Continuous improvement has a poor reputation in many businesses because it arrives as a programme, consumes attention for six months and stops. The versions that work are smaller, more boring and permanent.
Start with what customers actually experience
Improvement effort is frequently spent on internal efficiency that customers never notice, while the things that cost you customers go unaddressed.
The useful starting question: where do we fail customers? On-time delivery, order accuracy, response time, error rates and complaint volume all measure things customers feel.
Fix those first. Internal efficiency matters and it is second.
Map what actually happens
Process mapping sounds bureaucratic and is genuinely useful, provided you map reality rather than policy.
Walk the process with the people who do it. Ask what actually happens, including the workarounds. Most processes contain steps nobody can explain, duplicated data entry, waiting, and rework caused by something upstream.
The three most common findings in any operation:
- Waiting — for approval, for information, for the next stage.
- Rework — correcting errors created earlier, which is where a surprising share of effort goes.
- Movement — people and goods travelling further than necessary because of layout.
Fix the process before automating it
The most expensive improvement mistake is buying software to solve a process problem. Automating a badly designed process produces a faster badly designed process and a subscription.
Map, simplify, then consider whether software helps. Businesses that do it in that order configure the software to how they work; those that do not reshape their work around the software.
The people doing the work know
This is the most reliable and most under-used source of improvement available. The people on the floor know which steps are pointless, which equipment is unreliable and which procedures are routinely worked around.
What makes them tell you:
- Asking specifically rather than generally. “What slows you down on this task?” produces better answers than “any suggestions?”
- Acting visibly on something, quickly. One implemented suggestion generates more than any number of invitations.
- Closing the loop when you do not act, with a reason.
- No blame. If reporting a problem implies criticism of someone, reporting stops.
Measure a small number of things
Improvement requires knowing whether something got better. That means a measure before you change anything.
Useful operational measures: on-time delivery against what you promised, order accuracy, inventory accuracy against cycle counts, throughput time from order to dispatch, and rework rate.
Track trend rather than single points, and be honest about definitions. A measure whose definition drifts produces a trend that is an artefact.
Small and permanent beats large and temporary
The pattern that sticks:
- A short regular meeting — weekly, fifteen minutes, at the workplace rather than a boardroom.
- A visible list of issues raised, who owns each, and status.
- Follow-up as the first item. What happened to last week’s actions. This single discipline is what separates improvement from conversation.
- Small changes tested quickly rather than large changes planned extensively.
Programmes with a launch, a consultant and a name tend to end. A weekly meeting that nobody notices tends to continue.
Safety and quality are part of it
Operational improvement, safety and quality are frequently run as separate systems reporting to different people, which produces conflicting priorities.
They are the same conversation. A process that generates rework generates frustration and shortcuts; a layout that creates congestion creates both delay and vehicle-pedestrian risk. Near-miss reports are operational information as much as safety information.
Standardise before improving
A process performed three different ways by three people cannot be improved, because there is nothing stable to improve.
Agree one way, document it briefly, train to it, then improve from that baseline. Standardisation is not the enemy of improvement — it is the precondition.
Know when to stop
Not every process warrants optimisation. Effort should follow value: the process that runs a thousand times a day, or the one where errors are expensive.
Optimising a monthly task to save ten minutes is a poor use of the attention it consumes.
business.govt.nz publishes operations and productivity guidance, WorkSafe publishes material on worker engagement, and Stats NZ publishes productivity data under an open licence.
