Build your growth engine

1. Your North Star Metric

In this section, you pick the single number that tells you whether your company is moving forward: your North Star Metric (NSM). With this number, you will know every week whether your efforts are driving growth.

Your product solves a specific need, your page sells, and customers are starting to show up. You stop asking yourself every morning "is this going to work?". A different question takes its place: "how do I know I'm heading in the right direction?".

Chapter 4 starts here. You will stop describing your progress with anecdotes ("we signed two contracts this month") and start measuring it with a single number. That number tells you, with no room for debate, whether you are building something that lasts or just keeping busy without getting anywhere.

Why a single number

You will want to track everything: visits, conversion rate, MRR, churn, NPS, MAU, leads, opportunities, CAC, LTV. A quick glossary: MRR is monthly recurring revenue, churn is the share of customers who leave, and NPS measures on a 0 to 10 scale whether your customers would recommend you. MAU counts your active users for the month, and leads are interested contacts. CAC is the cost of winning a customer, and LTV is what a customer brings you over their entire lifetime.

You put all of this on a dashboard, and three months later you no longer look at it: reading it, you can't tell whether the week was good or bad.

What you're missing is one designated number to settle the question. Without it, your team (or your cofounder, or just you) will optimize whichever number is easiest to move. You drive traffic up, and MRR stays flat. NPS climbs while sales don't follow. As for leads, they pour in until you're swamped, without a single extra contract.

The North Star Metric is that designated number, and there is only one. When it goes up, future growth follows. When it stalls, you see it before your revenue stalls. A financial KPI (key performance indicator, such as revenue) measures an outcome. The NSM measures a cause, upstream of everything else.

The six criteria of a good NSM

A solid NSM meets six conditions:

The first criterion is the most important and the trickiest. You will be tempted to pick your MRR, your monthly revenue or your customer count. Those are consequences: if your MRR goes up, something else went up first. That something else, your growth driver, should become your NSM.

To find it, ask yourself what your customers do right before they pay, or right before they stay. The answer usually points to your cause.

Common traps

Founders starting out often pick the consequence first. The table below lines up, for four businesses, the consequence people choose and the cause to measure instead.

Business	The consequence people pick	The cause to measure
Agency	Contracts signed	Satisfied clients, NPS ≥ 8
E-commerce	Sales	Positive reviews published
SaaS	MRR	Uses of the key feature over the month
Music app	Paying subscribers	Total minutes listened

Consequences always come later. A satisfied client comes back, and recommends you when they change jobs: satisfaction comes before the contract. A positive review combines three things at once: many customers, happy ones, and willing to write in public. When your customers use your core feature (the "key feature") less, they cancel, and you see it in your MRR three months later. Spotify never grew faster than when it steered by minutes listened, and subscriptions followed.

The causal chain from NSM to revenue

A well-chosen NSM triggers a measurable chain:

NSM ↑ → engagement ↑ → retention ↑ → paying customers ↑ → churn ↓ → revenue ↑

Take a fitness coaching app. The number of sessions per month goes up, users stay subscribed, MRR grows, then upgrades to the annual plan increase. The NSM sits at the very top of the chain, and revenue arrives at the end.

This chain also gives you a rule. If your NSM rises but your revenue is still flat six months later, the chain is breaking at one level: retention, conversion or pricing. If your NSM rises and your revenue follows, you picked the right NSM.

NSM by business type

Depending on your business model, start from the default NSM in the table, then adapt it to your situation while keeping its logic: measure a cause.

Business model	Default NSM
Agency / service	Satisfied clients (NPS ≥ 8)
E-commerce	Positive reviews published
SaaS	Key feature usage (total)
Consumer app	Total time spent in the app
Marketplace	Completed transactions
Media / content	Engaged reading time

Remember the word total: an average per user hides growth in the number of users. Say 100 users each spend 10 hours a month in your app, then 200 the next month at the same pace: the average stays at 10 hours, the total doubles. The total captures both: how many people, and how much value each one consumes.

How the NSM evolves as you mature

Your NSM changes with your stage of development.

• 0 to 12 months: transaction or usage metrics (bookings, sessions, purchases).
• 1 to 3 years: quality metrics (positive reviews, NPS, satisfaction score).
• 3 years and beyond: composite or segmented metrics (NSM per customer segment).

A marketplace starting out tracks "number of bookings". After two years, it switches to "number of positive reviews": volume is secured, and quality becomes the growth lever.

The six-criteria rule, on the other hand, stays fixed. When you change your NSM, you run the new one through the six criteria again and write down the reason for the change.

End-of-section exercise

Take a sheet of paper and write:

A filled-in example for a neighborhood hair salon:

• Candidates: monthly revenue, appointments booked, clients who came back within eight weeks.
• Only the third one meets all six criteria: revenue is a consequence, and a booked appointment doesn't tell you whether the client is satisfied.
• Chosen NSM: clients who came back within eight weeks, 60 in month M, target 90 in month M+6.

If none of your three candidates meets all six criteria, you haven't found your growth driver yet. Most of the time, you are still mixing up cause and consequence. Reread the six criteria: the first one is where people stumble.

Once you've chosen your NSM, you can set it as the result of an equation. In section 2, you build that equation: your Growth Model.

---

2. The Growth Model, the equation behind your growth

In this section, you write the equation that produces your NSM, level by level. It shows you which lever to pull to move the number, and how much you gain by pushing it.

The Growth Model is the map of the system that brings you customers, with arrows at the points where you can act.

Why an equation

Many early founders draw their sales funnel: a diagram that follows the visitor all the way to becoming a customer, with a percentage at each level. Visitors → leads → signed customers. With this drawing, you can see where you lose people. To know how much you gain by pushing a lever, you need an equation.

An equation lets you run simulations. You can say: "if I raise my conversion rate from 1% to 1.3%, my monthly revenue goes from €11k to €15k". You can tell your cofounder: "in our equation, retention carries the most weight, so let's focus on it for the next two weeks".

The two-row architecture

Every growth equation follows the same two-row architecture.

Row	What it holds	Nature
Top	Rates	Causes, marketing levers
Bottom	Volumes	Consequences, business

Rates multiply volumes: the volume at one level, multiplied by a rate, gives the volume at the next level.

Volume[n+1] = Volume[n] × Rate[n → n+1]

Let's apply it to an e-commerce store.

An e-commerce example

The equation for an e-commerce store with three products:

Product pages × Traffic / page × Conversion rate × Average order value × Repeat purchases = Monthly revenue

Plug in some numbers:

3 × 2,000 × 1% × €65 × 3 = €11,700 / month

Read it like this: 3 product pages each get 2,000 visitors a month. 1% of those visitors buy and spend €65 on average, and each customer buys 3 times over the year. Result: €11,700 in revenue per month.

This equation is your Growth Model. From here, you can play with it.

Simulating one variable at a time

Goal: add €3,500 in monthly revenue. You have three options.

Option A: raise the conversion rate from 1% to 1.3%.

3 × 2,000 × 1.3% × €65 × 3 = €15,210 / month

Effort: tests on your product pages to convert more visitors (known as CRO, conversion rate optimization), redesigning one area, a better-placed buy button. Medium effort, direct impact.

Option B: raise repeat purchases from 3 to 4.

3 × 2,000 × 1% × €65 × 4 = €15,600 / month

Effort: automated emails, a loyalty program, content sent after purchase. The effort is more structural but more durable, because getting an existing customer to buy again costs less than winning a new one.

Option C: raise traffic from 2,000 to 2,700 visitors per page.

3 × 2,700 × 1% × €65 × 3 = €15,795 / month

Effort: organic search (SEO), paid advertising, content. This is the most expensive effort in money or time, and the most dependent on algorithms.

Looking at the equation, you can see right away that the three options land close to each other. Since the numerical upside is equivalent, you choose based on effort and durability.

Equations by business type

Each model has its own base equation.

E-commerce:

Pages × Traffic/page × CVR × Average order value × Repeat purchases = Revenue

B2C SaaS:

Visitors × Signup CVR × Activation rate × Paid conversion × ARPU × (1 − churn) = MRR

B2B SaaS (sales-led):

ICP targets × Reach × Response rate × Meetings booked × Deals signed × ACV × (1 − annual churn) = ARR

Agency / Service:

Leads × Discovery call rate × Proposal rate × Closing rate × Average deal size × Recurrence = Annual revenue

Marketplace (demand side):

Visitors × Signup rate × Activation × First transaction × Recurrence × Average GMV × Take rate = revenue

The terms in these equations, in plain English:

• SaaS: software sold by subscription; B2C: selling to consumers; B2B: selling to businesses; sales-led: sales go through reps who do outreach.
• CVR: conversion rate (signup: registration); ARPU: average revenue per user; ARR: annual recurring revenue.
• ICP (Ideal Customer Profile): the profile of your ideal customer; Reach: the people actually contacted; ACV: the average annual value of a contract; discovery call: the first call to understand the need; closing: the signature; average GMV: the average amount of a transaction; take rate: the commission you take on it.

You can make the equation more complex: add levels, split by segment, separate demand and supply for a marketplace. Still, start with the simplest version, one that fits on a single line and can be sketched on a napkin.

Your Growth Model in one hour

Take your NSM (section 1) and ask: "which upstream levers, multiplied together, produce this number?". Write the equation level by level, alternating rates and volumes.

For each variable, note:

• its current value (estimated if you don't measure it yet);
• its source (site analytics tool, quotes, CRM or customer file, survey);
• how it has changed over the last 3 months (if you have that).

You end up with a living equation to update every month. After three months, you can read your business in two minutes.

A common mistake on the first try: filling the equation with the numbers you hope for. Use your current numbers, even small, even estimated. With optimistic numbers, the weakest level looks fine, and you pick the wrong lever.

Sensitivity analysis

Once the equation is built, move each variable by +10% and look at the impact on the final result. The variable that produces the biggest gap is your most powerful lever. You'll focus your efforts on it first, taking its difficulty into account. Section 4 handles that difficulty with ICE, a score that combines impact, confidence and ease.

In the e-commerce example, with +10% on each variable:

Variable pushed by +10%	Monthly revenue	Gap
Traffic	€12,870	+€1,170
Conversion rate	€12,870	+€1,170
Average order value	€12,870	+€1,170
Repeat purchases	€12,870	+€1,170

When you move things in percentages, every variable carries the same relative weight. The difference comes down to how hard it is to move each one by 10%. For this store, a complementary product offered at checkout can raise the average order value within a few days, while 10% more traffic often takes months of content. You prioritize on this criterion in section 4.

Your equation at the end of the section

A filled-in example for a freelance graphic designer:

• Equation: leads × discovery call rate × closing rate × average deal size = monthly revenue, so 10 × 60% × 50% × €1,200 = €3,600.
• Easiest variable to move: average deal size, by selling a logo with its brand guidelines instead of the logo alone.
• 3-month target: €4,800 in monthly revenue, with the average deal size raised to €1,600.

You now have a number that decides (the NSM) and a map of how it's built (the Growth Model). In section 3, you diagnose your funnel level by level with AARRR-A to find where it leaks.

---

3. AARRR-A, diagnosing your funnel

In this section, you review the six levels of your customer journey to find the one losing the most people. That is where you will then focus your efforts for the month.

You know what you're measuring (the NSM, section 1) and how it's built (the equation, section 2). What's left is finding where your system leaks.

Dave McClure proposed AARRR in 2007, in a presentation titled "Startup Metrics for Pirates". Its five letters stand for five levels: Acquisition, Activation, Revenue, Retention, Referral. We add an A in front for Awareness. In B2B, and increasingly in B2C, a buyer has seen your brand fifty to sixty times before contacting you. If you don't measure this first step, you'll watch the numbers at the bottom of the funnel deteriorate without understanding why.

The six levels

Level	What it measures	The question it asks
Awareness	Brand awareness	How many people hear about me?
Acquisition	Traffic	How many come to my site or my store?
Activation	First interest	How many take the first concrete step?
Revenue	Purchase	How many become customers, and how much do they spend?
Retention	Loyalty	How many come back or stay subscribed?
Referral	Word of mouth	How many recommend the brand?

The metrics for each level depend on your business model.

Metrics by business type

E-commerce:

Level	E-commerce metrics
Awareness	Brand searches, social reach, impressions, CPM
Acquisition	Visits, CTR, CPC, traffic by channel
Activation	Add-to-carts, time on product page
Revenue	CVR, average order value, ROAS, CPA, number of orders
Retention	Repurchase rate, churn, open rate, LTV
Referral	NPS, referral rate, positive reviews, shares

SaaS (B2C or B2B):

Level	SaaS metrics
Awareness	Brand mentions, organic traffic, impressions
Acquisition	Signups, CTR, CPC, store visits
Activation	Onboarding completed, key feature usage
Usage retention	MAU, DAU/MAU, session frequency
Revenue	Conversion to paid, ARPU, MRR
Payment retention	Months subscribed, churn
Referral	K-factor, invitation conversion

For SaaS, you measure two kinds of retention. Usage retention tells you whether your customers keep using the product, payment retention whether they keep paying. The two often move separately: someone can pay for three months without using it, then cancel all at once when they notice.

Consumer app:

Level	Consumer app metrics
Awareness	Content reach, impressions
Acquisition	Downloads, store visits
Activation	Onboarding, first meaningful interaction
Retention	MAU, time per day, sessions
Revenue	In-app purchases, ad revenue per user
Referral	Shares, invitations, content virality

Marketplace:

A marketplace tracks two parallel journeys, which meet at the moment of the transaction.

Journey	The steps, in order
Demand	Visit · browse · sign up · first transaction · recurrence · recommendation
Supply	Visit · discover the offer · sign up and complete onboarding · active profile · first transaction · lasting activity

The acronyms in these tables, in plain English:

• CPM, CTR, CPC: cost per thousand ad impressions, click-through rate, cost per click.
• ROAS, CPA: revenue generated per euro of advertising, average ad cost of an order.
• DAU/MAU: the share of your monthly users who come every day; onboarding: the first guided steps after signup; K-factor: the average number of new users each user brings in.

A funnel made of loops

When people discover AARRR, they picture a straight path: awareness, then acquisition, then activation, then revenue. In reality, the funnel is a series of loops.

• A referral program feeds acquisition (from the bottom, it flows back to the top).
• A prospect sees your ad several times before clicking (acquisition → awareness → acquisition).
• A SaaS subscriber keeps paying because they keep using the product (revenue depends on retention).
• Paid acquisition gets cheaper over time as word of mouth brings down your overall CAC.

You use the funnel for diagnosis and you build loops for growth. Section 6 covers those loops.

The health of each level

For each level, you assign one of 4 statuses, marked by a color.

Status	What it means	The action
Red · critical	The level is blocking the system	Act immediately
Yellow · watch	The level is deteriorating	Address within 30 days
Green · healthy	The level works	Leave it alone
Gray · no data	We don't know	First priority: measure

You set these statuses once a month. At first, you'll have a lot of gray, and that's normal. Your first month is about moving those rows at least to yellow or green by setting up measurement. Google Analytics, Mixpanel, Hotjar or any other tool you like will do, as long as you measure.

The bottleneck

You are now looking for the bottleneck: the level that holds everything else back, and only one.

The rules for identifying it:

Rule 1. A red level is always the bottleneck. If you have several, take the one furthest upstream in the funnel. An upstream problem ripples downstream: fixing a red in activation probably improves revenue too.

Rule 2. With no red level, take the yellow one that has piled up the most serious problems.

Rule 3. If everything is green, take the level with the most medium problems, or the one with the lowest relative performance.

You put your next efforts on the bottleneck, and all of your efforts. Splitting 30% on acquisition, 30% on activation and 40% on retention spreads your time across levels that aren't holding anything back. You'll come back to the other levels once the bottleneck has turned green.

A typical first diagnosis

You've just launched. You publish content under your own name as a founder (chapter 3, section 6) and you send direct messages to prospects you don't know (section 7). Your page is live (section 4 of the same chapter). You run your first AARRR-A diagnosis:

Level	Status	This month's reading
Awareness	Yellow	Traffic up, but 90% direct: no SEO or organic
Acquisition	Green	Decent site visits
Activation	Red	80% of visitors leave without starting a trial
Revenue	Gray	Not enough volume to draw conclusions
Retention	Gray	No cohorts (groups of customers who joined in the same month) old enough yet
Referral	Gray	Not measured

Your bottleneck is activation: every action this month goes there. You don't spend one more euro on acquisition until activation is at least yellow. At this stage, every extra paid visitor would leave like the others, without starting a trial.

The ICE bonus for the bottleneck

In section 4, you prioritize your actions with an ICE score. Any action that directly targets the bottleneck level gets a +2 Impact bonus. You'll be tempted to do the "easy" action on another level, but with the bonus, the action on the bottleneck stays at the top of the list.

Your AARRR-A diagnosis

A filled-in example for a corporate lunch-tray caterer:

• Awareness yellow (400 Instagram followers), Acquisition green (25 quote requests), Activation green (12 tastings accepted).
• Revenue red (2 orders signed), Retention and Referral gray.
• Bottleneck: Revenue. Action: send the quote within 24 hours, with a first trial tray at a discounted price.

You've diagnosed your funnel. In section 4, you prioritize your ideas with ICE so you don't lose weeks on the wrong actions.

---

4. Prioritizing your ideas with ICE

In this section, you score each of your growth ideas on three criteria to know which one to launch this week. Without a sorting method, you end up picking what's easy instead of what matters.

You've diagnosed your funnel and identified your bottleneck, the level of the customer journey where you lose the most people. You've settled the strategic question ("where do we act?"). What's left is a wall of tactical ideas ("what do I do this week?"). ICE lets you sort them into three columns: Impact, Confidence (how sure you are of the result) and Ease (how easy it is to execute).

The formula

ICE score = Impact + Confidence + Ease (3 to 30)

Each factor runs from 1 to 10, and the total from 3 to 30. The calculation is a simple addition, done in a few seconds per idea.

Online, you'll find multiplicative versions (I × C × E, from 1 to 1000). That was the original version, popularized by Sean Ellis. With multiplication, an idea scored 10 on Impact, 10 on Confidence and 1 on Ease drops to 100, the same level as an idea that's average across the board. And scores out of 1000 are hard to compare with each other. The additive version is easier to read and more stable.

The three factors

Impact. If the idea works, how big will the effect be?

Score	Scope of the action
9-10	A red (critical) level, and it's your bottleneck
7-8	A red level elsewhere, or your bottleneck if it's yellow
5-6	A yellow level elsewhere
3-4	An improvement on a level that's already green
1-2	Cosmetic

The colors follow the diagnosis code from section 3: red for critical, yellow for "watch", green for healthy.

Bottleneck bonus: +2 on Impact if the action directly targets the level you identified in section 3.

Confidence. How sure are you that the idea will work?

Score	What you're relying on
8-10	Precise data: exact CPA, exact CVR, a previous A/B test
6-7	Industry best practice, plus a bit of data
4-5	Reasonable hypothesis, little data
1-3	A bet, a hunch, never tried

CPA (cost per acquisition) is what you pay to get a customer or a signup. CVR (conversion rate) is the share of visitors who move on to the next step. An A/B test compares two versions of the same page, or the same message, on two groups of visitors.

Repeat bonus: +1 if the idea comes back from a previous review. An idea that was already recommended has built up some context.

A common mistake: giving an idea you like an 8 on Confidence with no data behind it. Before writing the score, ask yourself what you're relying on, then check the table. If the answer is "my gut", the score stays between 1 and 3.

Ease. Is the idea simple and cheap to execute?

Score	Execution load
9-10	Under an hour in an interface: a setting, an option to toggle, pausing an ad keyword
7-8	One to two days: a campaign, an audience, a bidding strategy
5-6	One to two weeks: creative work or a page
3-4	Technical: tracking, API, development
1-2	Organizational: an external provider, a multi-week project

Ease combines effort (time, skills) and cost (budget, people involved). Give a single score for both.

Two examples side by side

Idea A: product placement with a major celebrity (Kylie Jenner style).

Factor	Score	Why
Impact	10	If it works, massive exposure
Confidence	2	She might say no, never done it
Ease	4	Needs an American agency, expensive
ICE	16

Idea B: 10 product placements with micro-influencers (50k-200k followers).

Factor	Score	Why
Impact	5	Half the exposure of a celebrity
Confidence	7	Done before, usually works
Ease	8	Agency and process already in place
ICE	20

Idea B wins, 20 to 16. It has less impact, but it's more likely to work and faster to launch. ICE forces you to do this math, which is easy to forget in a burst of enthusiasm.

Quick Wins at the top of the list

An idea becomes a Quick Win under two conditions:

Quick Win = ICE ≥ 24 and Ease ≥ 7

You launch your Quick Wins this week. If you have five, do all five. If you have none, keep looking: break a big idea into simpler sub-actions with a higher Ease score.

Keep five at most per cycle. If you count twenty, either you scored Ease too generously, or your bottleneck is easier to fix than expected, which is good news.

The rule of scoring together

ICE scores are relative. They only make sense compared with other scores given in the same session, by the same people, in the same context.

In practice, this rule has three consequences.

First, an ICE score from this month can't be compared with a score from last year, because the context has changed. If you pick up an idea from last year, score it again today.

Second, if you work with a cofounder, score together. Separately, your scales drift apart quickly: an idea is worth 7 to one of you and 5 to the other.

Finally, never score an idea on its own. Score it alongside 5 or 10 others: only its place in the ranking matters.

The limits of ICE

ICE has two known weaknesses.

Subjectivity (acceptable). The scores are arbitrary. Since you compare within a single session, they remain usable: you're sorting a bag of 20 ideas, and a sort can tolerate some arbitrariness.

Missing external factors (a serious limit). ICE ignores market opportunities, competitive threats and external shocks. Once your company has grown, you can add columns: estimated revenue in €, estimated cost in €, level of competitive risk. At the start, stick to the three basic columns.

ICE applied to your bottleneck

Go back to your bottleneck from section 3. Apply the +2 Impact bonus to every idea that targets it. Score 10 ideas, keep the best 3 to 5 (Quick Wins first), and launch them the following week.

Using the example from section 3 (activation in red for a founder just starting out), the list might look like the table below. Activation is the moment a new signup uses the product for real. Onboarding covers the first screens and messages that get them there.

Idea	I	C	E	ICE
Onboarding redesign with a guided tour	8 +2	6	4	20
Automated welcome email Day 0 · Day 3 · Day 7	7 +2	7	8	24 · Quick Win
90-second video on the landing page	5	6	7	18
Changing the call-to-action (CTA) button color	2	4	10	16
Personal call on Day 1 for every signup	6 +2	8	6	22

Idea 2 is the Quick Win of the week, idea 5 goes the week after, and idea 4 waits.

The same sort works outside software. Take a neighborhood florist whose bottleneck is repeat customers. A loyalty card, easy to launch and already seen at neighboring shops, will rank ahead of a custom mobile app that's heavy to build and never tested.

Your scored list at the end of section 4

A filled-in example for an online candle shop whose bottleneck is converting visitors into buyers:

• Abandoned cart email follow-up: 6 +2, 8, 9, so 25, Quick Win.
• Photos of lit candles on product pages: 6 +2, 7, 7, so 22.
• Samples sent to home decor Instagram accounts: 6, 5, 7, so 18, outside the bottleneck, keep it for later.

You have a method for deciding. What's left is choosing how to grow your business: that's the subject of section 5, on the three growth paths.

---

5. The three growth paths

In this section, you choose the type of growth you'll build over the next six months. You pick just one, because three half-finished efforts add up to nothing.

At this point, you have a funnel, a bottleneck and an ICE score: you know where to act. There are three types of growth. At the start, you'll want to run all three at once, and none of them gets anywhere.

One path at a time

Each path requires its own skills, tools and assumptions. Spending 30% of your time on each means seeing none of them through. Companies that succeed at the early stage (their first years) pick one main path and push it until it saturates. They open the second once the first brings in steady returns, like an annuity.

In the usual order, the three paths are virality, the sticky product and paid acquisition. None is better in absolute terms. The right one depends on your product, your stage and your ICP (Ideal Customer Profile, the portrait of the ideal customer you're targeting).

Path 1, virality through self-reinforcing loops

Virality is when your users bring in other users without you paying for it. It can emerge at any level of the funnel, including outside referral.

Loop type	Example	Levels crossed
Classic referral	Dropbox: invite a friend, get storage	Referral → Acquisition
Content / awareness	Respire: viral LinkedIn video, then a Monoprix deal	Awareness → Acquisition → Awareness
Embedded	Hotmail: "PS: I love you. Get your free email"	Usage → Awareness
Platform	YouTube: the player embedded on third-party sites	Usage → Acquisition
Cross-platform	Airbnb: listings auto-posted on Craigslist	Activation → Acquisition

Virality can pack two to three years of growth into six months. It is, however, hard to trigger and hard to sustain. You set up a mechanism that could go viral, then let it run, with no guaranteed outcome.

When to choose this path: your product makes people want to invite someone while using it (a collaborative, social or shareable tool). Or you produce content (media, app, community) and can create content that spreads on its own. A caterer whose every tray delivered to a company carries its name and an ordering QR code also fits this path: every guest at the lunch discovers the brand.

Avoid it if: your product is used alone and offers no occasion for sharing (a solo accounting analysis tool, a personal meditation app, an internal dashboard). In that case, virality won't happen, however hard you try.

Path 2, the sticky product

On this path, you grow through extreme retention: your customers stay. You build a product so useful that no customer leaves, and each new customer adds lastingly to your base. Churn is the share of customers who leave over a given period: a sticky product brings it close to zero.

Public examples: Facebook (your contacts and photos are there), Shopify (few serious competitors at its scale), ChatGPT (built into daily work habits since 2023). These companies don't have to chase new customers to replace the ones who leave: their users find it hard to walk away.

When to choose this path: you sell subscription software (SaaS) or an AI tool, your product fits into a user's daily work, and switching requires significant effort. You're playing the long game. Outside software, a weekly vegetable box subscription delivered to the same pickup point can follow this path, if customers plan their grocery shopping around it.

Avoid it if: you sell a one-time purchase (e-commerce without repeat buying), a one-off service (an agency on a single project) or an occasional-use service (event booking). In those cases, aim for another path.

Path 3, paid acquisition (CAC < LTV)

You buy customers, through advertising for example, for less than they bring you over their lifetime. The math is simple, the execution is hard.

Two acronyms sum up this math. CAC (customer acquisition cost) is what you spend on average to win a customer. LTV (lifetime value) is what a customer brings you on average, from their first purchase to their last.

The rule:

LTV / CAC ≥ 3

If you acquire a customer for €200 and they bring you €600 over 3 years, you can accelerate. The more customers you buy, the more gross margin you generate. Three constraints weigh on this path.

The platforms: ad costs go up when a competitor bids higher than you. Your positioning: your CTR (click-through rate on your ads) drops if your message no longer stands out. Your cash: you pay for acquisition in month M, and you collect the LTV from month M+1 to month M+36.

Public examples: Payfit ("it costs €560 to acquire a customer, who generates €1,000 in the first year"). Groupon at its peak ("a customer generates €90 in the first year, we want 500,000 new customers, budget = €45M").

Take an online candle shop. An Instagram ad costs it €15 per new buyer, and a buyer comes back three times on average, with €25 of margin per order. The LTV / CAC ratio is then 75 / 15, or 5: the shop meets the rule.

When to choose this path: your CAC is measurable, your LTV is predictable, your margins support the math, and you have cash or access to money (fundraising, financing, high net margin).

Avoid it if: your CAC exceeds your LTV (or remains too uncertain), you sell B2B very early with a 9-month sales cycle (you don't yet have enough data to judge), or you're short on cash. In that case, go back to paths 1 or 2.

The three-question test

Ask yourself these questions in this order, and answer honestly.

The question	If yes	If no
Does my product make people want to invite someone?	Path 1, virality, is in play	Move to question 2
Does it fit into daily life to the point of making switching painful?	Path 2, sticky, is in play	Move to question 3
Is my LTV at least 3× my CAC, and do I have cash?	Path 3, paid, is in play	None of the three is open to you yet: go back to the founder-led approach from chapter 3 (section 5), increase your manual volume and wait for the signals

Founder-led means sales driven by the founder personally: prospecting and demos done by their own hand.

You may have two paths in play at once. Still, choose ONE as your main path and give it 80% of your efforts for the semester, while the other waits its turn.

Three paths in parallel

You'll be tempted to tell yourself "I'll do all three". Run in parallel at the start, the three paths stay half-built:

• Your viral content doesn't take off, because you give it 2 hours a week instead of 10.
• Your product doesn't become sticky, since you build 3 average features instead of one excellent one.
• Your ad campaigns don't convert, for lack of time to tune them.

Six months later, no path works, because none got enough time to pay off.

Stacking two paths

Once one path is running, you can open the second, and the two reinforce each other.

With paid (path 3) on top of a sticky SaaS (path 2), your LTV rises because customers stay, and your paid acquisition math improves. With viral (path 1) on top of a sticky SaaS, your new customers don't leave, and your base keeps accumulating. By combining viral and paid, you lower your effective CAC, since some of your new customers arrive for free.

This stacking comes after a first path has succeeded.

Your path at the end of section 5

A filled-in example for a freelance graphic designer who creates visual identities for local shops:

• Main path: virality through content.
• Why: her clients don't buy again every month and her budget rules out advertising, but they display every logo in public.
• Number one lever: post each project as a before/after on LinkedIn and Instagram, with the client's approval.
• Path set aside: paid acquisition, for lack of a known LTV.

You have your path. Section 6 digs into virality, the least understood path and the most powerful when it works, with growth loops at every level of your funnel.

---

6. Growth loops at every level

In this section, you choose a growth loop to set up, meaning a mechanism where your current customers bring you new ones. With a loop, your growth depends less on your ad budget.

The funnel has a flaw: it reads left to right. You put awareness in (the fact that people know you), and you get revenue out. For more revenue, you need more awareness. This linear model is expensive and stops as soon as you stop feeding it.

In a loop, the output feeds the input. The more revenue you have, the more awareness you gain; the more users you activate, the more you acquire. This mechanism is called a growth loop. It's the difference between a company that grows once and a company whose growth compounds.

Loop versus funnel

With a linear funnel, your growth depends on the input tap: cut the advertising, and growth stops. With a loop, you open the tap once, then the loop runs on its own energy.

Four practical differences:

Funnel	Loop
Linear	Self-reinforcing
External tap	Output feeds input
Stable marginal cost	Decreasing marginal cost
Fragile as soon as you stop	Resilient once launched

No company that scales runs on a pure funnel: they all have at least one loop. A loop can exist at any level of the funnel, in addition to referral.

The five loop patterns

Most public cases fall into five patterns, meaning five models that recur from one company to another.

Pattern 1, classic referral. A user invites another and gets something in return. Dropbox made it the textbook case: "Invite a friend, get 500 MB of storage." Loop: Referral → Acquisition → Activation → Referral.

Pattern 2, viral awareness through content. A user publishes content about your product (or you publish it yourself), and lots of people discover your brand. Respire (zero-waste cosmetics) broke through on LinkedIn with a viral video that led to a Monoprix deal. Loop: Activation → Content → Awareness → Acquisition.

Pattern 3, embedded virality. In its normal use, your product shows your brand to third parties. Hotmail, in the 90s, gave the most copied example: every email sent from Hotmail ended with "PS: I love you. Get your free email at Hotmail." The recipient discovered Hotmail because the sender used it. Loop: Usage → Awareness → Acquisition.

Pattern 4, platform virality. Your product creates content that other sites embed on their pages, and those sites send traffic back to you. YouTube grew very fast because its video player could be embedded on every site on the web. Loop: Usage → Distribution on third-party platforms → Acquisition.

Pattern 5, cross-platform. You capture the audience of a third-party platform, legally or in a gray zone, and redirect it to you. In its early days, Airbnb automatically posted its listings on Craigslist, the big classifieds site of the time, and captured part of its audience that way. Loop: Activation → Cross-post → Acquisition.

Choosing the pattern for your product

Ask yourself three questions, in this order.

Question 1. Do your users already share an output of your product?

A Calendly link, a Loom video, a Figma design: users share them as part of normal use. A wedding photographer who delivers galleries online, with their name at the bottom of the page, is in the same situation. Choose pattern 3 (embedded) or pattern 4 (platform). Sharing already happens: you improve your brand's presence on what your users share.

Question 2. Does your product have a collaborative or social dimension?

A tool shared across a team, a product that works better with several people. Choose pattern 1 (referral) with an invitation mechanism. The user has a personal interest in inviting, because they benefit too.

Question 3. Your product has neither a shareable output nor a collaborative dimension.

Turn to pattern 2 (viral awareness through content). You produce the viral content yourself, founder-led (see chapter 3, section 6). This pattern takes longer to set up and more effort to sustain. It also works for products used alone (dashboards, analytics tools, personal apps).

A referral program bolted onto the product

You'll be tempted to add a referral program to a product that isn't suited to one: "Invite a friend, get 10% off." You'll spend weeks on it, for no result. A referral loop works under three conditions:

1. The user has a personal reason to invite (they gain something).
2. The invited person has a reason to come (the product already meets one of their needs).
3. Sharing is very easy (one click is enough, no form).

If your product doesn't meet all three, don't force the loop: change patterns.

Measuring a loop with the K-factor

The K-factor (or viral coefficient) measures how many new customers each customer brings you. It tells you whether your loop holds up over time.

K = Invitations per user × Invitation conversion rate

• K > 1: exponential growth without advertising (extremely rare, few products get there).
• K between 0.5 and 1: the loop doesn't sustain itself, but it lowers your effective CAC.
• K < 0.1: your loop isn't holding, revisit your pattern.

Example: out of 100 customers, 10% bring in a new one within 30 days. K = 0.1. The loop isn't enough on its own, but your CAC drops by 10% on new customers, which is already a gain. Above 0.3, your growth depends much less on paid advertising.

The single-loop mistake

A growing company rarely runs on a single loop. It has several, at different levels, which may or may not reinforce each other.

A typical B2B SaaS often runs three loops in parallel:

1. A content loop (founder-led, blog, podcast) that brings in awareness.
2. An onboarding loop (a new customer invites colleagues to use the product) that increases usage.
3. A product loop (each user action creates an output visible to a third party) that brings in new customers.

A local business can follow the same logic: Instagram content to get known, then a referral card slipped into every order.

Build one loop, stabilize it, then add the next. After a few cycles, your loops form a system that compounds your growth. Section 8 comes back to this in detail.

Your loop at the end of section 6

A filled-in example for a yoga class booking app:

• Pattern you can activate now: referral, with a free session for both the referrer and the invited friend.
• Lever surface: the confirmation screen after each booking.
• Target K-factor at 6 months: 0.3.
• Pattern number 2: embedded, with the studio's schedule built into its own website.

You have a diagnosed funnel (section 3), prioritized actions (section 4), a growth path (section 5) and a loop that can feed itself (section 6). Section 7 brings all of this together into a weekly ritual: the Growth Flywheel.

---

7. The weekly Flywheel, your ritual for getting it moving

In this section, you turn everything you've set up so far into a fixed two-hour meeting each week. Without this meeting, you manage two or three good weeks, miss one, and then never come back to it.

You already have the pieces. A funnel, the path that takes a stranger all the way to a purchase, level by level. An NSM (North Star Metric), the single number that guides your decisions. Prioritized ideas, a growth path and a loop.

This ritual is called the Growth Flywheel. A flywheel is a wheel that turns more easily with each rotation. You find this cycle in teams that grow fast. It repeats the same sequence every week, or every two weeks.

The full cycle

Gather Ideas → Prioritize (ICE) → Execute → Study Results → Decide → Identify Bottleneck → start again

Each turn of the flywheel closes one cycle and opens the next. You start on a Tuesday, finish the following Tuesday, and start over right away. The results come from keeping the rhythm week after week, with no spectacular trick or "growth hack".

The six stages of the flywheel

Stage 1: Gather Ideas. You collect the week's raw ideas, unfiltered, in a list of 10 to 30 lines in Notion or a Google Doc. They come from you, your cofounder, the team, customer feedback and your watch on competitors. At this stage, you don't cross anything out. For a neighborhood bakery, the list might include a loyalty card, an Instagram post about the Saturday batch, a partnership with the café across the street and pre-orders by text message.

Stage 2: Prioritize. You apply the ICE score from section 4 (Impact, Confidence, Ease). Spot the Quick Wins (ICE ≥ 24, Ease ≥ 7), and keep 3 to 5 actions. The rest leaves this week's cycle, but you keep it in a waiting list, the "ice box", for the following month.

Stage 3: Execute. You carry out the 3 to 5 chosen actions, and nothing else during the week. The brilliant idea from Thursday lunchtime goes on the list for the next cycle.

Stage 4: Study Results. At the end of the cycle, compare each result with the numerical target you had set (see the experiment template in section 9). Each action gets a marker:

Marker	Threshold
Succeeded	Beat the target
Promising	Within 10% of the target
Inconclusive	Below the target

Stage 5: Decide. For each action, you choose:

Decision	When you make it
Scale	It worked, you take it to a larger scale
Iterate	It almost worked, you adjust and try again
Kill	It didn't work, you stop and write down what you learned

Decide with a cool head. "We're keeping it because we spent time on it" doesn't count as an argument: the effort already spent changes nothing about the result you got.

Stage 6: Identify Bottleneck. Go back to your AARRR-A grid from section 3, the six levels of your funnel. Reassess the status of each level and spot the bottleneck, the level holding everything else back. It may have changed since the previous cycle, or stayed the same. You go back to stage 1 with this bottleneck in mind.

The concrete ritual: 2 hours on Tuesday morning

Tuesday morning covers four steps; execution fills the rest of the week.

Time slot	Step	What happens
Tuesday 10:00 - 10:30	Gather Ideas	Review the 10 to 30 raw ideas, add the week's new ones
Tuesday 10:30 - 11:00	Prioritize	Joint ICE scoring, pick 3 to 5 Quick Wins
Tuesday 11:00 - 11:30	Study Results	Review the previous cycle, Scale / Iterate / Kill decisions
Tuesday 11:30 - 12:00	Identify Bottleneck	AARRR-A review, declare the bottleneck for the new cycle
Wednesday → Monday	Execute	Carry out the 3 to 5 actions, nothing added mid-cycle

Two hours a week, not counting execution, is enough to keep the ritual running. That's less than the time spent in product meetings or sales reviews. Yet at the start, people often stop at "I have an idea, I'll do it". Without a ritual, your results depend on luck more than on your choices.

The self-discipline of the cycle

Three traps come up often.

Trap 1: skipping a week. An urgent client, a trip or a meeting makes you miss a Tuesday. You skip the week, then the next one, and three weeks later the ritual is dead. The rule: if you miss your Tuesday 10:00, you do the ritual Wednesday at 10:00, without waiting for the following week.

Trap 2: adding actions mid-week. "Brilliant idea on Thursday at lunch, let's launch it right now." No: you write it down in your document, and you'll score it in the next cycle. With this discipline, you run a growth program instead of lurching from one disguised emergency to the next.

Trap 3: lying to yourself about results. "It's too early to tell." If you set a numerical target (section 9), compare the result with it and decide. A result read honestly helps you improve, even when it disappoints.

The ritual by stage

From 0 to 6 months (the very beginning): weekly ritual, 2 hours on Tuesday. Your cofounder, if you have one, takes part. You run 3 to 5 actions per cycle.

From 6 to 18 months (first team): ritual every two weeks, 3 hours. You invite your first growth or marketing hire. You run 5 to 10 actions per cycle.

Beyond 18 months: monthly 4-hour ritual, in a review-meeting format. The whole team that touches the funnel takes part. You run 10 to 20 actions per cycle, with a quarterly roadmap alongside.

Whatever your stage, never drop below a monthly ritual. If you can't even hold one meeting a month, growth isn't among your priorities, and that's what needs to move back to the top of the list.

The over-tooling trap

You'll be tempted to add tools: Notion for ideas, Linear for execution, Mixpanel for results, a Google Sheet for the ICE score, a Slack channel for announcements. That's five tools, and three months later you're using none of them.

The rule: one tool to start, two at most. A Notion page or a Google Doc holds the ideas, the ICE scores, the results and the bottleneck. A Google Sheet or Mixpanel tracks the metrics. The day a third tool seems necessary, wait six more months: the need often disappears in the meantime.

Your ritual, at the end of section 7

A filled-in example for an online florist who runs her shop alone:

• Weekly slot: Monday 9:00, before preparing orders
• Participants: just her
• Single document: a Google Doc titled "Flywheel"
• Actions per cycle: 3

In section 8, you'll see why the ritual produces far more than the sum of its actions: that's the compound effect.

---

8. The compound effect: why multi-level growth beats acquisition alone

In this section, you compare, with numbers, three ways to grow the same store over 12 months. The goal: understand why growth that works several levels of the funnel pays off far more than acquisition alone, and turn that into a rule for your ritual.

You may be wondering whether 3 to 5 actions a week are worth the effort, compared with "I'll put everything into ads and see".

The starting point: a simple e-commerce store

Let's start from an online store selling products at €30 on average. The three scenarios below start from the same level and organize their growth differently.

Scenario 1: acquisition only (linear, fragile)

You pay for online advertising, "ads". You get 100 new customers a month, who each spend €30. None buys again, none brings in anyone else. In the table, LTV (lifetime value) means what a customer brings you in total, over their entire relationship with you.

Scenario 1	Value
Monthly revenue	100 customers × €30 = €3,000
Annual revenue	€36,000
LTV per customer	€30, one purchase and then they leave
Risk	Total dependence on ads

Your ads can stop working: an algorithm change, rising CPMs (the price paid per thousand impressions), audience fatigue with the same visuals. If that happens, your revenue stops dead the following month, because no customer base keeps buying.

Scenario 2: growing acquisition (linear but rising)

Every month, you reinvest 100% of the margin into more ads. You win 100 customers the first month, 110 the second, 125 the third, and so on. Growth stays linear, but it's rising.

Scenario 2	Value
Customers over the year	~1,860
Annual revenue	~€55,800
LTV per customer	€30, still a single purchase
Risk	Still dependent on ads, with more volume

It's better than scenario 1, but you remain dependent. Each customer also costs you a little more as ad auctions heat up, and each euro of advertising brings you less and less.

Scenario 3: acquisition + retention + referral (exponential)

Same ad budget as scenario 2, with two extra mechanisms.

• Retention (keeping your customers): 1/3 of a quarter's customers buy again the following quarter. You get there with automated emails, content sent after purchase and a quality product.
• Referral (recommendation): 10% of new customers bring in a new customer within the month, through word of mouth triggered by the product and a referral offer.

Modeled over 12 months, scenario 3 gives the following numbers.

Scenario 3	Value
Customers over the year	~1,860 acquired through ads, plus those brought in by retention and referral
Cumulative annual revenue	~€96,981
Effective LTV per ad-acquired customer	96,981 / 1,860 = €52, versus €30 in scenarios 1 and 2

With the same ad budget, you make almost twice the revenue of scenario 2. The gap comes from repeat purchases and word of mouth. Your dependence on ads also drops: by month twelve, more than 50% of revenue comes from repeat purchases and referrals.

The math holds outside e-commerce too. For subscription software, retention means renewals. For a plumber, it means customers who call back, and referral means the neighbors they send his way.

Why the gap widens over time

The compound effect follows an exponential curve. With each cycle:

• last month's customers buy again, with no new acquisition cost;
• they also bring in new customers, again with no acquisition cost;
• average LTV rises, which lets you bid higher on ads.

The longer you wait, the wider the gap. Over 12 months, it reaches 2x. Over 24 months, with retention that holds, it can climb to 4x or 5x. The mechanism resembles compound interest on an investment, applied to a customer base. Companies that grow sustainably never rely on paid ads alone.

The structural competitive advantage

With an LTV of €52 in scenario 3, versus €30 in scenario 1, you can pay up to €52 of CAC on ad platforms. CAC (customer acquisition cost) is what you spend to win a customer. A competitor stuck in scenario 1 can't go above €30, and you beat them in the auctions.

Who	Bid ceiling per customer
You, scenario 3	€52
Your competitor, scenario 1	€30, beyond that they lose money

You get the best placements, you show up first on the most contested searches, and you attract the best creators for your influencer partnerships. You run acquisition and retention side by side, and retention funds higher acquisition bids.

This advantage is structural: working harder won't close it. A competitor chasing you first has to build their own retention and referral machine, and meanwhile you pull further ahead.

What this means for your action list

The practical rule: never put 100% of your actions on a single level of the funnel.

At the start, you'll be tempted to bet everything on acquisition, because it feels the most urgent. You'd then be replaying scenario 1. Instead, split each cycle like this:

Share of the cycle	Where it goes
40%	Your bottleneck, at any level
20%	Retention, always, even when it's not the bottleneck
20%	Referral, always, even when it's not the bottleneck
20%	Your main path from section 5

These percentages are guidelines. With 5 actions per cycle, that gives, for example, two actions on the bottleneck, one on retention, one on referral and one on your main path. The principle doesn't change: every cycle, you put at least one action on retention and at least one on referral, even when neither is your bottleneck. These are the only levels whose gains compound without limit.

The most common mistake

At the start, people often put 80% of their effort on acquisition, 15% on activation and 5% on retention. Referral doesn't even make the top ten priorities. After 12 months, the store plateaus at €36,000 in revenue, with rising CAC, and its founder concludes that "marketing doesn't work". The marketing was working; the growth rested on a single level, and that's what made it fragile.

The opposite isn't any better: betting everything on retention without acquisition produces no growth, since you have no one to retain. The sensible order: acquisition first, to get volume, then retention and referral in parallel as soon as you have 50 to 100 customers.

Your compounding, at the end of section 8

A filled-in example for a neighborhood yoga studio (fictional numbers):

• Current scenario: 2, almost all new students come from Instagram ads
• Share of actions on retention: 0%, on referral: 0%
• Estimated LTV: €90, estimated CAC: €40

You can see why the weekly ritual from section 7 has to include actions on several levels. In section 9, you'll formalize how to document each of these actions: the experiment template.

---

9. Experiment template: Gas, OMTM, Objective, Result

In this section, you adopt the experiment template, a one-page sheet for documenting each action. With this format, every test teaches you something, even when it fails. It's the final piece after the funnel, the bottleneck, the prioritized ideas, the path, the loops, the ritual and the case for compounding.

Without this template, you launch actions, you get fuzzy results, and you don't know whether you learned anything. With it, each action becomes a short sheet: who you're testing, what, where, how, how much you're putting in, what you're aiming for and what you got. After 6 months, you have a file of 30 documented experiments, and you can draw patterns from it.

The four variables of a test

Every marketing experiment varies at least one of these four variables.

Variable	The question
Target	Who?
Message	What?
Channel	Where?
Format	How?

Target and Message are about substance. Channel and Format are about form, the container. Ideally, you test only one variable at a time, otherwise you don't know which one produced the effect. In practice, at the start, people often test two at once to move faster, accepting less rigorous attribution.

A meditation app that keeps its message but moves from Instagram to TikTok is testing only the channel. If it also changes its target, it won't know which change made the difference.

The four conditions of the experiment

Every experiment rests on four conditions, which you set before launch.

Gas (the fuel). The resources you put into it. For ads, it's the budget; for emails, the number of contacts targeted; for a page, the number of visitors allocated to the test. The more gas you put in, the more statistically reliable your result, and the more resources you use up.

OMTM (One Metric That Matters). The single metric that will say whether the experiment succeeded or failed. In exploitation mode, you pick a business metric: revenue, sales, leads (interested contacts). In progression mode, you pick a marketing metric: CPA (cost per action), CVR (conversion rate) or CTR (click-through rate). Just one. If two metrics seem equally important, you haven't yet decided what you want.

Objective. The numerical value you're aiming for on the OMTM. You set it before launch, using your Growth Model from section 2, the equation that links your funnel numbers to your revenue: you change a variable and look at the impact. The objective takes the form of a number you'll compare with the result, no discussion, and never a "it would be nice if…".

Result. The number you get at the end, marked in one of three states:

Marker	Threshold
Succeeded	Beats the objective
Promising	Within 10% of the objective
Inconclusive	Below the objective

Update the Result every week during the experiment, so you can adjust along the way instead of waiting for the end.

A common early mistake: setting the objective after seeing the first numbers, or lowering it along the way so the test "succeeds". Write the objective on the sheet before spending the first euro, and don't touch it until the verdict.

The hypothesis format

Before launching an experiment, write the hypothesis in this format:

This format is the standard in growth. It imposes three requirements on you:

1. Name the action precisely: "make the call-to-action (CTA) button at the top of the homepage orange" rather than "improve the site".
2. Name the expected result: "+15% conversion rate on page X" rather than "more conversions".
3. Give a data-backed reason: "because the heatmap (the map of clicks on the page) shows visitors don't see the current button" rather than "I think that".

Without a written hypothesis, you change things at random; with one, each test answers a precise question.

For a new accounting firm, it might read: "We believe that calling each prospect within an hour of their quote request will produce 2 more signed clients per month, because prospects we call back the next day often tell us they've already chosen another firm."

A complete example, from brief to result

Take a cosmetics brand sold online, testing placements with Instagram micro-influencers.

The whole experiment fits on one page, which you can copy and paste for your own.

Two families of OMTM: rates and volumes

To choose your OMTM, ask yourself whether you're in progression mode (improving efficiency) or exploitation mode (increasing volume).

Mode	Metrics	When you use them
Progression, the rates (causes)	CPA, CAC, CVR, CTR, activation rate, signup rate	When you're optimizing the machine
Exploitation, the volumes (consequences)	Revenue, sales, leads, customers, LTV	When you're growing the machine

At the start, you'll alternate. One experiment in two is about progression: you improve a page's conversion rate or a campaign's click-through rate. The other is about exploitation: you aim for 50 leads this month, or 750 sales this quarter. The classic mistake is optimizing rates without ever looking at volumes, or the reverse. Run both in parallel.

The three verdicts: Scale, Iterate, Kill

At the end of the experiment, you decide what comes next.

Succeeded → Scale. You beat the objective. You go bigger: more gas, a new cycle, and a process built around the action.

Promising → Iterate. You're within 10% of the objective. Change one variable (target, message, channel or format) and relaunch. You keep going, without scaling yet.

Inconclusive → Kill. You're below the objective. Stop and write down what you learned. Don't rerun the same test in three months "just to see": if your hypothesis hasn't changed, you'll get the same result.

Kill takes the most discipline. At the start, people often keep dead experiments alive because they spent time on them. This trap is called sunk cost: the cost already incurred that you refuse to lose. The weekly ritual from section 7 and the Scale / Iterate / Kill decision from section 9 protect you from it.

The 6-month experiment archive

After 6 months of the ritual, you've documented 50 to 80 experiments, at 2 to 5 per weekly cycle. Set aside half a day to review them.

For each experiment, note:

A filled-in line for a peer-to-peer equipment rental platform: Activation level, Message variable, Kill verdict, learning "promising zero deposit didn't push owners to post more listings".

You end up with a dataset. You can see which variables produce the most Scale verdicts, which levels resist you the most, which learnings keep coming back. This meta-analysis is well worth the 6 months of ritual.

Takeaways from chapter 4

CH4 conclusion

You have all nine pieces: an NSM that guides your decisions, a Growth Model that simulates, a diagnosed funnel, a prioritization method, a chosen growth path, an active loop, a weekly ritual, the case for working several levels and an experiment template.

Together, they form a machine. Other people can run it day to day, because you've documented everything, and its gains compound over time, because you work several levels. You've left behind the survival mode of CH1, the handcrafted selling of CH2 and the tactical marketing of CH3: you're building a system.

This system can keep the same shape for years. Two things will change: the team running it (you alone, then with a hire, then with a team), and the thresholds, which you'll adapt to your stage. You don't manage 50, 500 or 5,000 customers with the same ICE scores.

---

CH4 . Your engine repeats and compounds

You no longer do marketing piecemeal. Every week, you choose your actions based on a number. You have levels with statuses, ideas scored with ICE and loops that feed the top of your funnel. The question "what are we trying this week?" has become "where is our bottleneck, and what are we trying on it?".

This is the last step of the guide: you move from shots taken at random to a system you repeat.

The end-of-chapter deliverable

By the end of this chapter, you should have:

• An explicit North Star Metric: a cause, upstream of revenue, that you can write on a whiteboard in five words.
• A Growth Model: your funnel × rates equation, written down. You know which variable to move, and how much you stand to gain.
• A diagnosed AARRR-A: six levels with color statuses, the bottleneck identified, and the conviction to tackle it before anything else.
• An ICE backlog, your bank of scored ideas, fed every week: 10 to 30 raw ideas, scored, with 3 to 5 launched per cycle.
• One main growth path chosen among virality, sticky (retention) and paid (paid advertising). A single one, which you stick to: if you run three at 30% each, you dilute your efforts.
• An active growth loop: referral, content virality, embedded virality, platform virality or cross-platform. It turns the output of your funnel into new input.
• A ritualized weekly Flywheel: two hours on Tuesday morning, in six steps (gather, prioritize, execute, measure, decide, rediagnose).
• Efforts spread across several levels: your attention is split among acquisition, activation, retention and referral according to what your model says, far from 80% on acquisition.
• Experiment templates: one page per test, four variables, four conditions, a written hypothesis, a verdict.

If you can check these nine boxes, you have an engine: something that produces revenue while you work on the next level.

Two weeks of setup, zero experiments

You'll be tempted to measure everything: install Amplitude, GA4, Hotjar and Mixpanel, then spend two weeks configuring tracking for every click. Meanwhile, you won't have launched a single new experiment.

Three or four well-chosen numbers are enough to move forward; a forty-metric dashboard adds nothing at the start. A weekly Flywheel you actually keep teaches you more than perfect tracking installed too late. If you can't say in two sentences "our bottleneck is this level, and on Tuesday we're trying this action on it", you're tracking too many metrics and not making enough decisions.

After the guide

The guide stops here because what comes next is execution. You'll run the Flywheel fifty times and kill ideas that seemed brilliant. You'll see a channel explode, then cool off, and another take over. You'll go through quarters where nothing moves, then a month where everything compounds at once.

These swings are normal: compounding gains build up for a long time before they show.

At this point, you have the method. What you need is time, and conversations with other founders who have been through the same sequence.

That's what we try to offer at swanbase.

For the short definition and the comparison with traditional marketing and growth hacking, read Growth marketing: definition and comparisons.