Role-based first runs and AI personalisation

A generic first run is a compromise that serves no one well. The fastest path to value for a founder evaluating your tool is not the fastest path for the analyst who will use it daily, and the activation rate of a one-size-fits-all first run is always lower than it should be — it is the average of several persona-specific paths, each of which would convert better if it ran alone.

Role-based first runs fix this by routing each persona to the value that matters to them first. You ask one or two questions at sign-up — 'what is your main goal?' or 'what role describes you best?' — or you infer the answer from the enrichment data you already have, and you build two or three doors into the same product. The evaluator gets a pre-filled sample that shows the outcome in one screen; the daily operator gets the configuration path that makes the tool theirs; the team lead gets the collaboration feature first. Same product, different doors, each opening onto the relevant aha moment. The full mechanics for designing and instrumenting this are in the role-based first-run experiences with AI personalisation playbook.

AI personalisation at scale. The static role-based fork is the baseline; AI personalisation is the next layer. An AI that adapts the first-run path, the sample data, the next-best-action prompt, and the support content to the specific user in real time can add a further 15 to 30 percent lift on activation on top of a well-built static baseline, because it does at scale the thing a skilled onboarding specialist would do by hand for one customer. The personalisation signal does not need to be elaborate: the job-to-be-done from the sign-up question, the industry from enrichment, and the first two in-product actions are usually enough to branch meaningfully.

Worked example. A solo founder running a financial-reporting SaaS found that CFOs and finance analysts both signed up but churned at wildly different rates. Digging in, she found CFOs wanted to see a board-ready output in the first session; analysts wanted to start from a raw data import. She added a single post-sign-up question and split the first-run flow into two paths. CFOs got a pre-built board pack they could customise; analysts got an import wizard. Activation lifted 28 percent within the first cohort that ran through the split, with no change to the product itself — only the routing.

This kind of personalisation is exactly what the visitor-to-lead conversion work sets up: learning enough about a visitor before they sign up to route them to the right door the moment they do.