Improving flow or reducing variation: two improvement programmes

Comparison criteria

Criteria are stated explicitly and neither option is declared a winner: which one fits depends on the constraint that binds hardest in your operation.

The two traditions came from different problems and still show it

One grew out of production systems where the enemy was inventory and the time a product spent waiting between operations, in plants making broadly similar items repeatedly. The other grew out of high-volume manufacturing where the enemy was a defect rate driven by variation nobody could see without measuring. That origin explains their instincts. Flow work asks where the product stops and why; variation work asks what makes the output spread and by how much. A plant with weeks of queue and a stable process wastes its time on designed experiments. A plant with a two-day lead time and a stubborn yield problem gains nothing from mapping a flow that is already short.

Belt training is not a programme, and certificates are not results

The most common way both methods fail is identical: an organisation buys training, produces qualified practitioners, and never builds the governance that turns their work into money. Nobody selects problems against the plant's real losses, nobody protects the time, and finished projects hand their improvements to an area with no mechanism to hold them. Within a couple of years the training investment has produced a folder of closed projects and an unchanged cost base. What distinguishes plants that get value is unglamorous: a maintained list of losses ranked by size, a named sponsor per project, a defined handover into standard work, and a review that checks whether the improvement is still in place months later.

The bundled version works when the diagnosis comes first

Programmes combining both are sold as a single curriculum, and the combination is legitimate — the toolkits genuinely complement each other. The failure happens when practitioners reach for whichever tool their training emphasised rather than diagnosing the loss. A useful sequence is to characterise where the money is going before deciding on a method: measure lead time against work content, separate losses into waiting, defect and capacity categories, and let those numbers point at the approach. Plants that do this often find the first year of work is overwhelmingly flow-related because queues are visible and large, with variation work becoming the productive frontier once the obvious structural waste has gone.

Frequently asked questions

Can a small manufacturer run either of these seriously?

Flow-focused work scales down comfortably, because observation, standard work and short improvement events need attention rather than infrastructure. Structured variation projects are harder in a small plant, since they assume someone can be released from daily firefighting for weeks and that enough data exists to analyse. Smaller operations often get more from a lighter structured problem-solving discipline applied to their few recurring defects, reserving heavier statistical work for the one or two problems that genuinely resist it.

Which approach should a plant start with?

Start with whichever matches the loss you can actually quantify. If a product takes weeks to move through a factory that touches it for hours, the queues are the problem and no amount of statistical analysis will shorten them. If lead time is already short and the plant is scrapping or reworking a meaningful share of output for reasons nobody has explained, that is variation work. Diagnosing first also protects the programme's credibility, because the first result is visible and defensible.

Do improvement savings ever show up in the financial accounts?

Only when the resource freed is actually removed or redeployed to something that generates revenue. Saving operator hours in a cell changes nothing financially if those hours stay in the cell doing less. The disciplines that make savings real are agreeing beforehand what will happen to freed capacity, involving finance in how a benefit is counted, and re-baselining the standard cost once a change is embedded. Programmes that skip this accumulate claimed savings that nobody in finance recognises.