By Kevin Gray and Byron Sharp
Marketing scientist Kevin Gray asks Professor Byron Sharp of the Ehrenberg-Bass Institute about key mistakes many marketers are making.
KG: Big Data, under various definitions, has been getting a lot of buzz in recent years. Targeted online ads are now commonplace and Customer Relationship Management (CRM) is big business. Do you have thoughts on each of these you’d like to share?
BS: There is quite a bit of confusion about the term Big Data. It doesn’t mean ‘lots of numbers’ i.e. any large data set, it refers to continuous data that flows in usually from unobtrusive measurement. So clicks on webpages, calls to call centers, TV audience data, that sort of thing. The key attraction for marketers is to use this data to tailor offers to reach people who are good prospects, especially people who are interested in buying now. This sounds very powerful, but it’s been over-sold, the power has been over-estimated. Take Amazon for example who have been in this business for longer than most. They have data on your past purchases, plus vast amounts of continuously updated data on what other products were bought by people who bought items you bought. So real hard behavioural data, sounds exciting.
Yet look at Amazon’s recommendations for yourself, I think you’ll find they are pretty unimpressive, usually rather obvious recommendations e.g. another book by the same author. A lot of the time they simply recommend books you’ve already bought, for me they even recommend books I’ve written! It’s not that Amazon are stupid, or they haven’t hired the right quant jocks, it’s that there are severe limits to the ability of complex models to produce valid predictions. Adding a whole bunch of weak correlations together does not improve predictive power. So don’t let someone try to ‘pull the wool over your eyes’ by claiming that artificial intelligence or machine learning can overcome this fundamental problem, they can’t.
Possibly the most useful piece of big data driven targeting is simply display ad retargeting. If you went to an airline’s website and looked up flights to Rio then for the next week ads for that airline and its flights to Rio will follow you around other websites you browse. This works, sometimes, because it keeps nudging us to do something we got halfway through before being distracted. It’s pretty basic, but that’s probably why it can work.
Otherwise the strategic assumption that hyper-targeting is sensible, let alone possible, is unrealistic thinking. I’ve explained why with many different examples in my books.
KG: Any critic of the status quo is likely to encounter resistance. What kinds of reactions do you receive to the things you mentioned earlier? Do you feel you and your colleagues at Ehrenberg–Bass are being misinterpreted or misconstrued in any way?
BS: Mainly the reaction has been extraordinarily positive. Marketers are hungry for evidence, after being starved for so long. They are not upset by seeing popular theories fall as the evidence comes in. It’s harder for some marketing ‘gurus’, consultants and academics, because they have based their reputation on selling fantasy theories. But even some of these people are big enough to say ‘OK I was wrong, in fairness my reasoning appeared sound but, hey, the world turns out to be a surprising place’.
Yes, Ehrenberg-Bass Institute discoveries do sometimes get misreported, or misinterpreted. Probably the most common is to say we discovered that there is no such thing as brand loyalty – when page 92 of How Brands Grow explicitly says ‘loyalty is part of every market’, and our textbook explains with lots examples how and why loyalty is natural behaviour. It’s easy to understand how people who haven’t read the books could get the wrong idea – ‘a little bit of knowledge can be a dangerous thing’.
KG: Finally, the way forward…How do we get people to change their ways?
BS: Marketers, like consumers, are habitual and hence change is difficult. That’s not to say it doesn’t happen. First, it requires expose to the evidence, and then re-exposure and re-exposure accompanied by thinking and doing – then insight builds (‘ah ha’ moments) as well as evidence-based skills. It does happen. Just look around at the impact of science on other disciplines (like medicine and engineering). The same will happen to marketing.
KG: Thank you, Byron!
Kevin Gray is president of Cannon Gray, a marketing science and analytics consultancy. Byron Sharp is Professor of Marketing Science at the Ehrenberg-Bass Institute, University of South Australia. He is the author of How Brands Grow and the textbook Marketing: theory, evidence, practice, and co-author of How Brands Grow: Part II – all published by Oxford University Press.
