IT industry analyst behemoth Gartner is having their Symposium shindig this week in Orlando, where they made such predictions as “one in three jobs will be taken by software or robots by 2025” and “By year-end 2016, more than USD 2 billion online shopping will be performed exclusively by mobile digital assistants,” among other deep and unquestionably thoughtful prognostications.
And of course, Gartner isn’t the only analyst firm who uses their crystal ball to make news. Forrester Research and IDC, the other two remaining large players in the IT industry analysis space, also feed their customers – as well as the rest of us – such predictions of the future.
Everybody knows, however, that predicting the future is never a sure thing. Proclamations such as the ones above boil down to matters of opinion – as the fine print on any Gartner report will claim. And yet, at some point in time, such claims will become verifiable matters of fact.
The burning question in my mind, therefore, is where are the analyses of past predictions? Just how polished are the crystal balls at the big analyst firms anyway? And are their predictions better than anyone else’s?
If all you hear are crickets in response to these questions, you’re not alone. Analyst firms rarely go back over past predictions and compare them to actual data. And we can all guess the reason: their predictions are little more than random shots in the dark. If they ever get close to actually getting something right, there’s no reason to believe such an eventuality is anything more than random luck.
Of course, anyone in the business of making predictions faces the same challenge, dating back to the Oracle of Delphi in ancient Greece. So what’s different now? The answer: Big Data.
You see, Gartner and the rest spend plenty of time talking about the predictive power of Big Data. Our predictive analysis tools are better than ever, and furthermore, the quantity of available data as well as our ability to analyze them are improving dramatically.
Furthermore, an established predictive analytics best practice is to measure the accuracy of your predictions and feed back that information in order to improve the predictive algorithms, thus iteratively polishing your crystal ball to a mirror-like sheen.
So ask yourself (and if you’re a client of one of the aforementioned firms, ask them) – why aren’t the big analyst shops analyzing their own past predictions, not only to let us know just how good they are at prognostication, but to improve their prediction methodologies? Time to eat your own dog food, Gartner!
predictive analytics, Gartner research, big data analysts