The frameworks behind every engagement.

Most teams treat activation as a single metric — percentage of signups who 'activate' — which hides where users actually drop off. Split it into four stages instead: Signup → First Value → Habit Formation → Expansion, and instrument each transition separately.

Signup → First Value is almost always the biggest leak. If a user can't reach the moment your product proves its worth within their first session, everything downstream is compensating for a broken first impression. Measure time-to-first-value in minutes, not days.

First Value → Habit Formation is where onboarding either compounds or dies. The teams that win here build a specific trigger — a scheduled reminder, a usage streak, a second use case unlocked — rather than hoping users come back on their own.

Last-click ROAS rewards whichever channel touched the conversion last, regardless of what actually drove the decision. It also says nothing about how long it takes to earn back what you spent — a channel can show '3x ROAS' and still be cash-negative for eight months.

Build CAC payback by cohort instead: group users by signup month, track cumulative revenue per cohort against acquisition spend for that cohort, and plot the month payback crosses zero. Compare that curve across channels, not the vanity ROAS number each platform reports.

In practice, this reshuffles budget fast — channels that looked mediocre on ROAS often have the shortest payback because they attract lower-churn users, and vice versa.

Pure PLG stalls on complex, high-ACV deals that need a champion inside the buying org. Pure SLG is too slow and expensive for the long tail of smaller accounts who'd rather self-serve. Most software companies eventually need both — the failure mode is running them uncoordinated.

Define sales-assist triggers from product usage, not firmographics alone: a free account hitting a usage ceiling, multiple users from the same domain signing up independently, or a feature-gate hit repeatedly. Route those signals to sales; let everything else self-serve.

The blueprint only works if product, marketing, and sales agree on the trigger definitions upfront — otherwise sales chases signals that don't convert, and self-serve users who should've been routed to sales churn silently instead.

A churn risk score that only produces a monthly report changes nothing. It earns its keep when a risk tier automatically triggers an intervention — an in-app nudge, a CS outreach, a win-back offer — while there's still time to act.

Score accounts weekly, not monthly. Churn risk compounds fast in usage-based products; a monthly cadence means you're often intervening after the decision to leave has already been made.

Tier interventions by account value and risk severity. A high-value account at moderate risk deserves a CS call; a low-value account at high risk might only justify an automated email sequence. Treating every at-risk account the same wastes your highest-leverage resource — your team's time.