Where the flywheel idea comes from
The flywheel as a business metaphor comes from Jim Collins' work on how good companies become great ones, summarised on the source page cited below. His argument is that no single push produces a breakthrough. A company turns a heavy wheel one push at a time, always in the same direction, and the turns accumulate until the momentum does much of the work. The companies he describes could state their wheel as a short sequence of steps in which each step makes the next one almost inevitable.
Marketing borrowed the metaphor to describe the customer relationship as a cycle rather than a pipe. The version most marketers know, attract, engage and delight, keeps the two ideas that matter from Collins: the wheel stores energy, and the energy comes from consistent pushes in one direction rather than from occasional campaigns.
As a thinking tool, the framework makes the contribution of existing customers to new growth visible. What it hides, when used carelessly, is the sequence of decisions a buyer actually goes through before the first purchase. That sequence is the funnel's job, and the two models are complements rather than rivals.
The marketing flywheel
The marketing flywheel has three stages, and the output of each is the input to the next. Attract brings strangers into contact with the brand; engage turns contact into a first purchase or commitment; delight turns a customer into someone who buys again and tells others, which feeds attract. The table shows what each stage contains and the signals that tell you whether it is adding or losing speed.
| Stage | What happens | Who does most of the work | Signals it is turning |
|---|---|---|---|
| Attract | Strangers discover the brand through content, search, advertising, reviews and referrals | Marketing, with customers contributing referrals and reviews | New visitors, branded search demand, referral share of new customers |
| Engage | Prospects evaluate, try and buy; the first promise is kept | Sales, product, onboarding and service | CVRShare of visits or clicks that convert. on trial or purchase, activation, time to first value |
| Delight | Customers get the outcome they wanted, return and recommend | Product, service, community and lifecycle marketing | Repeat rate, renewal, reviews, referrals, LTVTotal value a customer brings over time. |
The point of drawing the stages as a circle is that they never end. A customer who is delighted re-enters attract on behalf of the brand, and a customer who is let down at engage never reaches delight, which means the wheel loses a push it would otherwise have gained for free. Every stage also has a cost of failure that lands in a different stage, which is why the framework is useful for cross-team conversations: the marketing team's acquisition results depend partly on what service and product did last quarter.
Force and friction
The framework's working vocabulary is force and friction. Force is anything that speeds the wheel: a referral programmeA scheme that rewards existing customers, and often the friends they invite, when a recommendation brings in a new customer., a product that is easy to recommend, content that customers share, onboarding that gets people to their first result quickly, pricing that is simple to explain. Friction is anything that slows it: a slow checkout, long support queues, confusing renewal terms, a product that works but is awkward to describe, a handover between sales and service where context is lost.
The practical exercise is to list forces and frictions by stage, then ask where marketing can act. Marketing adds force mostly at attract, through demand creation, and can remove friction at engage by matching the message to the experience that follows the click. At delight, marketing usually amplifies what product and service deliver rather than creating it; lifecycle programmes, covered in the lifecycle marketing guide, are the main lever it owns there. The honest result of the exercise is often that the biggest friction sits outside marketing's control, and the framework's value is that it says so plainly.
A useful discipline is to pair every force with the friction it is meant to overcome, and to assign a KPIA metric tied directly to an objective. to the pair. A referral programme is a force; a cumbersome referral flow is the friction that cancels it; the metric is referred customers as a share of new customers. Without the pairing, teams add forces (more campaigns) when removing a friction (a shorter form) would turn the wheel faster at lower cost. The tutorial on designing a referral programme builds that force step by step, with a sharing flow simple enough not to become the friction.
Worked flywheel
The example applies the force-and-friction exercise to a subscription business. The figures are invented; what matters is where each push lands and which stage it moves.
A meal-kit subscription maps its flywheel
A meal-kit service attracts about 12,000 new visitors a month: 40% from paid social, 35% from search and 25% from referral links sent by customers. At engage, 3% of visitors start a trial box, around 360 trials. At delight, 55% of trial customers are still subscribed after four weeks, and each retained customer refers on average 0.4 new customers within six months. The team lists frictions: a sign-up form with eleven fields, delivery slots on weekdays only, and a trial cancellation flow that takes four screens. They list forces: a referral credit of £15, recipe cards designed to be photographed and a weekly email with the next menu. The model they build says that cutting the form to five fields and adding weekend delivery could lift trial starts to 3.6% and four-week retention to 60%; if that happens, the referral share of new visitors rises from 25% to roughly 31% within two quarters, and blended acquisition cost falls from £42 to about £36 per retained customer. The numbers are invented to show the arithmetic; the lesson is that two frictions removed at engage did more for attract than any planned increase in paid social budget.
Flywheel, funnel and loops
Three models describe the same business from different angles. The marketing funnel framework is a sequence with a start and an end; it is the right tool for finding where prospects drop out before a first purchase, and it has nothing to say about what happens after. The flywheel is a cycle; it is the right tool for seeing how existing customers feed new demand, and it is vague about the steps inside each stage. The growth loops framework sits between the two: it describes specific, measurable loops (a referral loop, a content loop, a paid loop that reinvests revenue) with explicit inputs, actions and outputs, where the flywheel describes the whole business as one loose cycle.
In practice the funnel is the diagnostic, the flywheel is the narrative that aligns teams, and loops are the engineering. A team that only uses the flywheel tends to produce inspiring diagrams and few decisions; a team that only uses the funnel tends to treat retention as somebody else's problem. The comparison of the marketing funnel and the flywheel sets out which planning questions each model answers and how to use both in one plan. The growth marketing fundamentals hub sets out how the three fit a growth programme, and the retention guide for e-commerce covers the delight stage in the detail a flywheel diagram cannot.
Limits
The flywheel is a metaphor, and metaphors hide as much as they show. Momentum is not a metric; it has to be approximated with proxies such as repeat rate, referral share and the trend in acquisition cost, and those proxies move slowly, so the framework is poor at judging a single campaign or a single month. It assumes that delight produces referrals, which is true in categories people talk about and far less true in categories they do not. It is weak for businesses with long purchase intervals or formal procurement, where a customer may be perfectly satisfied and still produce no measurable push for years.
It is also easy to use as a slogan. Declaring that the company "runs on a flywheel" changes nothing unless the forces and frictions are listed, owned and measured. And because the framework blurs stages into a circle, it can obscure the hard, sequential work of converting a stranger into a first-time buyer; that is where the funnel and the growth loops frameworks linked above should take over.
When a rewards programme sits on the wheel, the guide to using the flywheel for customer loyalty shows how to tell rewards that add force from rewards that only pay for existing habits.
Common mistakes
- Treating it as a reporting structure. The flywheel aligns teams; it does not replace a funnel report or a KPI tree.
- Adding force where friction is the problem. More campaigns at attract cannot compensate for a trial flow that leaks at engage.
- Assuming delight is automatic. A customer who got what they paid for is satisfied, not delighted; referrals need a reason and a mechanism.
- Ignoring purchase frequency. In a low-frequency category the wheel barely turns, and a journey map is the more honest model.
- Measuring it weekly. Momentum proxies move over quarters; weekly swings are noise.
- Drawing the circle and stopping. The value is in the lists of forces, frictions, owners and metrics, not the diagram.
