Big Data – How much is hype?
Big data is certainly the talk of the data industry – see how popular it has become on google searches over the past year:
What is it?
Basically, it’s the effluent of the digital world. That’s not supposed to sound as bad as it does: Big data is the by-product of the digital age – it’s the information left over from online consumer behaviour.
Why is it called Big Data?
The term big data arises because there’s so much of it. Literally terabytes of the stuff. Enough to theoretically make analysts like us very happy. But it doesn’t quite do the job as there’s too much of it – and I didn’t think I’d ever say that.There is so much of it, standard super-powerful data software (SAS, etc) is struggling to handle it. There are now new processor sharing capabilities such as Hadoop, which move closer to being able to manipulate it. They work by sharing processing across different systems to allow the computing to be done.
What can you do with it?
This is another big stumbling block, as the data needs to be sifted to get to the insight – not an easy task.There have been a few interesting examples, such as with Linkedin – they increased their members networks by suggesting connections with the ‘people you may know’ function. This was the brainchild of Jonathon Goldman, a data scientist who saw the opportunity to use data to suggest connections – the upshot was a new growth trajectory for Linkedin and millions of new page views. However, for the most part big data is not leveraged to good use – sitting around taking up lots of expensive storage space.
Flash in the pan?
Does big data live up to its hype? This interesting graph from Gartner (click here) suggests we are reaching the crescendo of the hype. Now it’s a case of working with the smaller part of the Big Data that’s usable and seeing what is possible to build from there.
For marketing analytics, we see the major benefit to stem from access to more detailed data, both on the channel side and performance side. This will mean that major opportunities will arise in analysing much more detailed impacts and the ability to assess performance across different consumer types. For example, instead of looking a TV’s total ROI, it could be broken down to assess the ROI of copy rotation, day of week, time of day, channel (sales house), and so on.
From a consumer perspective, the opportunity to split data down to a less aggregated level and segment market mix models by customer profiles will lead assessment of marketing ROI by the groups that deliver more value.
Useful, but doesn’t solve everything
So there is significant opportunity here, but possibly not quite as much as initially hyped. For big data to be useful you need the right data capture, good data analysts/data scientists, senior project sponsors and a bit of patience – not a small order…
Written by Michael Cross
Brightblue is a specialist marketing ROI consultancy. Our experience ranges from consulting to market mix modelling (econometrics) to global media budget setting and optimisation. We take a clear, dynamic and commercial approach to ensure results are used to maximum effect.