HADI loop
The HADI loop is a four-step cycle, hypothesis, action, data, insight, that turns a growth plan into a weekly rhythm of testing instead of a single plan reviewed once a quarter.
The HADI loop is the testing cycle Growth Lab runs inside RoadMap, once RoadMap has named which decisions to test first. Each loop tests one assumption about a channel, message or audience against a number agreed before the test starts, hypothesis, action, data, insight, and ends with a decision to scale, fix or drop the idea.
A loop is short on purpose. Where an annual plan reviews a channel once and lives with the result for a year, a HADI loop runs on a weekly or two-week cycle, so a team gets several results before a budget is fully committed. The loop does not replace a growth plan. It is how a plan gets tested piece by piece instead of approved once and left unchecked until the next planning cycle.
What makes a loop a loop, rather than a single test, is the fourth step. The insight has to produce the next hypothesis, and the next loop has to start from it. A result that nobody turns into a decision was not a loop, it was a report.
Across a Growth Lab engagement, most weeks run more than one loop at once, each testing a different channel or message, so the roadmap moves on several fronts rather than waiting on one test to finish before the next one starts.
How it works, step by step
- Hypothesis. Write down a specific belief: which channel, message or audience will move a named metric, and by roughly how much.
- Action. Run the smallest version of the test that could prove or disprove it, on a fixed budget and timeline.
- Data. Measure the result against the number agreed before the test, not against how the test felt.
- Insight. Decide to scale, fix or drop the idea, write down why, and start the next loop.
When to use it
Use it once a growth plan has more ideas than time to test them, and no consistent way to decide which one to run next. It fits a team that already has some traffic or customers to test against; a company still finding its first customers needs interviews before it needs a testing cycle.
Common mistakes
- Testing without a number agreed in advance, so any result can be called a win afterward.
- Running loops longer than a week or two, which turns the cycle back into a slow annual plan.
- Scaling a result nobody re-tests at the new, larger budget.
- Dropping an idea after one loop instead of one fix attempt first.
