SaaS Activation Rebuild

A workflow SaaS with strong retention but a leaky top-of-funnel. A prior agency had rebuilt onboarding twice and made it worse. We shipped a new activation flow, instrumented it end-to-end, and ran three experiments in eleven weeks.

The Challenge

A workflow SaaS had strong retention but a leaky top-of-funnel — and a prior agency had rebuilt onboarding twice and made it worse. Retention proved the product was worth using; the funnel just kept new users from finding that out. Two previous onboarding rebuilds — done on intuition — had made the numbers worse.

Our Approach

A squad of four shipped a new activation flow on Next.js, Postgres, and pgvector, instrumented it end to end, and ran three controlled experiments in eleven weeks — treating activation as a first-class metric, not a UI concern. Instrumentation came first: every funnel step measurable before anything was redesigned. The first shipped variant reached real users in four weeks, and three controlled experiments ran back-to-back inside eleven.

What We Built

A new activation flow, shipped as a live variant in four weeks; end-to-end funnel instrumentation — every step measurable; and three controlled experiments run back-to-back in eleven weeks.

Key Decisions

Activation as a first-class metric — the engagement was scoped to a number (Day-7 activation), not a redesign; every decision traced back to whether it moved that metric. Experiments over opinions — a prior agency had rebuilt onboarding twice on intuition and made it worse; we shipped variants to real users and let cohort data decide.

The Outcome

Day-7 activation up 112% cohort over cohort, median time-to-first-value down from 14 to 5.2 minutes, onboarding tickets down 41% — with the first variant live to real users in four weeks. Day-7 activation more than doubled, cohort over cohort — with pricing untouched and onboarding support tickets down 41%.

“The first agency we've worked with that treated activation like a first-class metric — not a UI concern.”