Interpreting data is a balance of science and art – of knowing when to question the figures, and when to let them speak for themselves.
At an in-house training workshop recently, I found myself grappling with a participant’s question on how to tell the ‘story’ of numbers. I’ve been mulling the question ever since.
It’s simple when the numbers say good things – sales are up, business leads have multiplied, productivity has grown. It’s also easy when the increments are big – 20 per cent, double, three-fold.
But when the numbers deliver news that might mislead, befuddle or disappoint executives or business partners, it’s time to get creative. You could group several categories if they are weak individually, to imply a trend. Or you could search for numbers that illustrate the flip side of an occurrence.
What you’re realising by now is this: numbers alone don’t tell stories. Circumstances do. When asked to interpret any data, your first task should be to find the reasons, people, trends, beliefs and other circumstances surrounding those numbers.
Draw on different kinds of research to do this, such as qualitative, applied and conceptual, either to flesh out a hypothesis or to find relationships between different groups of data. Here’s a closer look.
Scenario #1: You have a hypothesis.
Starting with a hypothesis gives you some direction. Say you’ve been tasked with proving that a beleaguered TV network does offer commercial merit for potential advertisers. You know that audience numbers for the network are down overall, but you’ve heard anecdotally that some of their new shows are faring well.
You could use your analytical research skills to break the numbers down further – to uncover what demographic is tuning in, which shows are doing well, and which ones are flat-lining. You could then harness your qualitative research skills, perhaps designing a field study or interviewing a focus group to uncover why those shows are popular, and how their success compares with similar programming at rival networks. You could broaden the research field further to include correlational research (see P5 of this SlideShare presentation), and try to identify a social trend that mirrors what’s happening in your business.
With these facts in hand, edit your hypothesis and you’ll see you have something to talk about – a newly female-centric audience, an evolving program lineup that’s ditching the old and embracing the new, a new breed of celebrity that’s drawing a cult following. That’s your story.
Scenario #2: No hypothesis, all loose ends.
When there’s no hypothesis and only raw numbers, your foremost task is to find comparisons, groups, likenesses. For example, you might be writing an annual report that seeks to explain why sales volumes are up, but corporate revenue is down.
You could start by going back to your quantitative figures, and see what insights you gain from grouping categories differently. Do you get more pronounced highs or lows in the data when looking at only new-release product sales, or at product lines that invested in marketing? Next, you could use empirical research to compare different fields over one year or several years – to identify which divisions did well, which ones underperformed and which increased expenses have led to the revenue decline. You might then use analytical research to find out why those numbers are up or down – for example, the managerial issues that caused one division to use extra contractors for the year, the R&D expenses that went over budget (but are forecast to bring profits back in line soon), the skills shortages that affected productivity that year.
You now have a hypothesis – which you should re-test, of course – and a story emerging of a trend happening in your business that might otherwise have remained hidden.