Every marketing person has been hounded by the term ‘big data’. In case you have not heard, big data is seen as the answer to successful marketing nowadays. One wonders whether it will end up being one of those management fads we had got accustomed to having in the 1980s and 1990s, to be forgotten by the time we moved into the 2000s. Anyhow, big data is here with us and we need to understand it because it does have its benefits. However, I would like to state at the outset big data is not enough.
The concept is that by looking at past consumer behaviour, following consumers as they browse through the internet, and analysing demographics, marketers believe that they can understand consumers fully. So the more data sources one has and the more the data can be integrated, one would learn all one needs to know about consumer behaviour. Someone did make the famous claim some eight years ago – “with enough data, the numbers speak for themselves”. However, is that so? It can be said to be so as traditional market research is based more on what consumers claim to do rather than on what they are actually doing.
I do not believe so. Let us take one simple aspect of big data. It tells us what customers have done in the past, which is indeed useful information.
Big data is not enough and canonly be useful if integrated with marketing research expertise
However, does that information provide knowledge? Is past behaviour a guarantee of future behaviour? Does past behaviour tell us what consumers are thinking or why they behave the way they do? Does it anticipate trends? And what about latent needs or unmet needs? To put in theoretical terms, can statistical correlation take the place of causation?
For example, big data will show what I bought on the internet or what I bought at a supermarket and link that purchase with my demographic profile. If I made the purchase on the internet, it would even tell you what other pages I browsed before I got to make the purchase.
Bit it does not tell you whether I would have chosen another product if it were available. Nor does it tell you why I chose that particular product or how I wish to use it. And what if I decide not to make the purchase? Big data excludes from its analysis potential consumers.
Thus it provides information on my behaviour but it does not explain my behaviour. This is why big data is not enough and can only be useful if integrated with marketing research expertise.
Big data is also being threatened by the need for data privacy. Surveys (ah! marketing research again) show that demand for privacy has never been greater, that data privacy is the next big consumer rights issue and that the greatest sin that a company can commit in the eyes of the public is losing their personal data, followed by selling their personal data. So what happens to big data if consumers opt out from passing on their personal details or its use becomes so much controlled that it would lose a great deal of its value?
The nightmare of market researchers is sampling error and sampling bias. Today there is nothing that provides comfort that big data is free of these errors and biases.
Maybe one needs to read once more the story of the 1936 presidential elections of the US and the ability of George Gallup to predict correctly the results. He based his results simply on surveys, using an appropriate methodology.
The publication Literary Digest also sought to predict the results using a methodology that is not dissimilar to today’s big data – a questionnaire among its subscribers. Literary Digest was way off mark in its predictions.
As with any other tool, big data has its benefits if it is used well. However, it needs to be understood that data do not speak for themselves, no matter how much is available.