Numbers Don't Speak for Themselves

A number is not an insight. It is the raw material an insight is made from, and the gap between the two is where most data presentations quietly fail. This week has been about turning analysis into a story people act on, and today we face the hardest raw material of all: the numbers themselves. This is where the recent work on reading and interpreting data meets the work of communicating it. Getting to a trustworthy number is a real achievement, and it is only half the task. A trustworthy number that nobody understands the meaning of is just a more accurate way to be ignored. The other half is making that number mean something to a person who has to decide, and that is a different skill from producing it. The trap is that the number feels self-evident to you because you lived through the analysis that produced it. You know that fourteen per cent is alarming, that this figure is up and that one is flat, that this gap is the whole story. The audience knows none of that. They see a digit on a slide with no sense of whether it is good, bad, big, small, normal or catastrophic. Your job is to supply everything the number cannot say for itself.

The so-what test

The single most useful habit in data communication is to interrogate your own findings with one relentless question: so what? You have a number. So what? It means this. So what? That matters because of this. So what? Which means we should do this. You keep climbing until you reach a decision, and the level you stop at is the level you should actually present. Most people present two or three rungs too low, down at the raw findings, and leave the audience to climb the rest alone. Take a real example.

The data point: mobile checkout completion is fifty-eight per cent. So what? It is fifteen points below the web, and it has been falling since April. So what? That gap is where most of our lost mobile revenue is coming from, roughly the figure we keep blaming on the market. So what? If we fix the mobile checkout, we will recover more revenue than the entire campaign we are about to fund is projected to bring in. Now you have something worth an executive's attention, and notice it is the same number the whole way up. What changed is that you climbed it to the decision instead of dropping it, raw, on the table.

When it comes to showing data, the discipline that matters most comes from Cole Nussbaumer Knaflic's work on storytelling with data, and it is stricter than most people expect: each chart should make one point, and that point should be impossible to miss. Not a chart that contains six trends and invites the audience to find the interesting one. A chart built so that the single thing you want them to see is the loudest thing on it, with everything else quieted down to support it. If you have three things to say, you have three charts, not one chart with three arguments fighting inside it. Two moves make this real.

• First, strip the clutter: every gridline, label, colour and series that is not carrying your point is competing with it, so remove it. A chart is not a place to prove how much data you have.
• Second, and this is the move most people miss, put the insight in the title. Not the topic, monthly mobile completion rate, which tells the reader nothing, but the point, mobile checkout has fallen fifteen points since April. The title is the most-read text on any slide, and wasting it on a topic label instead of the takeaway throws away your best chance to be understood. When someone reads only your chart titles, in order, they should get the whole argument.

Order the charts as an argument, not an archive

When you have several things to show, the order is not neutral, and the default order, the one you discovered them in, is almost always wrong. An archive is arranged by when you found things. An argument is arranged by what the audience needs to believe first in order to accept what comes next. Ask what the audience has to accept before your recommendation can make sense, and put those pieces in that order. Usually, it runs: here is the situation, here is the specific thing that is wrong, here is what it is costing, here is what fixes it. Each chart earns the next, so that by the time you reach the recommendation, it feels less like your opinion and more like the only sensible conclusion, because the audience has been walked to the edge of it themselves.

A simple check: read only your chart titles, in order, top to bottom. If they tell a single connected story that ends at your recommendation, the sequence is right. If they read as a disconnected list of topics, you have built an archive, and you need to reorder it into an argument before anyone else sees it.

Make the number something a person can picture

Large numbers are almost meaningless to the human mind. Chip Heath, in his work on making numbers count, shows that abstract figures only land when you translate them into something a person can actually picture. 2,300,000 is a blur. The same figure, restated as the entire annual budget of our support team walking out the door every year, is a punch, because now the listener can see it. The number did not change. It was made concrete, given a human scale and a comparison the audience already understands.

So whenever you present a figure that matters, do not leave it naked. Anchor it against something the audience already has a feel for: a familiar budget, a known target, last year, a competitor, a per-day or per-customer breakdown that shrinks a huge number to a human one. A GBP 400,000 annual loss can become more than a thousand pounds a day, every day, since spring. Same truth, but one version a person can hold, and people only act on what they can hold.

Show the conclusion, not all your work

There is a deep instinct among people who do rigorous work to show that work, to lay out every step so the audience can see how carefully you got there. In a presentation to decision-makers, that instinct betrays you. They do not want to retrace your analysis. They want your conclusion and just enough evidence to trust it, with the rest available if they ask. Keep the detailed work in an appendix or a backup, ready for the question, and out of the main line of the story. Showing all your work does not read as rigorous to a busy audience. It reads as an inability to tell what matters from what does not, which is the opposite of the impression you want to leave.

The mistakes that keep data inert

A few habits reliably stop numbers from ever becoming decisions.

1. The data dump: every chart you made, shown because you made it, with no single point guiding the audience through.
2. Topic titles instead of insight titles, so the reader has to work out the point of each chart for themselves.
3. Naked big numbers, presented with no comparison or human scale, so they blur instead of landing.
4. Stopping at the finding instead of climbing the so-what ladder all the way to a decision.

The cost of leaving your data inert is rarely loud, which is exactly why it is easy to keep paying. It is the insight that died in a dashboard nobody opened. The finding that was technically delivered, in a deck that was technically presented, and then quietly ignored because no one in the room could tell what it meant or what to do about it. Nothing dramatic fails. The work simply does not travel, and months later the problem it identified is still there, still costing what it always cost, while the analysis that spotted it sits forgotten in a shared drive. That is the real price, and it is paid not by the executive who did not act but by the analyst whose good work went nowhere, and who slowly gets a reputation for producing things that do not lead anywhere.

Go out and be successful.

Oluwatosin Ogunkoya | Flotog BA Insights  |  www.flotogbainsights.com

Tomorrow: Speaking to the Person Who Has to Decide. The bottom line first, the one number they will repeat, and making your ask impossible to miss.