Lies, Damn Lies and...

I've never had much truck with statistics. They always seem to be wrong, but people still frown seriously over them and make important decisions from the data provided. Take the weather forecast. Millions spent on predicting what the weather is going to do tomorrow and then, for no apparent reason, the wind changes and it tips it down when it's supposed to be a heatwave. What's the point?

Another example. A better one. In my industry, a language school examined their end-of-year stats and discovered a massive increase in one-to-one English lessons for students from Russia. The sales department was galvanised and a new campaign organised to reflect this sudden spike. A trip to Moscow and St Petersburg was even mooted to attend various fairs and workshops. Then some bright spark dug a little deeper and found that the surge in bookings was actually the result of just one Russian oligarch who lived in London and had booked a whole year of one-to-ones, worth thousands of pounds. Legend has it that he parked his pink Rolls Royce outside the school every day - and received a parking ticket every day too (another statistic which probably messed with the traffic violation data in that street). The campaign and trip were cancelled and the man who discovered the anomaly took the Russian out for afternoon tea at the Ritz.

Statistics are only as good as the humans who first, diligently enter all the data all the time and second, who take time to analyse the results carefully.

In the language school industry, schools are usually on the small size, which means that pretty much all statistical data is either futile or invalid. If you work for Nike, Coca Cola or Microsoft, you can mess with data harvested from millions of people. Suddenly, things seem to make a bit of sense and you can predict stuff, albeit very carefully. But when I see language schools investing in lavish, expensive new data analytical systems, I wonder why. The data is generally meaningless and sometimes outright harmful. Relying on flimsy information can destroy a company and certainly drain a marketing budget very quickly.

Much better to sit down with a few marketing and sales people and talk about 'gut' instinct in markets. This might sound suicidal in the 21st century, but what it all comes down to in the end, is the human element - something that only a very brave madman would even try to predict. Sure, statistics can help, but relying on them completely for policy is a disaster waiting to happen.

Afficionados of data analytics will scoff at my comments, of course, since computers are designed to give us all the answers, before we even know the questions. They are far better than humans and can provide invaluable insights into buying and selling patterns.

But I simply say that humans, by their nature, are fickle, unpredictable and generally do the thing no one has predicted. In a way, it's a like betting on a horse race. Statistical data is the favourite and generally wins most races, but what makes life interesting is that the 100/1 shot sometimes romps home first, for no valid, logical reason.