Blog
The experiment lifecycle: from research to revenue
The Cycle is the workflow that keeps research, decisions, and learning connected. Here’s how it works and why structure beats ad-hoc.
Most programmes start with good intentions. Research feeds hypotheses. Hypotheses become experiments. Experiments get analysed. Decisions get made. But when each step lives in a different tool or doc, the links break. Research stops informing. Decisions vanish. Learning doesn’t accumulate. The Cycle is the structure that keeps it connected.
The Cycle
Research: Surveys, heatmaps, interviews, session replays. The raw material. In a proper system it lives in a research library you can link from hypotheses.
Hypotheses: “If we do X, then Y will happen, because Z.” Testable, prioritised (PIE, ICE, PIG), and linked to the research that supports them.
Experiments: The A/B or MVT test. Designed, run, analysed. Bayesian analysis and stopping rules so you don’t p-hack or call winners too early.
Results & decisions: What happened. What you decided (ship, iterate, stop, investigate) and why. Optional revenue impact. What you learned. What you did next: the next test in the chain.
Why structure wins
Ad-hoc testing runs experiments. It doesn’t reliably connect research to hypotheses, or decisions to learning, or one test to the next. The Cycle enforces the connections. When you decide, you link the follow-up. When you prioritise, you see the research. When someone asks what the programme has delivered, you can trace from experiment to revenue to programme-level impact. Structure is what makes it repeatable and auditable.
From research to revenue
Revenue enters at the decision: control and treatment revenue, incremental impact, and conservative annualisation (novelty decay, statistical adjustments). That gives you experiment-level and programme-level numbers you can report up. The lifecycle doesn’t end at “winner/loser”. It ends when you’ve captured the decision, the learning, and the next step. That’s how you get from research to revenue in one connected flow.
Run the Cycle in Experiment OS
Research, hypotheses, experiments, analysis, and decisions. One system. See The Cycle.
Start free