reactive

Reactive

Short description placeholder

Reactive. Tier 2 of the OperatorAI Maturity Model

What this tier means

Your team runs A/B tests when something's clearly broken or when an obvious opportunity surfaces, a new product launch, a paid-traffic campaign that's underperforming, a Black Friday push. Testing happens, but it's not on a regular cadence. When tests do run, they default to the testing tool's settings.

3-9 tests per quarter. 95% significance (whatever the tool's default is). Sometimes a winner, sometimes a confusing result. Mostly tactical, rarely compounding.

What it looks like in practice

3-9 tests per quarter (concentrated around campaigns, launches, or noticed problems)
Hypotheses come from dashboard suggestions, AI-tool variant prompts, what worked for a competitor
Sample size calculated only when traffic is obviously low
95% significance gates the winner-call (tool default)
Some peeking happens under deadline pressure
Losing tests noted in passing, not formally documented
Self-serve AI tools used inconsistently

Why this matters

Reactive programmes have started building the testing muscle but haven't yet enforced the four disciplines that compound wins. Each test produces a result, but the results don't compound across the programme because losing tests aren't informing future hypotheses and false positives at 95% are slowly degrading the baseline.

The transition to Structured (Tier 3) happens when the team commits to a regular cadence, a documented hypothesis-priority framework, and a sample-size gate before reading any test.

Recommended next move

Speed Sprint — £1,500

Page-speed audit + 3 high-impact fixes shipped + 30-day revenue measurement. 14 days end-to-end. The same playbook that took BeeFriendly Skincare from $48K to $1.45M annual revenue after a 2.24-second LCP reduction.

If page speed is already passing Core Web Vitals, AI Headline Lab £500 is the next-best entry into operator-led testing.

Score your CRO programme in 7 questions

Take the diagnostic