never return. Onboarding was a five-email drip that fired on a timer — the same sequence whether or not the user had done anything — and the customer-success team could only hand-hold the largest accounts. Nobody could say which step a given account was stuck on until it had already churned.

The journey

The team designs the activation flow as a storyline — a graph the agent follows with each new account:

1. Welcome. On first sign-in the agent introduces itself and offers to help connect the first data source, right there in the app.
2. Connect a data source. The step doesn't advance because the user says they did it — it advances when the product reports the real event back to the agent through a callback. Until that signal arrives, the agent stays on this step and offers a lighter path if the user goes quiet.
3. Build the first report. Same shape: the agent guides, and the product's own "report created" event moves the account forward.
4. Invite a teammate, then form the weekly-review habit — each a step, each advancing on a real signal, each able to branch to a nudge or a hand-off when someone stalls.

Because progress is saved to the person, the journey continues across the in-app assistant and email: a user who drops off after step 2 gets picked up on step 2 three days later, not restarted from the top.

What changes

The agent answers from the company's own help docs, so it speaks about this product, not a generic one. It runs around the clock, so a signup at midnight is met immediately. And it scales the judgement of a good onboarding specialist — what to do next, when to push, when to back off — to every account at once, routing a human only to the accounts that genuinely stall.

The numbers here are structural, not measured: one saved journey per user rather than a broadcast sequence, 24/7 coverage, and zero steps advanced on a user's say-so rather than a real product event. It's an illustration of the shape, not a report of a specific deployment.