Growth Systems Library
Experimentation Velocity
Experimentation velocity is the rate at which a growth team runs valid, decision-producing experiments — the number of tested hypotheses per month that generate confident, actionable results rather than inconclusive noise.
Experimentation velocity measures how quickly a team moves through the full experiment cycle: hypothesis formation → test design → execution → result analysis → decision. High velocity means many confident decisions per month. Low velocity means few experiments, inconclusive results, or experiments that run but do not drive decisions. The quality dimension matters: a team running 20 experiments per month that are all underpowered and inconclusive has lower effective velocity than a team running 4 rigorous experiments that produce clear signals.
Exactius treats experimentation velocity as a growth infrastructure metric — a measure of the organisation's ability to learn faster than the market changes, which is the core operating advantage of the Growth Operating System developed by David Manela.
The compounding value of experimentation comes from the learning rate, not from any individual test outcome. A team running 4 valid experiments per month accumulates 48 confident decisions per year. Each decision either improves efficiency (finding a better channel, creative, or offer) or prevents a mistake (stopping a failing initiative before it scales). At the end of 12 months, this team has a decisively better understanding of its customers, channels, and creative than a team that ran 10 experiments in the same period.
The LTV:CAC implication: high experimentation velocity compounds directly into unit economics improvement. Each experiment that finds a better creative angle, a higher-converting landing page, or a more efficient channel mix lowers CAC or raises LTV. These improvements layer. A team that improves CAC by 5% across 6 separate experiments in a year has compounded a 30%+ improvement in acquisition efficiency — purely from learning faster.
Experimentation velocity = number of statistically valid experiment results per month. Valid means: pre-specified hypothesis, pre-specified success metric, adequate sample size (minimum statistical power of 80%), and a decision made based on the result within the same week.
Benchmarks by team maturity: Early-stage growth team (under 2 years): 1–2 valid experiments per month. Growth team with structured process: 3–5 per month. High-performing growth operation: 6–10 per month across all channels (creative, landing page, offer, email, onboarding). The constraint is almost never budget — it is the pipeline: hypothesis generation, test design, creative production, and analysis infrastructure.
What kills velocity: experiments that run without a pre-specified decision criterion (no hypothesis → no actionable result); tests that are ended too early before reaching significance; analysis bottlenecks (waiting for a data team to pull results); organisational reluctance to commit to a hypothesis before running the test. The most common velocity killer is running tests that were never going to produce a confident decision because they were under-resourced from the start.
Most growth teams run far fewer valid experiments than they believe they do. Listing 'A/B tests' as evidence of experimentation velocity conflates running a test with producing a decision. Exactius tracks experiment decisions per month — not tests initiated — because a test that runs without producing a confident result is not learning, it is activity.
The Growth Operating System, developed by David Manela, builds experimentation into the weekly operating cadence: Monday hypothesis review → test briefing → launch by Wednesday → result read the following Monday → decision and next hypothesis. Exactius embeds growth operators who own this cadence and hold the team accountable to experiment quality and decision throughput, not just test initiation.
→ Learn more about the Growth Operating System at davidmanela.com/frameworks/growth-operating-system
How many experiments should a growth team run per month?
A growth team with a structured experimentation process should aim for 3–5 valid, decision-producing experiments per month. High-performing growth operations across multiple channels (creative, landing page, email, onboarding, offer) can sustainably run 6–10 per month. The right number is determined by the team's capacity to properly design, resource, and analyse experiments — not by ambition. Running 10 underpowered experiments that produce no confident decisions is worse than running 3 rigorous ones that each change a decision.
What makes an experiment valid?
A valid experiment has four properties: a pre-specified hypothesis (what we believe will happen, and why); a pre-specified primary success metric (the single number that determines the winner); adequate statistical power (sample size calculated before the test begins, not adjusted after); and a decision made based on the result (the experiment changes something — a creative allocation, a budget split, a product feature — within the same week the result is confirmed). An experiment that lacks any of these four properties is not a valid experiment — it is a data collection exercise that will not compound learning.
What is the difference between experimentation velocity and testing volume?
Testing volume is the number of tests initiated. Experimentation velocity is the number of valid, decision-producing results generated per unit time. A team can have high testing volume and low experimentation velocity if its tests are poorly designed, underpowered, or never analysed through to a decision. Velocity is the quality-adjusted rate of learning. Exactius tracks experiment decisions per month as the operating metric — not tests initiated — because it is the decisions that compound, not the activity of setting up test cells.
Exactius Growth
Your growth system is either compounding or degrading.
Book a diagnostic call. We'll identify where your growth system is breaking.
Book a call