When it comes to Big Data, nothing in the commercial sphere can compare to the unimaginable 30 petabytes collected annually by the Large Hadron Collider at CERN in Geneva. (A single petabyte is 1m Gigagbytes – roughly equivalent to four times all the data contained in the US Library of Congress, or the combined memory capacities of around 800 human brains).
But as many businesses have discovered, however big a pile of noughts and ones you’ve collected, it’s not much use if you can’t figure out what it’s all trying to tell you.
‘Big data has been a business buzzword for years, companies collect tons of data but it’s clear that it is not yet delivering value. You do need a database but that’s just the start’ says Prof Michael Feindt, founder of predictive analytics business BlueYonder and a former CERN researcher who has used his experience of interrogating masses of experimental data to bring the academic rigour of particle physics to the world of e-commerce.
Prof Feindt will be appearing at MT Live in London on June 23rd, talking about how to extract business value from big data. Rather than the data itself, the secret to extracting that missing value is insight, says Feindt. And insight means asking the right questions. ‘You need algorithms to find the answers to the questions you are really interested in. Many software vendors are really only selling a database, but it’s the algorithms that really matter.’
SEE PROF FEINDT - PLUS LORD BROWNE, SIR MARTIN SORRELL AND MANY MORE - AT MT LIVE, LONDON, ON JUNE 23rd
Feindt’s company BlueYonder uses a combination of human brainpower and machine learning automation to help its clients – in online retail, insurance and healthcare among other sectors – to maximise their sales and profitability through dynamic pricing. That means changing prices on the fly depending on data inputs ranging from sales history and social media feeds to the stockmarket and even the weather.
The results can be impressive – from a German retailer which saved 20m Euros on wastage of fresh sushi (with a shelf life of only two days one of the most perishable of perishable goods), to the online fashion retailer which has banished end of season sales because thanks to predictive analytics it no longer has overstock to be got rid of.
The right algorithms and good implementation can result in boosted turnover, market share and profitability, he says, as well as reduced returns (a big problem for many e-tailers). ‘It is pretty hard to do all those things at once according to the conventional rules of doing business.’
Ultimately he reckons that 90% of the decisions a typical online retailer makes could be automated – but that doesn’t mean that the machines are about to take over. ‘I have no fear that computers will end up ruling the world, that’s rubbish. Clever algorithms can beat humans, but you have to have human knowledge to come up with the algorithms in the first place.’