Work from top to bottom: first get your North Star clear, then fill in the funnel steps, and only then assign the experiments and channels.
01
North Star Metric
The North Star Metric is the one metric that measures value delivered to customers: not vanity (sign-ups), not output (features) and not purely financial (revenue alone). A good NSM is a leading indicator of sustainable revenue and correlates with retention. Examples: 'rides per active rider per week' (Uber), 'nights booked' (Airbnb), 'minutes of music listened to' (Spotify). One NSM for the whole team stops marketing from optimising for sign-ups while product works on retention.
Questions that help
- ·Which event proves that a user is really getting value from our product?
- ·Does this metric correlate with retention and revenue over three months?
- ·Can every function (product, marketing, sales, support) influence this metric?
Example
For a planning app for construction firms: 'number of fully published weekly schedules per active job planner per week'. It proves the core task was completed, not just that someone logged in.
Common mistake
Choosing a vanity metric ('total number of accounts'). It always goes up and steers nothing.
02
Acquisition
Acquisition is about how new users arrive in the first place. For each channel you state the volume, the quality and the CAC that come with it. Not every channel works for every product: a viral consumer app benefits from social and organic; a B2B SaaS from SEO, content and outbound. Good acquisition choices match the ICP: choose channels where your audience already is.
Questions that help
- ·Which 2 or 3 channels currently bring in the most qualified users?
- ·What is the CAC per channel, and how does it compare with the LTV?
- ·Which channel could we make three times bigger without quality collapsing?
Example
A SaaS tool for accountants: SEO on long-tail questions (40%), referrals through partner networks (35%), outbound LinkedIn (25%). Deliberately no paid ads: the CAC is too high for our cycle time.
Common mistake
Doing a bit of every channel at once. Better two channels done seriously than six done superficially.
03
Activation
Activation is the step in which a new user experiences the aha moment: the first time they realise why this product makes their life better. Often it is a specific behaviour in the first session or first week. Those who reach the aha stick around far more often than those who miss it. The biggest lever for growth sits here: a few percentage points of activation lift the whole funnel.
Questions that help
- ·Which behaviour in the first session correlates with 30-day retention?
- ·What percentage achieve that behaviour now?
- ·Which friction sits between sign-up and the aha moment?
Example
A newsletter tool: 'first email sent within 24 hours'. Those who do it have 70% retention at 90 days; those who do not, 12%.
Common mistake
Defining the aha moment as 'sign-up completed'. That is a tick in a box, not an experience of value.
04
Retention
Retention tells you whether people come back, and so whether your product delivers real value. You usually measure it in cohort curves: of a group that started in week 0, what percentage is still active in weeks 1, 4, 12, 26? A healthy product shows a 'smile curve': the curve drops at first but levels off (or climbs) after a few weeks. Falling to zero means no retention, and no business.
Questions that help
- ·What does our retention curve look like per cohort?
- ·Which users stay, which leave, and what sets them apart?
- ·Which product or onboarding changes would lift retention?
Example
A fitness app: after 90 days, 18% are still active weekly. Almost all power users turn out to have connected with a friend within the first week.
Common mistake
Looking only at churn among paying customers and ignoring free users. Free users are often the signal of whether the product itself works.
05
Referral
Referral measures whether existing users bring in new ones: a viral coefficient. Loops can be organic (word of mouth, social shares), incentivised (referral credits, friend discounts) or built into the product (multi-user tools where sharing is core usage). A viral coefficient above 1 means every user brings in more than one other on average: that is exponential growth. Between 0.3 and 1 is a healthy supplement to acquisition; below 0.1, referral is negligible.
Questions that help
- ·Do our users bring anyone along, and how do we find out?
- ·Which form of incentive suits our product without lowering lead quality?
- ·Which part of the product experience is inherently social?
Example
A planning tool that works per team: every new user invites an average of 2.4 colleagues within 14 days. Viral coefficient: 0.6 (60% of invitations are accepted).
Common mistake
Setting up a referral programme without an organic signal first. If users do not recommend it of their own accord, a discount rarely fixes that.
06
Revenue
Revenue is how you turn delivered value into money: conversion from free to paid, ARPU, expansion (upgrades and cross-sell) and net churn. A good growth model shows per cohort how the revenue curve runs and where the levers are: pricing, packaging, expansion. Expansion with existing customers is often cheaper than acquiring new ones.
Questions that help
- ·What conversion do we see from trial or free to paid?
- ·Which pricing tier is growing fastest, and why?
- ·How much expansion revenue comes from existing accounts?
Example
B2B SaaS: 9% trial-to-paid conversion, average ARPU of €120 a month, net revenue retention of 118% (expansion outweighs churn).
Common mistake
Looking only at new MRR and ignoring expansion. Existing customers are often the quietest growth engine.
07
Experiment Backlog
The experiment backlog is a prioritised list of hypotheses, each linked to a funnel step. For each item: a short description, the hypothesis, the linked metric and an ICE score (Impact × Confidence × Ease, each from 1 to 10). Every week the team runs the top X. A healthy backlog holds 20 to 50 items, is topped up continuously and pruned consistently. Not every experiment has to be a 'win'; most will not work, and that is a valuable insight.
Questions that help
- ·Which 5 experiments are at the front of this sprint, and which metric do they affect?
- ·What did we learn last sprint: won, lost, inconclusive?
- ·Which hypotheses would deliver a lift of 5% or more if they succeed?
Example
Top 3 this sprint: (1) add an onboarding checklist [activation, ICE 8/7/9], (2) a referral pop-up after the 5th use [referral, 6/5/8], (3) a re-engagement email after 14 days [retention, 7/6/7].
Common mistake
Running experiments without writing the hypothesis down first. Afterwards you cannot learn anything.
08
Channels & Resources
Channels and resources are the practical reality: which channels you have available, what budget per month, which team members are free for experiments, and what tooling (analytics, A/B testing, email, attribution) is in place. Many ambitious plans run aground here: no developer capacity for experiments, no analytics to measure with, no budget for paid tests. This block forces you to be realistic about what you can actually run this month.
Questions that help
- ·Which channels can we seriously use this month?
- ·What experiment budget and how many developer hours are available per sprint?
- ·Which measurement infrastructure are we still missing, and what is that blocking in our backlog?
Example
Available: 16 developer hours per sprint, a €3k experiment budget per month, analytics through Mixpanel, A/B testing through PostHog. Not available: a paid social budget or attribution tooling, so the top-of-funnel test of LinkedIn ads is on hold.
Common mistake
Building an experiment backlog without facing the reality of capacity. You will never get past item 5.
