Did you see that piece in the news earlier in the year? Without wanting to upset any accountants out there, the question was whether current UK financial auditing practices were giving effective warning of possible financial problems. The article by accountancy professor Prem Sikka cited Carillion and BHS as examples.

The question then arises why the managers themselves did not spot they had a problem from their numbers. Unfortunately the managers in these and many other cases are not alone. A fundamental problem is that too many business owners and managers are not aware that there are better ways to use performance data or Key Performance Indicators (KPI) as they are often called.

The least useful way to present "the numbers", as they are often referred to, is to put a whole lot of different types of data in a table. You might add budgetary figures and even the numbers from the last reporting period. Crucially this presentational approach is purely that, a *presentation*. It cannot as such analyse your numbers. OK you might comment on the ratios of various numbers. But what does any of this really tell you about what is happening? Not a lot.

World leading statistician Dr Donald J Wheeler says,

'Data has no meaning without a context'.

The simplest and most useful context is to look at how your numbers change *against time*. A simple line graph will do this for you and help you link data points to real-world events.

Without looking at your latest number in the context of other data over time you are just treating it as a piece of *count data*. That is, how much or how many, which can have its place although it doesn't help to better understand how an organisation and its processes work and can be improved. It is, however, more useful, when managing a business or any organisation, to use your numbers in the form of *analytical data**, *which means *properly *analysing it. In this way you get a sound basis for decisions supporting management action to improve performance. So what is this proper analysis?

If you have your numbers in a graph against time, as above, you are already one step closer to proper analysis. The time axis links your numbers to real-world events, hence possible causes, and it may be possible to see any trends that are present.

Talking of trends take the UK retail sector. As everybody knows Christmas produces a strong trend in retail sales. As did the January sales until some bright spark came up with Black Friday sales! It is essential for managers to be aware of these seasonal effects and other patterns or trends and how they are varying. It's much harder to spot trends in a table of data.

When real-world events produce unusual effects in the performance of a process managers need to know as soon as possible. The key here is to plot your numbers in real time as soon as they are available, otherwise you might as well be trying to drive around only looking in the rear view mirror!

The above a graph was of despatch data showing that possibly there was a falling trend in performance over the first three weeks or so. Was the drop off on day nine significant? How would we know? A proper analysis would be to use a *Process Behaviour (PB) **chart*.

A PB chart has been a proven data analysis method for a very long time. It is known by many other names, mostly variations on Shewhart Control Chart, a name which acknowledges the man who developed it, Walter A Shewhart. What Shewhart did was to add three decision lines to a line graph of the data being studied. Below is an example of the three lines added to our previous line graph of daily despatch data.

The green line is the average for this set of data, which immediately gives your eye a reference point that can be useful for identifying possible overall patterns and trends. The two red dotted lines were Shewhart's real breakthrough. He said that if your process was stable and predictable (what he called in statistical control) you might expect future points to lie anywhere between these two lines providing the process being studied did not change. Any data point outside these two lines would be caused by some exceptional event different to stable and predictable running of the process that is worth investigating. We call this **rule 1**.

So in the PB chart above data point A is outside the lower red dotted decision line and would have been worth investigating immediately at the time because something exceptional had occurred. Perhaps a machine or vehicle had broken down or the supply chain had failed, the latter perhaps even due to a weather event such as a typhoon or tsunami!

One of the other interesting times that you can learn from a PB Chart is when you get a run of *eight or more data points above or below the green average line*. This is highly unusual for any system or process and we call this **rule 2**. In the PB Chart above you can in fact see nine points at B all above the average. This is telling us that on average over these nine days the process was more stable and operating at a higher average. Why was this? What had changed in the operating conditions or the supply chain over this period?

So by drawing a line graph and then applying these three decision lines you can create your own PB chart and get more from your numbers, which will give you a better understanding of your organisation. With this knowledge you should be in a position to improve performance and, who knows, one day you might achieve World-class performance! And you don't necessarily need a fancy computer programme to do this. You can even do it with a piece of graph paper and a pencil!

If you would like a simple worksheet to help you calculate the decision lines do please use the contact page on this site.

#PBchart #DataAnalysis #ImproveKPIs