Monthly Archives: August 2016
If you have ever been on any sort of management course, you will have been shown a variety of matrixes to represent data, strategy etc. These can be quite visual and can help to identify products or people who fit into different areas allowing you to target specific areas rather than try and tackle everything. Most of the time, you will see these in PowerPoint or Word, and have been manually created and are not particularly accurate and are not dynamic in any way.
So how can we create a useful matrix on which we can plot data?
First of all we have to decide on the type of chart that will best display our raw data. In the case of matrix displays a scatter chart is best, plotting results based on two set of values, effectively giving us x and y co-ordinates.
Using recent enquiries I have had, some examples of this might be:
- strategic importance of work compared to the revenue it will generate
- knowledge transfer from long serving staff due to retire – level of critical knowledge vs. the amount of time before retirement
In both of these examples, the business wants to identify work or people that might have a critical impact on the business so that they can be prioritised ahead of less critical work or where very specific and unique knowledge sits with one person, and ensure that an individual’s knowledge is recorded or transferred before they leave.
Let’s take the critical knowledge transfer and create a chart to track that.
One piece of information I will need is the amount of time between today and the day the person is scheduled to retire, and I might break that down into three segments;
- less than 3 months
- 3-6 months
- 6 months to one year
The other piece of information I will need is the level of knowledge an individual might have and assign scores or levels (these are likely to be subjective);
So we may find ourselves with data like this;
Because we can’t plot “words” in a scatter chart I need to change the Low-Medium-High ratings to numerical values. Depending on how you measure this you can use a simple 1-2-3 or have something with a few more breakpoints – choice is yours.
So now our new table might look like this (I’ve used a basic data validation list for the rating);
We can now create a simple scatter chart using the due retirement and rating columns.
The data has plotted OK but nothing jumps out at us to start dealing with those who will be shortly retiring and have critical knowledge level of high. What we need is some sort of background with clear segments to help identify target employees.
We could create or find an image that contains the number of segments we need but the proportions may not be that accurate, so the image will have to be stretched or simply won’t fit in with our data. So we are going to create the background manually using dummy data.
As we have three levels of criticality and 3 time periods we will need a 3 x 3 grid pattern.
Away from your real data, type the following values;
Using these values create a stacked column chart…
…and get what you see above. At this point we need to do a few minor adjustments;
- Fix the maximum value of the Y-axis to 3
- Set the gap between series to no gap or 0%
Our chart will now look like this;
Before we start applying different colours to the various segments that make up the chart, we have to move the axes out of the way. The axes need to be moved because we need to copy the “real” data onto the chart, and in order for the data points to look as they did in the first basic chart (with no background) we need to use secondary x and y axes. If this is not done, the real data looks back to front and upside down as you need to read the values off the secondary axes rather than off the normal x and y axes. It’s not a major issue but best to sort out at this stage.
Right click on either axis. If you select the x axis first then select the axis option VERTICAL AXIS CROSSES AT MAXIMUM CATEGORY.
Right click on the y axis and from the options select HORIZONTAL AXIS CROSSES AT MAXIMUM AXIS VALUE.
The chart should now look like this;
Apply formatting of some sort to your segments. Use any of the fill options available to you at this stage – solid colours, gradients, images or patterns…more options than you can shake a stick at.
The path is now clear to add the real data to the chart.
Highlight the due retirement and rating columns as we did in the original plain scatter chart and copy them.
Depending on how much data you have, you can create a separate column using a formula to only identify people who have a leave date within the next year…no need to plot people who are not going for another 10 years!
Click on the chart that is being used to represent the 3 x 3 grid of our matrix. On the HOME tab click on the arrow below PASTE and select PASTE SPECIAL. Instead of the normal options you see if you do this (paste values, paste link etc.) you will see this dialog box;
Select all the options as in the picture above and click on OK. The chart that appears looks very odd as it is trying to plot your copied data as stacked columns – the same chart type as the base chart with our grid.
If you’re using 2010 right click on the new series you just added and select CHANGE SERIES CHART TYPE, and select SCATTER WITH ONLY MARKERS type chart. If you’re using 2013 or later then it will open the COMBO chart dialog window. Select the correct series and select SCATTER.
Now, it’s purely down to formatting. Depending on what you are presenting and how you want to highlight data points will be entirely up to you. In my version here, you will see I have added shading to differentiate between the high/mid/low ratings and some extra labels to identify the criticality levels, as well as some arrows to highlight the timescale – the latter having been added manually.
The only other thing I have added are some custom labels. As we are working with a scatter chart we can’t attach any useful labels to the data points. I have added a new column with a formula to pick out the names of people who have a criticality level of high and their leave date is within the next three months. I then used Rob Bovey’s xy chart labeller which is a free add-in that is perfect for this sort of job. If you are not familiar with this add-in, read my blog http://wp.me/p2EAVc-et which explains how it works and where you can get it from – an absolute must for these sorts of jobs.
So that’s one way of creating a custom background to make a matrix type display using Excel. Once you have your basic set up for the background, i.e. number of segments you want to measure against, it’s down to your raw data. Depending on your own Excel skills, use data validation, dynamic ranges, formulas etc. and whether you are working with static or continuously evolving data, how much interaction and dynamism you want to build into your matrix. With a little thought and preparation you can construct a pretty useful matrix type chart that will highlight data points, be they people or activities that need attention.