The Limits to Marketing Analytics

From what you read today, you would think that having more customer data and better analytical tools is the answer to every marketer's dream. With more data and better tools, marketers could develop better statistical models, which they can use to target customers better, with better offers, and with better timing. Marketing would finally be a hard science and customers would just hand their money over, resigned to the fact that those canny marketers knew them better than their spouses.

But is more data and better tools really the answer? Or are they false marketing gods?

Let's get one thing straight. I am not against analytical marketing. I have helped many companies gather more data, develop better tools and do better marketing. The fewer data that companies have and the fewer tools that they use, the more they can be helped. One of my financial services clients is currently achieving a 30% campaign response rate through this approach.

But what about more advanced companies, the ones who already collect lots of data and throw a range of analytical tools at it. The analysts cry that if only they had more data and better tools they could do even better. Implicit in their cries is the assumptions that if they could gather the right data, and have the right tools at their disposal, that they could develop better campaigns.

All three assumptions may be wrong.

The Wrong Data

More advanced companies gather extensive transaction and customer demographic data. But they know hardly anything about the circumstances in which customers purchase, the motivations behind them and the influencers of customer behaviour. We now know that much of customer behaviour is driven by sub-conscious cognitive pathways which take the mental effort out of hum-drum things like shopping. So that it can be used for more interesting things like living life. That's why brands and branding are just as important today as they have ever been. A good brand makes the shopping decision easier. It reduces cognitive complexity and the brand gets bought. Companies need to gather very different data to supplement the data they already gather. This requires new collection approaches, such as ethnographics and other field techniques, in addition to the standard approaches. Some companies are investing in gathering this softer, more qualitative, contextual data, but they are still in the minority.

The Wrong Tools

The same companies have invested in suites of analytical tools to analyse their transactional data. They assume that if you bought A, statistically, you are more likely to buy B as well. But that the two are not related to each other. They assume that customers live in a world of independent data and Gaussian statistics. As a recent posting by John Hagel on his Edge Perspectives blog points out, the real world is increasingly driven by dependent data and long-tail or power-law statistics. This dependant world is a sign of our networked times, where everything from the choice of your mobile handset upgrade to the choice of your next automobile is heavily influenced by the opinions of friends and family. Companies need to understand how the data they have relates to other data. Whether it is really independent and to use the right statistical approaches. Again, some companies are investing in analysing data with influence networks in mind, particularly mobile telcos, but they are also in the minority.

The Wrong Campaigns

One of the mainstays of analytical marketing is the champion-challenger approach. The best campaign is the champion. Subtle changes are made to the campaign and tested in controlled marketing experiments against it. These are the challengers. The best of the current champion and the challengers becomes the new champion. But this assumes that the basic campaign is already right and that it just needs to be improved to reach the champion of champions position. We now recognise that in today’s fast-moving customer environment, the likelihood that a particular campaign structure is right rapidly falls over time. The longer a basic campaign has been running, the less likely it is to be the best one available. As a recent article by Duncan Watts in the Harvard Business Review points out, companies need to actively search for the best new campaigns in addition to operating champion-challenger for their current best ones. Again, some companies actively search across the marketing fitness landscape for the best campaigns, but, well, you know the rest by know.

The wrong data, analysed by the wrong tools too often turn into the wrong campaigns. With response rates for direct marketing campaigns averaging 2.61% according to a recent DMA study, it is pretty clear that analytical marketing has a problem. Maybe a radical rethink is the answer. That doesn’t mean throwing away the tried and trusted approach to analytical marketing, but it does mean adding new tools, techniques and approaches to it. To bring it into the complex networked world we live in today.

What do you think? Is analytical marketing a victim of the winners curse? Or are things just doing fine, thank you?

Post a response and get the conversation going.

Graham Hill