Review
Growth was suddenly half my job and I wanted to know whether there was a method under the folklore. This is the closest thing to one, and its real contribution is procedural rather than tactical.
The argument that matters is the order of operations. You don't run growth experiments until you have evidence that people would be upset if the product disappeared. Teams skip that step constantly, then spend a year optimising the top of a funnel that leaks at the bottom.
My honest caveat is that the operating model assumes conditions most companies don't have: clean instrumentation, a cross-functional team that can ship without negotiating for engineering time, and an executive who'll tolerate a high failure rate. Without those three you get the vocabulary and none of the results, and the book is light on how to get them.
The weekly cadence is the part I actually kept. A standing meeting where you look at one metric, pick the next test, and kill the losers is unglamorous and it works. Read it if you're setting up a growth function. If you want a list of tricks you'll be disappointed, and that's to the book's credit.
Key Takeaways
The parts worth keeping:
Earn the right to grow
- Before any growth work, test whether the product is a must-have. The survey question is blunt: how would people feel if they could no longer use it? If a large share wouldn't be disappointed, growth spend is wasted.
- Find the moment where a new user first understands the value, then measure how many reach it and how fast. Almost every retention problem is really people never arriving there.
- Growth can't fix a product nobody misses. It only accelerates whatever is already true.
The funnel, and the question each stage answers
| Stage | The question |
|---|---|
| Acquisition | Can we reach the right people at a cost that works? |
| Activation | Do they reach the moment of value, quickly? |
| Retention | Do they come back without being prompted? |
| Revenue | Does the money follow the value, and at what point? |
| Referral | Does using it create a reason to tell someone? |
- Work the stages in order of leakage, not in order of interest. The most interesting stage is rarely the one costing you the most.
The team and the cadence
- The growth team is cross-functional by design: product, engineering, data and marketing in one group, with the authority to ship. A growth team that has to file tickets is a suggestion box.
- Run a fixed weekly loop: look at the data, generate ideas, prioritise, test, review. The cadence matters more than any individual idea.
- Expect most tests to fail. The value is the rate of learning, so the number to protect is tests shipped per week, not win rate.
- One metric owns the room. Pick the number that best expresses delivered value and make every experiment argue its case against it.