AARRR, the flywheel, North Star, ICE and RICE

AARRR is Dave McClure’s five stages of growth: acquisition, activation, retention, referral, revenue. A North Star is the main metric; ICE and RICE rank ideas.

From stages to a decision

In the last lesson you laid the path out in stages. Now three things are needed: what to count as the main thing (one metric), how to look at growth by stages (AARRR and the flywheel) and what to do first (prioritizing). They are linked: the main metric says where to go, the stages show where it hurts, the priorities choose what to fix first. All the numbers in this lesson are approximate: you’ll get real ones from a counter in the part on metrics, and for now it matters more to learn to ask the right questions.

AARRR: five stages of growth

Where from. The American investor and entrepreneur Dave McClure gave a five-minute talk, “Startup Metrics for Pirates”, at the Ignite conference in Seattle in 2007. The name is a pun: “AARRR” sounds like a pirate’s cry. They are five stages at each of which a startup needs to measure something:

1. Acquisition. Where people come from: search, social media, recommendations, ads.
2. Activation. Whether the person got a first benefit: a first “happy visit”. For a site this is usually the first useful action: opened the prices, read the answer to a question, began filling in the form.
3. Retention. Whether they come back.
4. Referral. Whether they advise others.
5. Revenue. Whether they pay.

The order differs slightly with different authors (sometimes Revenue goes before Referral): what matters more is that each stage has a metric of its own, not just one, “visitors”.

An example for a project with services (the invented Kate Alder). This is a service people buy once in a few years, so the Retention and Referral stages look different than for an app.

• Acquisition: visitors from search per month.
• Activation: opened the page with prices and terms.
• Retention: for a service, rarely “came back to the site”; more often “got in touch again” or “stayed in touch” (signed up for useful reminders).
• Referral: got in touch on someone’s recommendation. Measured with one question in the form: “How did you hear of us?”
• Revenue: paid.

Where it lets you down.

• It was devised for startups with an app. For a site with a single one-off service, Retention and Referral are harder to measure, and Referral is often the most important stage.
• Vanity metrics. Views and “likes” will grow easily and nothing will change. Ask: would a decision change if the number were different?
• Numbers don’t give causes. AARRR shows where it sagged, not why.
• A pun doesn’t replace thought. Five letters are easy to remember and easy to take for a ready analysis.

The flywheel

Where from. The image of a flywheel (a heavy wheel that is hard to get turning but easy to keep turning) was brought into business by the consultant Jim Collins in Good to Great (2001). Amazon described its own flywheel at about the same time. In 2018 the company HubSpot replaced its funnel with a flywheel of three parts: attract, engage, delight.

How it is built. It’s a closed chain in which the result of one stage feeds the next and returns to the start. A happy customer recommends, people come to you who are easier to trust, they ask for discounts less often, the work goes more smoothly, customers are happier. Each turn is a little easier. And friction (a poor reply, a long wait) slows the wheel.

An example. “A person finds a clear page with prices, makes a request, gets an explained result and recommends us to acquaintances; new people arrive who need nothing proved to them.”

Where it lets you down. A flywheel is a metaphor, not a model: it’s hard to measure and easy to draw out of wishes. Check each link: does a happy customer really recommend? Ask. For services bought once in a lifetime the cycle is weak, and it’s more honest to write “so far it doesn’t close”.

The North Star Metric

Where from. The “North Star” metric: one metric that best reflects the value you bring to customers. The term was especially established in growth circles and by the company Amplitude, whose “North Star Playbook” is in the sources; a close idea of “the one metric that matters” is in Lean Analytics by Alistair Croll and Benjamin Yoskovitz (2013).

What it should be like.

• About the customer’s benefit, not your revenue: revenue comes later and confirms that there’s a benefit.
• A leading indicator: it grows earlier than the money.
• Understandable to the whole team and measurable with a tool, not a “feeling”.
• Specific: the action, the period and the audience are named.

For a consultation site: “the number of consultation requests per month from landowners”. That names the action from the brief, the period and the right audience. Bad candidates: “happy customers” (it isn’t clear how to measure, and it comes late), “visitors” (they grow without any benefit), “revenue” (late, and says nothing about the cause).

Where it lets you down, and this matters. The British economist Charles Goodhart formulated a rule often retold as: when a measure becomes a target, it stops measuring well. If you reward yourself for the number of requests, you’ll get spam requests and “anything so long as they leave a phone number”. So beside the North Star you put quality control metrics: what share of requests became customers, how many requests came from the right segment. One metric doesn’t replace the others; it only leads.

Check yourself: why is “the number of consultation requests per month” better than “happy customers”?

It can be counted by a counter or a form, it grows earlier than revenue, and it names the action, the period and the audience. “Happy customers” can’t be measured without surveys, and one learns of them late.

ICE and RICE: what to do first

Where from. You always have more ideas than strength. To choose among them you score them.

• ICE is associated with Sean Ellis, the author of the idea of “growth through experiments” (growth hacking): Impact, Confidence, Ease, each from 1 to 10; the result is the sum or the average. It’s handy for quick experiments.
• RICE was proposed by Sean McBride at Intercom in 2016: Reach (how many people in a period), Impact (0.25, 0.5, 1, 2, 3), Confidence (as a percentage) and Effort (in person-months). The formula: reach × impact × confidence ÷ effort.

The difference is in two things: RICE counts reach (how many people it will touch) and effort as numbers. So an idea with a large reach and little work rises in RICE, and may stay second in ICE. Try it in the diagram: change the reach or the effort and see how the rankings come apart.

• Idea
  ICE, from 1 to 10

  Impact Confidence Ease

  RICE

  Reach, people a month Impact 0.250.5123 Confidence, % Effort, person-months
• Idea
  ICE, from 1 to 10

  Impact Confidence Ease

  RICE

  Reach, people a month Impact 0.250.5123 Confidence, % Effort, person-months
• Idea
  ICE, from 1 to 10

  Impact Confidence Ease

  RICE

  Reach, people a month Impact 0.250.5123 Confidence, % Effort, person-months
• Idea
  ICE, from 1 to 10

  Impact Confidence Ease

  RICE

  Reach, people a month Impact 0.250.5123 Confidence, % Effort, person-months
• Idea
  ICE, from 1 to 10

  Impact Confidence Ease

  RICE

  Reach, people a month Impact 0.250.5123 Confidence, % Effort, person-months

By ICE

1. A page with prices and timing: 7.7
2. An email to customers after a repair, asking for a review: 7.3
3. Booking a repair through a messenger: 6
4. An article, “How to check your own brakes”: 5.3
5. An ad campaign on social media: 4.7

By RICE

1. A page with prices and timing: 1280
2. An ad campaign on social media: 375
3. An email to customers after a repair, asking for a review: 256
4. Booking a repair through a messenger: 200
5. An article, “How to check your own brakes”: 150

The idea “An ad campaign on social media” was number 5 by ICE and became number 2 by RICE: RICE counts the reach (1500 people) and the effort (2 person-months), and ICE doesn’t see them.

All the numbers are guesses: Confidence is there so that you can honestly say “I’m not sure”. Scores help you argue about priorities, but they do not replace the decision.

Where it lets you down.

• They are guesses. The number “impact 7” is the author’s feeling. That’s why both schemes have confidence: so that you can honestly say “I don’t know”.
• Scores can be bent. The team wants to do one thing and gives it high scores. Make your first scoring before you know what you want.
• No “harm”. Neither scheme has the question “aren’t we deceiving someone?”. The idea “put in an urgency timer” can score many points, but it’s a dark pattern (earlier lessons).
• Scores don’t replace conversation. Their value is that they make you state your assumptions aloud and argue about them. The decision stays with you.

Counting without spying

One more word before the numbers come in. To count visitors you need a counter, and a counter usually means cookies or a script that sees your visitors’ IP addresses. In the EU and the UK the ePrivacy Directive and PECR ask for consent before cookies that aren’t strictly necessary, and the GDPR treats an IP address as personal data. In the US, state privacy laws such as California’s ask for notice and an opt-out for some uses. That’s why a growing number of small sites use privacy-friendly counters (Plausible, Fathom, Matomo configured without cookies), while Google Analytics 4 is the common default. In Russia the usual counter is Yandex Metrica. The part on quality and the law returns to this. This is general information, not legal advice: measure as much as a decision needs, tell people what you measure, and have the setup checked.

How it all fits together

You can tell the same story with these tools in order: the North Star says what we want to achieve; AARRR shows at which stage it hurts; the flywheel reminds us that the customer after the purchase is part of growth too; ICE and RICE choose what to fix first. For a small site that’s enough. You need neither ten metrics nor dashboards: one main metric, five stages, one plan.

Practice

Build four things for your project.

1. A North Star in one phrase: what we count, for what period, from whom. The name of the action from the brief, in the same words.
2. AARRR. For each of the five stages: what we count, an estimated number and how to measure it (for example a visit counter, a form, an orders spreadsheet). If the right counter doesn’t exist, write which you’ll connect.
3. The flywheel. Four links, each strengthening the next.
4. Five growth ideas. Score them by ICE and by RICE and enter the numbers. Below, in one phrase, explain how the rankings differed.

The check looks at form: the action and the period in the North Star, numbers in the tables, ways of measuring, four links of the flywheel. The part on metrics will show whether your numbers are right. For now it matters more to write honestly that they are guesses, and to say how you’ll test them.