The incredible lightness of numbers

As my day job is in strategic marketing, I often have to deal with numeric data, such as market size, market shares, revenue forecasts, etc. That’s maybe the reason I stopped believing in naked figures. Particularly when presented in isolation, without any information about their source, logic or meaning.

Take this example. A few months ago I was preparing a presentation about disruptive market trends in telecom. While crafting a slide about the massive potential of the Internet of Things, I got confronted with growth forecasts ranging from 26 billion units by 2020 to 212 billion things in 2020.”  Yes, that’s a difference of a factor of almost ten. One would expect a bit more alignment between respected industry analysts like Gartner and IDC. It was even impossible to tell which estimate was the most accurate one, because it wasn’t also very clear what “units” or “things” they actually counted…

Here’s my point: numbers are meaningless without context or without a good explanation. There’s a quote attributed to Winston Churchill, saying that:

“The only statistics you can trust are those you falsified yourself.”

Although sources claim that Sir Winston never made such statement at all – which means that you should be as cautious when citing quotes as when showing numbers – there’s certainly some truth in it:

  • Most presenters use figures either to prove their point or to persuade their audience (of a point they aren’t able to prove.) Both may of course be honorable causes, but still, as a member of the audience this often gives me an uncomfortable feeling of being manipulated.
  • Even when facts and figures are not intentionally misleading, they still may be massaged to invoke more (or less) emotion (see e.g. Garr Reynolds’ example about the usage or tables and charts in my previous post.) And of course, the same numbers can mean different things to different people.
  • You can prove anything you want with numbers, statistics and correlations.  From a 2011 BusinessWeek article I learned that Facebook ignited the Greek debt crisis, and that Global Warming is caused by scientific research… If you (or the people listening to you) have no idea of what’s behind a correlation, you may claim any fact you like.
    Another more recent case of such correlation equals causation thinking – also known as the cum hoc ergo propter hoc fallacy – is a Princeton study saying that Facebook would lose 80% of its users by 2017. The numbers generated a  row between the Princeton University researchers and the Menlo Park social networking giant, as the latter on its turn “proved” that the renowned university would lose all of its students by 2021.

correlation_greece_facebook

As a conclusion, if you want to include numbers, statistics and correlations in your presentation, use them scarcely, carefully and wisely. Always mention their source(s), present them with the necessary reservation, and in the right context.

For what it’s worth: last week, Google announced a new tool that should help data analysts distinguish cause from correlation, when e.g.measuring sales generated by a web banner, or estimating the impact of a new feature on app downloads.

More reading:

To those who understand life

If you were to say to the grown-ups: “I saw a beautiful house made of rosy brick, with geraniums in the windows and doves on the roof,” they would not be able to get any idea of that house at all. You would have to say to them: “I saw a house that cost 15,000 euros.” Then they would exclaim: “Oh, what a pretty house that is!”

Just so, you might say to them: “The proof that the little prince existed is that he was charming, that he laughed, and that he was looking for a sheep. If anybody wants a sheep, that is a proof that he exists.” And what good would it do to tell them that? They would shrug their shoulders, and treat you like a child. But if you said to them: “The planet he came from is Asteroid B-612,” then they would be convinced, and leave you in peace from their questions.

They are like that. One must not hold it against them. Children should always show great forbearance toward grown-up people.

But certainly, for us who understand life, figures are a matter of indifference.

I should have liked to begin this story in the fashion of the fairy-tales. I should have like to say: “Once upon a time there was a little prince who lived on a planet that was scarcely any bigger than himself, and who had need of a sheep…”

To those who understand life, that would have given a much greater air of truth to my story.

Above is an excerpt from one of my favorite novellas, “The Little Prince” by French aviator and writer Antoine de Saint-Exupéry, which implicitly gives a great definition and provides a strong rationale for storytelling.

petit_prince

Indeed, many people like facts and figures, but those numbers become more meaningful if you embed them into a context that appeals to their emotion. For many professional speakers this should be no surprise at all. Anyone familiar with Aristotle’s ancient art of rhetoric knows that a well-balanced mix of ethos, pathos and logos motivates and persuades your audience – and makes your presentation memorable.

In their book, “Made to Stick: Why Some Ideas Survive and Others Die,” Chip and Dan Heath describe an experiment with students at Stanford University. All pupils had to prepare and deliver a one-minute persuasive speech. After everyone had finished their talk, the students were asked to rate each other on the effectiveness of the presentations and write down the key points they remembered.

  • On average, the presenters used 2.5 figures in their one-minute speeches
  • Only about 1 out of 10 used a personal story to make their point
  • 63% of the class remembered details from the speeches that used stories
  • But only 5% of the audience remembered the statistics

The little prince had it right. Figures are a matter of indifference and, to those who understand life (and IMHO to all the rest of us too), a good story can give a much greater air of truth.

Other articles about this topic that are worth reading: