A couple of weeks ago I was at a conference in Lisbon and spent five days listening to some very smart people share some really valuable information (see my summary of new findings in mobile market research).
However, many of the presentations at the conference were rendered less impactful because of the way the presenters showed numbers. There are some easy ways to make numbers more accessible and impactful, and in this post I share a few of them.
Note, this post does not focus on data visualisation – that will be another post. This post looks specifically at six easy steps you can take to ensure that the numbers you display can help tell the story, increase engagement, promote understanding, and make action more likely.
1) Use Fewer Numbers
The first tip is to simply use fewer numbers. Imagine that you had to pay 5 dollars for every number you included in your presentation; you would soon cut back the quantity of numbers.
The key question to ask is whether each specific number says something useful. If it doesn’t, drop it from the presentation (even if it stays in the background data, notes etc.)
- Instead of showing the top ten, consider showing the top five.
- Instead of showing before and after numbers, perhaps just show the movement.
- Rather than show volume and value, consider which of them relates to the message you are telling and show that one. (But in the background notes, report etc you would usually want to show both.)
- Show fewer views of data. Do you really need to show the data broken by gender, age, region etc? Decide which elements are core to the story. These should be shown, but showing other views (out of habit or completeness) hides the message.
For example, the table below shows (hypothetical) average (median) hourly rates for four regions of a country. Having looked at the data, let’s assume we have decided the story we want to present on this occasion is about the differences between the regions. Since the story we are telling is about the differences between the regions, the Total (i.e. the national) picture is not adding anything – so we remove it.
2) Differences Should Be Big
Showing people numbers that are very similar, is a method (but not a great one) of saying “Look, not much going on here!” But, if you want to illustrate that something is happening, you need to make sure the differences are big, they need to be big enough to jump off the page.
In the data below, the differences between the four regions do not appear very large, when expressed as hourly rates. However, when these median salaries are expressed as annual salaries, the regional differences become more visible.
3) Use Fewer Digits
Sometimes you may hear discussions about how many decimal places to display, but that is not the key issue. The key issue, when making numbers more visual, is how many significant digits to display, and the answer is usually two or three.
The right-hand table below shows the annual salary data with two significant digits, i.e. we show them as thousands. The numbers are becoming every clearer, and the gap between the North and South ever more noticeable.
4) Declutter The Numbers
Reducing the number of significant digits makes the numbers more visual, but we can complete the job by removing any recurring symbols. In the example below, this means removing the $ signs, so the message in the numbers stand out even more.
5) Sort The Information For Clarity
It is amazing how often data is presented in an order that makes it harder to interpret. For example, data is often arranged in the order it comes out of the data collection process or statistical analysis package. In the example below I have sorted the data in descending order of median salaries by region.
Now it is very apparent that the people in the North are paid much less than the people in the South, $16 thousand a year less – in terms of the differences between medians.
6) Use Interpretive Or Summarising Treatments
If the message we want to convey from this data is that there are wide variations in income between the regions, then the table above is sufficient. However, perhaps our message is not just factual, but also emotional, we want people to see unfairness in these differences in salaries.
One way of using interpretative or summarising treatments with the earnings data would be to show them as one of:
- Differences from the national picture
- Differences from the lowest paid region
- Differences from the highest paid region
The differences mentioned above could be shown in as absolute values (e.g. the West earns $12K a year less than the South) or as percentages (e.g. the East only earns 84% as much as the South. However, all of these treatments stay within the language used by the raw data, with its focus on money. There are alternatives, such as focusing on days worked.
We can express the difference between the median salaries in terms of how many extra months the people in the other regions have to work to reach the same median income that the South earns in one year.
Months that the regions have to work to earn the same amount as the median annual amount for the South:
- East – 14 months
- West – 18 months
- North – 21 months
We have reduced the original number, to three simple numbers, describing how many months extra the other regions would have to work to earn the same as the South – now the numbers are helping tell the story. Another way to show these numbers would be just to report the number of extra months each region had to work to achieve parity, i.e. East 2 extra months, West 6 extra months, North 9 extra months.
As I mentioned earlier, this post is about working with numbers. Another post will tackle the topic of visualising. However, it is worth noting that if we had tackled the initial data (regions and earnings per hour) with visualisation we might have produced something less impactful than it could have been, because the visualisation might have dealt with the data, rather than the story that was revealed by simplifying the numbers
However, if we take as the starting point for visualisation the extra months the other regions need to work, the task for the visualisation becomes much more straightforward.
Want to know more?
I have recently broadcast two webinars on the topic of working with numerical information to find and communicate the story in the data. You can access the slides and recordings from the NewMR Play Again page.
Note, most of the things I have mentioned in this post were described in the 1980s by Andrew Ehrenberg, and I warmly recommend his 1982 book “A Primer in Data Reduction”.