Sample 01
From 61 Dashboards to 20 Numbers That Each Decide Something
Sixty-one dashboards couldn’t say which customers were getting value. One metric could.
KPI & North Star Framework · $31.4M ARR B2B SaaS · 5 pages · Built on the companion analytics audit (Sample 03)
The situation
Leadership ran the business on 61 dashboards with nine definitions of “customer.” The engagement metric behind CS prioritization counted any login as activity — and more than a third of supposedly “active” accounts had nobody doing any real work in them. When every team reports a different number, no number can trigger a decision.
A third of accounts looked engaged. CS prioritized — and deprioritized — on that basis.
Those accounts had completed zero core workflows in 30 days. The business was measuring logins, not value.
The finding
The redesigned North Star — Weekly Value-Delivering Accounts — requires two distinct users completing real workflows that end in a downstream action, every week. In the fictional data, accounts above the threshold for eight straight weeks renewed at 96% and contracted −2.1%; those below renewed at 71% and contracted −11.8%. The signal was readable four to six months before renewal — long enough to act.
The 25-point renewal gap is an observed association in constructed data — a reason to investigate, not proof the metric causes retention. The framework treats it as a leading indicator to act on, not a causal claim.
Leadership stopped debating dashboards and started operating one number.
WVA coverage — the share of ARR in value-delivering accounts — is reviewed every Monday: the coverage, its four-week trend, the ARR that moved in or out, and any guardrail in breach. Below 78%, onboarding and reactivation capacity is reallocated before any new-logo spend is approved. Accounts out of the set for four straight weeks enter the renewal-risk queue automatically. Every metric on the scorecard names its owner, its threshold, and the action its movement triggers.
Your team has plenty of dashboards, but no agreement on which metrics actually indicate customer value — or what anyone should do when they move.
The evidence
The North Star, defined. Deliberately requires two humans and a downstream action — so it does not move when an administrator logs in to pull an export. That is precisely the failure that made 35% of “active” accounts meaningless.
An account is value-delivering in a week if, within the rolling seven days, at least two distinct users each completed at least one qualifying core workflow, and the account completed at least three in total. A workflow qualifies when it reaches a terminal success state and produces a downstream action — an export, an approval, or an integration write. Excluded: admin-only sessions, template edits, and re-runs of a failed workflow within 24 hours. The board-level form is WVA coverage: the share of ARR held in accounts that were value-delivering in at least six of the last eight weeks.
Observed association in fictional data · not a causal claim · signal readable 4–6 months before renewal
Why this metric survived and the alternatives did not. A North Star must predict the outcome that matters, lead the decision, be moveable by named teams, and resist accidental inflation.
| Test | How WVA performs |
|---|---|
| Predicts the outcome | 96% vs 71% renewal; −2.1% vs −11.8% contraction. That spread is the entire retention problem, visible in a weekly number. |
| Leads the decision | Stable and readable four to six months before a renewal conversation — long enough that intervention is still cheap. |
| Moveable by named teams | Onboarding moves first-week breadth, Product moves workflow completion, CS moves reactivation. Each team can point to its own input. |
| Resists inflation | Two known gaming vectors — nominal second users, trivial re-runs — are mitigated in the definition and the guardrails, with quarterly hand-audits of the largest movers. |
Candidates rejected — and why. The rejections are the part of this exercise that saves the most time later.
| Rejected | Why |
|---|---|
| Monthly active users / logins | Measures access, not value. 35% of accounts counted as active had nobody doing any work in them. |
| Total workflows run | Rewards volume and tolerates concentration: one power user running 400 trivial workflows outranks a team of six running thirty that matter. |
| Composite health score | A score of 62 tells nobody what to do. Composites hide which input moved and invite quiet reweighting when the number is unflattering. |
Also rejected: ARR growth (lags 6–12 months), NPS (11% response rate), customer count (top 25 carry 41% of ARR) · full list in the PDF
Guardrails — a breach escalates regardless of what the North Star is doing, so the metric cannot be improved by degrading something it doesn’t measure.
The driver tree. Read it downward, operate it upward — each function reviews the three inputs it owns.
Acquire (VP Sales / VP Marketing): risk-adjusted pipeline coverage · win rate by segment · blended CAC. Activate (VP CS / Head of Product): time to first value · week-4 account breadth · implementation→activation rate. Expand (VP CS): fixed-cohort expansion rate · % of ARR with an expansion event · expansion pipeline coverage. Retain (VP CS / CFO): revenue-weighted retention · contraction rate at renewal · ARR at risk. Six board metrics, fourteen operating metrics, everything else diagnostic — each with a single named owner, a person, not a function.
What changes on Monday
The weekly operating review opens with one slide: WVA coverage, its four-week trend, the ARR that moved in or out of the value-delivering set, and any guardrail in breach. Twenty-five minutes. The measure of whether the framework worked is not whether the dashboards look better in Q1 — it is whether the Q2 planning conversation reaches a decision faster than the Q4 one did.
How it was done
3-week project. Tidal Arc designed the hierarchy and wrote every definition; the client’s analytics team built the models. This framework answers the companion analytics audit — the assessment that found the reporting layer directionally wrong (Sample 03).
Wondering whether your dashboards measure activity or value? Discuss it on a discovery call →