Unit economics: how to know if each customer is profitable
Learn unit economics in practice: how to calculate CAC, contribution margin, LTV and payback, with a worked example, common mistakes and ways to improve.
Growth in practice: how to choose, prioritize and run experiments that move revenue, with a hypothesis template, the ICE method, test ideas and common pitfalls.

Growth, in practice, means running short, measured experiments at points in the funnel that affect revenue, such as checkout conversion, new customer activation, pricing and retention. Each experiment starts from a written hypothesis, has a success metric defined before it begins and ends in a decision: keep, discard or adjust. Results come from a steady testing cadence, not from one brilliant idea.
The most common mistake is starting with an idea someone saw at another company. The right starting point is the funnel. A widely used breakdown is AARRR: acquisition, activation, revenue, retention and referral.
Put numbers on each stage and look for the biggest leak close to the money. A hypothetical example: 10,000 people visit the offer page, 600 start checkout and 300 pay. Half of the people who decided to buy drop out along the way. That is the first candidate for an experiment, because any gain there turns straight into revenue without needing more traffic.
Improvements at the top of the funnel matter too, but their effect gets diluted in the stages that follow. The closer you are to payment, the faster a test shows revenue impact.
A useful hypothesis fits in one sentence with four parts: if we make this change, then this metric will go up or down, because we have this evidence. For example: if we offer an instant payment option at checkout, then the share of completed payments will rise, because part of the drop-off happens at the card step.
Before you start, write down:
Guardrail metrics prevent false wins. An aggressive discount can lift conversion while crushing margin or attracting customers who cancel the following month.
Ideas are never in short supply. Time to test them all is. The ICE method scores three criteria from 1 to 10: expected impact, confidence in the hypothesis and ease of execution. The average of the three orders the queue. An example with hypothetical scores:
| Experiment | Impact | Confidence | Ease | ICE score |
|---|---|---|---|---|
| Fewer checkout fields | 7 | 8 | 9 | 8.0 |
| Discounted annual plan | 8 | 6 | 7 | 7.0 |
| Abandoned cart email | 6 | 7 | 8 | 7.0 |
| New guided onboarding | 8 | 5 | 4 | 5.7 |
The scores are subjective, and that is fine. ICE exists to structure the discussion and keep the loudest person's idea from jumping the line. Revisit the scores after each result, because what you learn changes your confidence in the next hypotheses.
Some types of test sit close to the money and deserve a place on the list:
Acquisition experiments, such as new channels and ad creatives, also belong in the queue, but they usually cost more and take longer to show revenue impact.
A badly read test is worse than no test, because it gives confidence to a wrong decision. A few precautions:
Many tests will show no difference at all, and that is information too. Keep a simple log with hypothesis, dates, metric, result, decision and lesson learned. Over time, that history becomes one of the growth team's most useful assets.
Growth works as a routine. A weekly 30-minute meeting is enough to review the week's results, decide what stays and pick the next tests. Each experiment needs an owner who sets it up, monitors it and presents the outcome.
In small companies, one person with access to development, data and marketing can run this cycle. The bottleneck is usually depending on several departments or vendors for every small change.
Bridgemoon Capital runs marketing, develops systems and has its own payment processing, which makes it possible to test everything from the ad to the checkout without relying on several vendors. In the businesses it joins as a partner, this experiment cycle is part of the growth routine.
Growth hacking became associated with quick tricks to grow, often hard to repeat. Growth, in the sense used here, is a continuous process of measured experiments across the whole funnel, focused on revenue and retention. What produces lasting results is the discipline of the method.
It depends on your traffic and the size of the difference you expect to detect. As a practical rule, run at least one or two full weeks to cover day-to-day variation, and only stop once you reach the planned sample size. Tests stopped too early tend to crown winners that do not hold up.
To detect small differences, yes. With little traffic, the way forward is to test bigger changes that produce differences easier to see, or to compare before and after periods with caution. Customer interviews and usability tests also help you pick good hypotheses without depending on volume.