| July 27, 2020
As this blog nears the end of its existence I cannot help but reflect on the different themes that run through its 14 years of content – don’t worry, this is not another post about the importance of perceived difference to brand growth – however, it is a related subject: how often brand marketers get led astray by data.
It seems strange to in this day of data-abundance and machine learning to talk about data leading people astray, but nevertheless it happens all too often. Ignoring the issues of data quality – which is hard to do given the messy nature of most freely available data – the biggest problem with data is that we assume too much and question too little. We want simple answers to complex questions and are not willing to think about what our data really represents or how it should be analysed.
With this in mind, here are my top three reasons why brand marketers get led astray by data.
We analyse the data we have, not the data we need
Back in 2006 I wrote a post titled Incomplete, dirty and not even free (but it’s the only information we have!) which warned that all research is an approximation and that the reliability of the approximation is directly related to the cost of obtaining the information involved. However, my original post talked about a very granular aspect of data capture, that of getting enough data to be reliable. But what about the data we are simply missing? I would argue that brand builders have been led astray for decades – since the advent of scanned sales data – because they have focused too much on what people do, not why, and that, if anything this problem is worse today than it has ever been because there is so much more data that tells us what people do, see, say and even feel from sales data, media data, search and social data.
Unless you bring all that data together you will have an incomplete view of what is happening to your brand. Even then, the one place where a complete view of your brand exists is in people’s heads. If you want to know why people buy the brands they do, and why they say what they do, you must do more than observe them using passive data, sometimes you have to dig deeper and ask them questions. Yes, words are imperfect signals of meaning, but at least you have a chance to get beyond the obvious and uncover a real insight that offers the potential to unlock significant growth, not just marginal gains.
Brands are built over time, but a lot of analysis ignores time as a variable
This is the one that is insidious and has led countless brand builders to focus on the wrong metrics, often to the detriment of their brand’s profitability and future innovation and marketing. Why on earth do we assume that point in time data can tell us everything we need to know about how brands grow over time? It’s like looking at one scene from a movie and assuming that you can figure out the entire plot. It is not that point in time data is useless but without a good framework for understanding how brands over time to inform your analysis it is all too easy to miss important findings.
Perhaps the worst offenders when it comes to ignoring the influence of time are the simplistic attribution models that do not take account of prior influences on behaviour when optimizing media spend. In theory, these models optimize all the time but they do so based on what is happening in the moment, so in essence they are a stream of snapshots not a continuous flow of data. As Simon Peel, Adidas Global Media Director, demonstrates in this video taking time into account using market mix modelling can give you a very different understanding of what is driving sales.
Brand builders need to take account of much longer time frames than the activation cycle. They need to look across the interpurchase interval which dictates the rate at which existing users have the opportunity to buy again or not. And they need to look beyond this year to the people who do not even know they will need the product category today but likely will do in future. Which is why Kantar uses a balanced attribution approach to try to give marketers a more holistic view of what drives behaviour and combines both short and long-term influences on sales to understand total marketing ROI.
Our own assumptions create the lens through which we look at data
Over the life of this blog I have sounded off about assumptions made from data that I felt were more informed by assumption than empirical proof (the effectiveness of how companies use search terms and behavioural data to target people has been a perennial favourite).
Unfortunately, we all suffer from this problem, me included. Over the years I have been indoctrinated to view the world in a certain way. Smart people have told me things, I’ve read learned journals and explored different data sets, and all that experience has given me a certain viewpoint which guides my understanding of how brands grow over time. This understanding means that when I hear that a brand has grown or gone bankrupt one of the first things I look at it BrandZ to see what the brand equity data has to say, potentially ignoring the fact that brand success or failure is informed by many more things than what people think and feel about it. Only when I have ascertained that there is no clear smoking gun in the equity data do I start to consider other factors, and when there is clear evidence of equity changes being related to growth or decline I have to force myself to ask, what else might be going on, is this a cause or an effect?
So those are my top three reasons why marketers get led astray by data. And yes, they are the product of my own experience and viewpoint, so I believe they are important!