The second part of analyzing the problem is to review performance data about the process or activity. And I want to say a few words about that I often management reports are just single point comparisons, comparing one data point, this period, say, with one other data points, either last period or last year or with some budget or target. The trouble with these single point comparisons is that they ignore the fact that variation exists within all processes. So comparing a small number of data items with some target or budget, or with this time last year actually tells us very little, the difference between them could be just random variation. What we actually need to do is review the data over time. By presenting our business data as a time series, we can begin to filter out the random variation and actually start to highlight the signals of change in the process.

So Let's look at how we do that. And here on the next slide, we see some performance data, an on time delivery performance mapped out over about 10 weeks there. And this method of presenting performance data allows a much more informed analysis than other approaches. For example, we can see from this chart that the average on time delivery performance is around about 89%. And with this current process, there's no way that management can consistently achieve their on time delivery target, which is 95%. So the process needs to change in order to achieve this target.

In the second example, here, we have the number of days from order entry to delivery for a process. And again, the time series presentation reveals some interesting facts about performance. For example, the range of performance from order entry to delivery is quite wide and this reflects very Variable performance in the process. And clearly we see here that there's a problem around weeks sort of 27 to 29. When the number of days from order entry to delivery increased quite dramatically. Also, we see that from week 31 onwards, the number of days is reduced.

And we wonder if this is an improving trend if some improvement activity has taken place here, and I'll share a few performance measures here, which are useful when analyzing data and the performance of the process. So some of the indicators include flow, which is the end to end elapsed time from the start of a process to the end of the process. So from ordering tree to delivery as we saw in the previous graph there and the flow time of a process and the amount of variation in it shows us how stable the process is, and shows us how that is performing against customer expectations. flow time is also sometimes called lead time. The next indicator is quality, which will be the right first time Or first time through indicator, meaning what percentage of parts are of work is done correctly the first time. As an alternative, we could use the number of scrap items at a processing step, or the amount of rework.

Customer satisfaction is another good performance measure, we might use customer scores, or you might identify pain points in the process from the point of view of the customer. And also the people work in the process. And working process into country is another good measure, which would be the level of inventory at each stage. Because this helps identify the bottlenecks where work in progress inventory is high, then we potentially have a bottleneck in front of that point on time delivery is a good performance measure. And we would present that as a time series as we've seen in the previous example, to show the peaks and troughs and the range of performance. And finally, safety.

We might want to identify the points in the process, where risk is the highest key points then for analyze Using the data? Well, the time series chart gives us a way of understanding and improving business processes. And the spread of data, ie the range tells us how stable are processes. The wider the range, the less consistent the process. And the only way to reduce the range is to reconfigure the process. And interestingly, the cost of running a process is related to the spread of performance.

So the wider the range, the more inconsistent the cost of running the process will be. Because there's much more fluctuation, the closer together the range of data, that means the much more stable processes and therefore the more stable the cost of running the process will be. Where we see anomalies in the performance data, we may need to do some further analysis in order to identify the root cause of problems. For example, we may want to look at it by day of the week by shift and any other factors or characteristics that we can identify a pattern If data points also tells us stuff, for example, it can indicate any changes in the process itself. We should investigate any sudden troughs or peaks. See what was different, what occurred when those peaks or troughs happened so that we can start to isolate what's causing the issues.

Brian joiners quote from his book fourth generation management is the opposite here. Our ability to produce a rapid sustained improvement is tied directly to our ability to understand and interpret variation. And the key point here is that the wider the range of data, the less consistent the performance of the process and an inconsistent and variable process is much more likely to cause problems and also causes cost fluctuations. As I mentioned,