Small batches and fast feedback beat large batches and slow feedback, almost without exception. It’s the closest thing delivery work has to a law of physics, and it shows up everywhere once you look: commit size, release cadence, test-cycle length, review turnaround, even how often leaders check in on an initiative.
Why it holds:
- Defect cost compounds with delay. A problem found in minutes is a fix; found in weeks it’s an investigation, a re-learning of context, and often a negotiation about whose problem it is.
- Big batches hide risk. A hundred changes tested together give you one pass/fail signal for a hundred hypotheses. You learn almost nothing about which change did the damage.
- Little’s Law does the math: average WIP = throughput × cycle time. When everything “feels slow,” either WIP is too high or throughput dropped — usually both, and usually because batches grew.
Gene Kim’s Second Way (from The Phoenix Project) generalizes it: amplify feedback loops right-to-left so problems surface immediately, not weeks later — which is most of the actual argument for shift-left testing, expressed as flow rather than as a slogan.
Related:
- Theory of Constraints — small batches make the real constraint visible
- Build quality in — end-of-line inspection is a large batch of feedback, delivered late
- Measure outcomes, not activity — lead time and deploy frequency are batch-size metrics in disguise