Open app

September 16, 2026

The Lifecycle section arrives with the four stages, the activation funnel's Overview, Board, List and Cohorts, funnel management, lifecycle and score sections on every profile, the evaluator that computes it all from your synced profiles with the lifecycle definition editor in settings, and — closing the set — the four shipped scoring models, the Retention and Expansion monitors, and a scoring model editor in settings.

The Lifecycle section

Admin

The workspace sidebar gains Lifecycle, with four fixed entries: Activation, Retention, Expansion and Reactivation. Each is a stage over a population of profiles — new signups, active accounts, churned and dormant accounts — and every profile is in exactly one of four lifecycle states: new, dormant, active or churned. A stage with one funnel opens it straight away; with several it lists them. Retention and Expansion say their monitors are not built yet, but funnels inside them work today.

Analyze a funnel

Admin

Every funnel has four tabs over a shared Signed up in the last window of 30, 90, 180 or 365 days. Overview shows entered, converted, stuck and median time to convert, a step-by-step chart that counts completions per step with the conversion from the previous required step and the median time to get there, and a fit × intent quadrant with an explicit unscored bucket. Board is a read-only triage view — one column per required step plus Converted, the top 25 cards by intent, and chips for every step each card has completed. List adds current step, days in step, status, fit and intent to the profiles table and sorts by score with Not enough data and Not computed grouped after the numbers. Cohorts shows rolling weekly or monthly signup cohorts with the share completing each step, a History starts here marker on cohorts older than your first connected source, and a marker on the bucket where the funnel or lifecycle definition changed. Each page says when its figures were last computed.

Create, edit, archive and delete funnels

Admin

Workspace admins can add a funnel to any stage from a template — Activation, Win-back, Feature adoption — or from scratch, up to ten per stage. Edit funnel opens the steps and entry-criteria editor: each step has a name, a stuck threshold in days, a weight, whether it is required, and a Completes when rule built from property comparisons, days since a date, activity counts, lifecycle states and milestones, or marked by hand. A rule on an undeclared property is flagged inline. Changing steps or criteria creates a new version of the funnel; enrolled signups keep the version they were evaluated under. A funnel with history can be archived, and stays reachable from Show archived; a funnel that never enrolled anyone can be deleted.

Lifecycle, scores and enrollments on the profile

Admin

A profile's page now shows its lifecycle state and how long it has been there, its activated, churned and won-back milestones (with backfilled ones marked as approximate), every profile-level score with the factors that produced it and how fresh its inputs were, and one section per funnel it is enrolled in with each step's completion and the funnel's intent score.

Lifecycle states and funnels are now computed from your data

Admin

Lifecycle states, funnel enrollments and step completions are now evaluated from synced profiles rather than seeded. Three passes run in order — lifecycle, enrollments, scores — after every successful sync, every six hours in between so the dormant and churn timers advance, and whenever an admin saves the lifecycle definition or clicks Recompute now. Churn has a grace period: an active account must fail the Active while rule for the configured number of consecutive days before it reads churned, so a quiet long weekend never flips it. The first evaluation places every profile from what is true today, with backfilled activations marked as approximate, and enrolls every profile in the activation funnel from its signup date so conversions that happened before your first sync still count. Editing a funnel now takes effect on the next evaluation: in-progress enrollments follow the new version, converted and exited ones keep theirs.

Define the lifecycle in settings

Admin

Settings → Lifecycle holds the definition every stage reads: the Activated when rule, the Active while rule, and the Dormant after and Churned after timers. Preview runs the lifecycle pass under your draft without saving and shows, per state, how many profiles are in it today and how many would be after a recompute. Save and recompute creates a new version and recomputes every profile under it in the same request. The Evaluation section lists recent passes with what each did or why it failed.

A timeline on every profile

Admin

A profile's page gains a Timeline: when it was created in your source and every lifecycle transition since, newest first, each with the state it moved from and to. Times Uptend did not watch happen read "around" a date rather than an exact moment.

Scores are now computed

Admin

Every workspace ships with four scoring models, and the scoring pass now computes them after every evaluation: ICP fit on every profile, Activation intent on every activation enrollment, and Engagement and Churn risk on every active account. Each score names the factors that produced it and how fresh its inputs were; an account too thin to score fairly reads Not enough data with the missing inputs listed, never a low number. Activation intent is built from step progress, pace and recency, so a signup from yesterday gets a real number. Engagement and risk are cleared when an account churns rather than left to read as live figures. The shipped models read placeholder properties — company_size, plan, users_total, projects_total, last_seen_at — so map those keys, or expect Not enough data until you do. Existing workspaces receive the four models on their next evaluation.

The Retention monitor

Admin

Lifecycle → Retention is now a monitor over your active accounts: how many are active, how many churned in the window and the churn rate that implies, how many were won back or went dormant; a retention curve by activation cohort showing the share of each weekly cohort still active one, two, four, eight and twelve weeks later, with History starts here on cohorts that activated before your first source was connected; an At risk list ordered by churn risk with engagement and the top risk factors inline; and this window's Transitions, newest first. Funnels you add to the stage are listed underneath.

The Expansion monitor

Admin

Lifecycle → Expansion shows the fit × engagement quadrant over active accounts, the Thresholds your scoring models declare — ten users, five projects — with how many accounts are past each line now and how many crossed it in the window, and Crossed this window, the accounts first seen past a line, newest first. A crossing is recorded once, when an evaluation first sees the account past the line, and again when it falls back; crossings also appear on the account's profile timeline.

Edit scoring models in settings

Admin

Settings → Scoring lists the four scoring models with their category, what they are anchored to, their funnel, version, status and when each was last computed, and opens each into an editor. Workspace admins can change a model's factors — which property or activity each reads, its weight, its transform and half-life, whether it is inverted — along with the minimum weight that must have an input, the bands, and, on profile models, the threshold lines the Expansion monitor watches; the name and the active flag are editable too. The key, category, scope and funnel are fixed, and models cannot be created or deleted. A factor on an undeclared property is flagged as you type and refused on save, and a live warning names any factor whose absence alone would leave every account missing it reading Not enough data. Preview scores the accounts the model covers under your draft without saving and shows, per band, how many are in it today and how many would be after a recompute, with the commonest reasons for Not enough data. Save and recompute creates a new version of the model and recomputes every score under it in the same request; a rename or an on/off change saves without a recompute.

On this page