Six Sigma was born at Motorola in the 1980s and spread in the 1990s when General Electric adopted it. Its core claim is simple: reduce variation and you reduce defects.
In the ERP world, Six Sigma is often mentioned but rarely practised. There is a good reason for that: the full method is too heavy for most companies. Still, some of its parts are directly useful even in a company with no statistical infrastructure at all.
What does sigma mean?
Sigma is a measure of the variation in a process’s output. The higher the sigma level, the less often the process falls outside its limits. It is commonly expressed as defects per million opportunities:
| Sigma level | Defects per million opportunities | Roughly |
|---|---|---|
| 3σ | 66,807 | 93.3% correct |
| 4σ | 6,210 | 99.4% correct |
| 5σ | 233 | 99.98% correct |
| 6σ | 3.4 | 99.9997% correct |
The real message in the table is not the percentages but the distance between them. Moving from 3σ to 4σ cuts defects more than tenfold. Most companies’ transactional processes sit around 3σ, and even getting from there to 4σ makes a big difference. A 6σ target, on the other hand, is neither necessary nor economical for most business processes.
Two paths: DMAIC and DMADV
Six Sigma has two main methods:
- DMAIC — for improving an existing process. Details: our DMAIC note.
- DMADV — for designing a new process or product from scratch: Define, Measure, Analyse, Design, Verify.
You encounter both in ERP projects: improving the existing purchasing process is a DMAIC job; setting up the processes of a new business unit is a DMADV job.
The belt structure and the question of scale
Six Sigma has its own organisational model: Yellow Belt, Green Belt, Black Belt, Master Black Belt. It requires training, certification and people who dedicate part of their work to it.
This structure makes sense in manufacturing companies with thousands of employees. In a 50-person firm, the cost outweighs the benefit. The reason most Six Sigma programmes are left half-finished at the scale of Turkish SMEs is usually not the method itself but the mismatch in scale.
What we take in an ERP context
The definition of a defect. This is Six Sigma’s most valuable contribution: clarifying what a “defect” is. In transactional processes, a defect is not a faulty part; it is a wrongly opened customer account, a re-issued invoice, an order entered twice, a delayed confirmation. Without a clear definition, there is neither measurement nor improvement.
The number of opportunities. Thinking in “defects per million opportunities” instead of percentages makes processes of different volumes comparable. You can measure a unit that issues 50 invoices a month and one that issues 5,000 with the same ruler.
The idea of variation. Averages mislead. “The average delivery time is 3 days” can hide the fact that some customers wait 6 days. The habit of looking at the spread rather than the average gives you the right reflex when reading ERP reports too.
The measure–improve–control discipline. The DMAIC cycle can be applied without any certificate.
What we leave out
We don’t use the heavy statistical toolset — design of experiments, hypothesis tests, process capability analyses — in most ERP work. These were designed for physical variables on a production line; in transactional processes such as order approval, invoicing or stock counting, they are usually far too heavy.
We also generally don’t recommend setting up a belt programme. Instead, we prefer to work with the team that is experiencing the problem, on its own data.
ERP data makes the job easier
The most expensive step in Six Sigma is measurement; in manufacturing, you often need to set up a separate rig just to measure. In a company running an ERP, however, measurement of transactional processes is largely ready: who entered what and when, how many times it was corrected, how many days it waited.
That is why, in companies with an ERP, the lightweight version of the method — a clear defect definition, measurement from existing data, root cause analysis and controls built into the system — delivers disproportionate benefit. We run this work as part of Process Consulting and make the measurements continuous on ErpwareBI.