Continuous Improvement and KPI’s

Continuous Improvement follows the ‘Plan – Do – Check – Act’ cycle (also known as the Reichart Cycle.   The cycle looks like this:

'Plan - Do - Check - Act' Cycle

We’ve looked at the ‘Plan’ element in designing the factory for our Chair Factory.  We’ve considered what we need to do, and what resources we need to achieve our goals.

The next step is to follow our plan, the ‘Do’ step, and monitor our performance against our targets.  To perform this step we need to look at KPI’s. 

When we’re monitoring our performance we can  use our KPI records and other triggers (such as customer complaints, jams in production, etc.) to identify the root causes of problems as they arise, and amend our processes to resolve them.  This represents the ‘Check’ and ‘Act’ stages of the Cycle, and we continue back around the Cycle once more.

But first, let’s discuss KPI’s and how we identify them.


KPI’s are Key Performance Indicators.  KPI’s are important as they provide a measure of how successful the process is operating and can be graphed to show performance trends over time periods.  KPI’s can measure many things … from rates of production, rates of material turnover, etc. but can also measure things such as quality.

However, choosing the wrong KPI’s can measure the wrong things in a process, and drive the wrong influencing behaviours as a result.

For example, consider our chair factory.  We could choose one primary KPI, the number of complete chairs manufactured per day.  We can easily measure this and record it, so at first glance it appears a good KPI.  The next step is to choose a target for our KPI to achieve … say five chairs per day.  So, now we can measure our production rate.  But the question is … is this actually a good KPI for quality?

If our factory produces six excellent chairs … our KPI looks good.

If our factory produces four chairs … our KPI now looks bad.  What we don’t know at this point, is what has caused the KPI to look so bad.

For this reason, it can be useful to identify KPI’s for each major stage of a process to try to identify problems before they affect the rest of the process.

Our process assumes everything is working well and as expected.  However, real life is never quite so simple.  What we need are KPI’s that tell us how the other elements are performing.

Think about a car, for example.  The computer in a modern car is monitoring many different elements of the car’s operation … fuel mixtures, temperatures, mixtures of exhaust gases, water levels and pressure, oil levels and pressure, position of the throttle, etc. and many other sensors … which it then analyses and uses it’s programming to determine what it needs to do to optimise the engine’s operation, from changing the idle speed, fuel mix, and other factors.  All of this takes place without the driver’s (the customer) involvement or interaction.  This is what we want for our customer … the factory monitors itself and knows what it needs to tweak to optimise the chair production without the customer having to be involved.

In the absence of a computer, our factory is going to do it manually but it’s the same principle … but it all relies on what is measured and the targets.

So, if we think about our factory again, we need to identify KPI’s for our stages.  To do this we need to identify a few elements for each KPI:

  • The stage the KPI sits against
  • What it measures
  • How it is measured
  • What the acceptable levels are
  • What happens on overperformance
  • What happens on underperformance
  • Who owns the action triggered

If we work through an example, hopefully the relevance of each element will become clear.

Let’s consider the first stage … where the wood is selected from the warehouse.  What are our goals for this stage of the process?

Our output for this stage is ‘Suitable lumber moved to Production area’, so we can derive some key elements for this stage to be successful:

  • Sufficient lumber for a selection to be made
  • Lumber of sufficient quality for suitable lumber to be found
  • Lumber stored in suitable conditions for immediate use by Production when moved

There may be others but we can use these as a starter.  We can summarise these as follows:

What it measuresLumber quantity   Lumber quality Lumber environment

Our next step is to consider how we measure each of these KPI’s and what performance is acceptable.  For this exercise, we can assume that we are always buying the right type of lumber for this exercise (e.g. pine, in certain dimensions)

  • To measure lumber quantity, it could be as simple as checking our stock of lumber and working out how many days of production we could support from stock.  Our aim would be to always have enough stock to run production between lumber deliveries and perhaps a little extra, say 8 days’ worth of stock for weekly lumber deliveries.
  • To measure lumber quality, we might decide to inspect every piece of lumber as it arrives to ensure the lumber isn’t damaged, or splitting, and reject the lumber that doesn’t meet our standards.  Our aim would be to be only accept lumber that is of a high enough quality for Production to use.
  • To measure our lumber storage environment, we might decide to measure the humidity, temperature or rate of insect infestations (all of which can seriously affect the quality of lumber).  For this example, let’s measure humidity.  Humidity that is too high can cause wood to rot, and too low can cause it to split, so we would measure the relative humidity in the air around the lumber (ideally between 50% and 70% in the UK) to monitor this

We can now add these to our KPI table, which looks like this:

What it measuresLumber quantityLumber qualityLumber environment
How is it measuredStock level in days of consumptionLumber acceptance rateRelative humidity in Lumber Storage
Acceptable levels8 days95% or more50-70%

Our next step is to consider what action (if any) is triggered when the KPI deviates from its acceptable levels.  It should be noted that it is perfectly acceptable if no action is taken on overperformance or underperformance … but if no action is take on either, you should question if it is a suitable KPI in the first place.

So, back to our example:

  • Stock levels of lumber
    • Overperformance – More than 8 days’ worth of stock – Perhaps reduce the next weekly order to help bring stock back down
    • Underperformance – Less than 8 days’ worth of stock – Perhaps increase the next weekly order to bring stock levels back up
  • Lumber Rejection Rate
  • Overperformance – Acceptance of 95% or more of delivered lumber – Doing nothing is an acceptable action at this point
    • Underperformance – Acceptance of less than 95% delivered lumber – Numerous actions could be taken, including challenging lumber supplier, buying better grade of lumber, etc.
  • Lumber Storage Humidity
  • Overperformance – Relative humidity above 70% – Warming the storage area can reduce relative humidity
  • Underperformance – Relative humidity below 50% – Cooling the storage area, or using humidifiers can increase relative humidity

We can again add these details to our KPI table

What it MeasuresLumber QuantityLumber QualityLumber Environment
How it is MeasuredStock Level in Days of ConsumptionLumber Acceptance RateRelative Humidity in Lumber Storage
Acceptable Levels8 Days95% or more50%-70%
What happens on OverperformanceReduce next weekly orderHeat storage area

The next step is to identify who is responsible for the actions that are triggered by over and underperformance.  We do this for clarity.  If we don’t identify an individual, we have no recourse if the action is not taken when needed.

Fortunately, this stage is also usually the easiest to perform.  In our case, we can assign all the triggered actions to the Lumber Store Manager.  This gives us our completed KPI table for the first stage of our process:

What it MeasuresLumber QuantityLumber QualityLumber Environment
How it is MeasuredStock Level in Days of ConsumptionLumber Acceptance RateRelative Humidity in Lumber Storage
Acceptable Levels8 Days95% or more50%-70%
What happens on OverperformanceReduce next weekly orderHeat storage area
What happens on UnderperformanceIncrease next weekly orderChallenge lumber supplier to provide better quality lumberCool storage area Use humidifiers if necessary
Action OwnerLumber Store ManagerLumber Store ManagerLumber Store Manager

Once we’ve identified the KPI’s relevant for a stage, we can also add them to our Analysis Table for stage one.

Activity NameSelect Wood from WarehouseMeasure and Cut Wood into PiecesAssemble Chair
InputLumber available in warehouseSuitable lumber in Production AreaLumber parts
MaterialsFuel and oil for forklift truckLumberLumber parts Glue
EquipmentForklift truck Hi-Viz VestSaws Chisels Mallets Saw benchClamps Assembly table
EnvironmentWarehouseProduction areaAeembly area
SkillsExperience of assessing lumber for visible grain Licence to drive forkliftBench joineryBench joinery
InformationLumber specification Chair designChair designChair assembly instructions
OutputSuitable lumber moved to Production AreaLumber partsAssembled chair
KPI’sLumber Quantity
Lumber Quality
Lumber Environment

Repeating the KPI identification across the table will give us our initial KPI’s to monitor the operation with.  As the operation runs, we may identify better ones or adjust the targets/trigger actions but we have to start somewhere and this is our baseline.

The final step is to identify the records that will be generated.  Having KPI’s is one thing, but if you don’t keep records of their performance, spotting trends and working from objective evidence (rather than ‘gut feel’) is a lot harder to do.  So, keeping records of the KPI’s, complaints received, issues found, and the root cause analysis done to resolve them, etc. are all useful records that can help the organisation learn from its mistakes and gaps.

The Continuous Improvement activity kicks in as we monitor the performance of the KPI’s and maintain our records to identify weak spots.  These may be in the performance of the KPI’s, the adherence of staff to the system, or even new areas we hadn’t considered.  By feeding that information back into our plan again, we complete the cycle and start the next one, thereby giving us a Continuous Improvement environment.