Metric Tree
Decompose one north-star metric into 4-6 movable input metrics.
## 1. Pick 1 north-star metric — ## 2. Break it into 4-6 input metrics — ## 3. Assign one metric per motion — ## 4. Instrument the tree in real time — ## 5. Run weekly decisions against the tree —
Why it works▾
A metric tree is 2026’s control panel: 1 north-star metric, 5 input levers, 0 vague dashboards.
When to use▾
Use a metric tree when 1 team owns growth, marketing, and product, and you need 1 shared operating model instead of 12 dashboards. It works best when the north-star metric is stable, the inputs are measurable in PostHog, Clay, or Attio, and every weekly decision needs a clear trade-off. If you cannot move an input in 14 days, it is not an input, it is vanity.
Steps in detail▾
1. Pick 1 north-star metric
Choose 1 metric that maps to retained value, not activity: qualified pipeline created, activated teams, or weekly retained users. In 2026, I would not pick sessions or CTR because Meta Advantage+ and PMax can inflate both. Example: a B2B SaaS team uses PostHog to define “activated account” as 3 users hitting 5 key events in 7 days, then ties that to Attio opportunity creation. If the metric does not survive an LLM audit in Claude, it is not a north-star metric.
2. Break it into 4-6 input metrics
Split the north-star into 4-6 movable drivers, each with a clean owner and weekly target. Example: north-star = qualified opportunities; inputs = target-account coverage in Clay, reply-to-meeting rate from 11x, product-qualified lead rate in PostHog, landing-page conversion in Framer, and stage-to-stage speed in Attio. Use Clay + LLM scoring to rank accounts by fit and intent, then stop adding inputs when the tree explains 80% of variance.
3. Assign one metric per motion
Map each motion to exactly 1 input so nobody hides behind blended reporting. Example: paid media owns landing-page conversion, outbound owns reply-to-meeting rate, lifecycle owns activation rate, and product owns time-to-first-value. In 2026, you can wire this through n8n: Meta Advantage+ feeds leads into Clay, Clay enriches them, and an 11x agent books meetings only for accounts above the fit threshold. If 2 teams share 1 metric, 0 teams own it.
4. Instrument the tree in real time
Build the tree from live sources, not Monday screenshots. Example: PostHog streams product events, June tracks attribution, Attio tracks opportunity stage, and Clay updates firmographic fit every night. Then use Claude Code to analyse drops in conversion and output a decision note like: “stage 2 to stage 3 fell 18%, driven by ICP mismatch in healthcare accounts.” If the metric tree updates monthly, it is already dead.
5. Run weekly decisions against the tree
Use the tree as the agenda for every growth review: 1 metric up or 1 metric down, then 1 action per input. Example: an AI-native operator sees qualified pipeline down 12%, uses Common Room to find which PLG signals vanished, asks Clay to rescore those accounts, and spins up a Lindy workflow to trigger personalised follow-up only on high-intent segments. Every weekly meeting should end with 3 numbers, 2 owners, and 1 next test.
Pitfalls▾
- 01Picking a north-star metric that is 1 proxy with no business value, like traffic or opens.
- 02Using 7+ inputs, because a tree with 12 branches is just a dashboard in disguise.
- 03Mixing leading and lagging metrics in the same layer, which makes ownership impossible.
- 04Letting Sales and Marketing share 1 input, because shared accountability becomes no accountability.
- 05Updating the tree in spreadsheets instead of PostHog, Attio, Clay, and a workflow tool like n8n.
Get one interactive framework in your inbox each week
Every Monday. No fluff.