Metrics Trees
Decompose your North Star metric and top business KPI into a tree of sub-metrics to align teams and estimate impact
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 3
- Confidence
- 92%
A metrics tree maps how a top metric (the North Star metric, measuring value delivered to users, or the top business KPI, measuring value captured) breaks down layer by layer into the sub-metrics that drive it — effectively the math formula that produces the top number. The two trees usually overlap in the middle, revealing the highest-leverage metrics. Teams use it to align the whole org around two shared top metrics, to assign sub-metric ownership, to shape team topology around goals, and to estimate how much a given experiment will actually move the top-level number.
Origin
Presented by Itamar Gilad as the goals-layer tool in GIST, building on the value-exchange loop (deliver value to the market, capture value back). The North Star metric concept is credited in the conversation to Lenny Rachitsky's widely-read article on the topic; Gilad refines the definition to specifically mean value created for the market (e.g. WhatsApp messages sent, Airbnb nights booked), distinct from the top business KPI.
Core principles
- 01The North Star metric measures value delivered to users; the top business KPI (revenue/profit) measures value captured — the two sides of the value-exchange loop
- 02Every increment of the North Star metric is a small unit of user value (e.g. each WhatsApp message, each Airbnb night booked, each document created)
- 03Metrics in the overlap of the two trees are super-important because moving them moves everything else
- 04The tree lets you estimate an experiment's impact: moving a bottom-of-tree metric like activation rate 10% probably moves the global metric only a tiny amount
- 05If you don't know the math formula behind your top metric, working that out should itself be the work
How to run it
- 1
Pick the two top metrics
Choose one North Star metric that measures core value delivered to users (documents created per month, nights booked, messages sent) and one top business KPI that measures value captured (revenue or profit).
Pro tip A published 'North Star' is often just a company's single most-important metric; the true North Star specifically measures how much value you create for the market.
Watch out Measuring only revenue/business metrics with no user-facing value metric is a telltale sign you are not evidence-guided.
- 2
Decompose each into a tree of sub-metrics
Break each top metric down through successive layers into the variables that mathematically drive it, until you reach levers teams can directly work on. Identify where the two trees overlap.
Pro tip Mapping it out explicitly (not just holding levers in mind) is what lets you estimate the impact of a candidate experiment before you build it.
Watch out An experiment on a deep, bottom-of-tree metric will usually move the global number only a small amount — check this before committing resources.
- 3
Assign ownership and shape topology around the tree
Give teams sub-metrics to own on an ongoing basis for a sense of ownership and mission, and arrange team topology around the goals in the tree rather than around software structure or hierarchy, re-adjusting when goals change.
Pro tip The shared tree drives alignment — teams will sometimes set aside their own ideas to jump on another team's idea once they see it moves the North Star metric more.
Watch out Team topology built around software structure or a hierarchy, rather than around goals, misdirects effort; revisit it from time to time as goals shift.
In the wild
WhatsApp measured messages sent for a long time 'because every message sent is a little incremental of value for the sender' — rich media, from anywhere, free to the receiver, versus costly SMS. Going from one billion messages a year to two billion roughly doubles delivered value.
→ A North Star metric that directly tracks user value delivered, distinct from any revenue figure.
Amplitude measures weekly active learning users: users who found an insight in the tool so important they shared it with at least two other users who consumed it. It captures the core value moment, not just logins.
→ Demonstrates a North Star metric engineered to encode genuine value delivery rather than vanity usage.
Common mistakes
Confusing 'most-important metric' with a true North Star metric
Companies label their number-one metric as the North Star, but if it doesn't measure value created for the market it misdirects the whole tree; the North Star must track user value, the KPI tracks value captured.
Setting siloed per-department goals instead of shared top metrics
Separate engineering, design, and marketing goals push people in different vectors and make prioritization impossible; a shared metrics tree gives the whole org two aligned top metrics to move.
Is it for you?
Best for
Scaling companies with an established business model that need to align many teams and objectively estimate the impact of competing ideas
Not ideal for
Very early-stage startups that don't yet know how they create value and should be iterating toward product-market fit instead of building elaborate metric trees
From the transcript
“the metric we usually use to measure value delivered is called the North Star metric”
“the northometric is measuring how much value we're creating for the user the core value that they're getting”
“the top kpi is what we expect to get it should be revenue or profit”
“you can uh further break them down into what I call metrics trees”
“if you're gonna work on something at the bottom here like activation rate like say you move that ten percent how much is that going…”
“there's like a math formula that equals your North Star metric”
From the episode
Becoming evidence-guided
Itamar Gilad (Gmail, YouTube, Microsoft)