Measuring the influence of brand marketing on sales

by Guest Contributor Bill Pink | October 29, 2018


Bill Pink
Kantar Millward Brown
Head of Brand Guidance Analytics in North America

Effective brand marketing increases the probability of consumers either picking a brand or paying more for a brand when it is time to buy.  But what data and analytic approaches help us to understand the influence of brand marketing on sales?

We know that brand and sales are not the same thing, and we know there are many influences on consumer choice beyond branded memories and associations such as availability, discounts, new product entrants, and changes in consumer needs. But even after controlling for these influences we still find it difficult to isolate the contribution of brand marketing. How can we do better?

The answer lies in going back to some first principles of analytics and putting them in a brand context. If we could build the ideal data set to understand brand impact, that data set would need to observe consumer choices and feelings at a highly granular level and cover a variety of users. Here is why:

First, granularity is key because we know different consumers are influenced differently by messages, product benefits, pricing, etc. When our data exists at a granular level, we can easily see these differences in action. Contrast this with analytics run solely at a macro or trended level. We may see flat trends in aggregate but we would miss movements within particular consumers, or groups of consumers, that average out to no change. Granular data uncovers these patterns, which is why so much of our work in this space focuses on bridging sales and brand data at a segment, region, or household level.


Second, we know that history matters. Ask yourself how often you are looking at two to three years of trended data for a mature brand? Or do a bit of homework and see how long that brand has been in existence in that category and market? I recently tried to understand the impact of brand equity on sales for a brand that was over 70 years old, but we had data going back two years and only at the monthly level. No surprise the data was rather flat and provided minimal variation to leverage. It’s extremely rare to have data on consumers going back so far in time to capture the initial launch, but if we were working with brand and sales data at a granular level then we would have a mix of recurring, new, and lapsed buyers in our data to work with.

Said bluntly, variation is the mother’s milk of analytics. And when trying to use analytics to understand the impact of brand marketing, we need to leverage data assets that allow for granular observations as much as possible to have variation in brand feelings, consumer outcomes, and different brand histories. Do you agree? Please share your thoughts below.


Leave a comment
  1. Robert MacKenzie, ubina consulting, November 28, 2018
    This article seems to be implicitly talking about larger brands which probably have very high ad investments and consistently high brand awareness. In this context, I think that having granularity of data doesn't necessarily help because you're not going to see many consumers who don't already know the brand. The (growing) problem of walled gardens in terms of digital data, and the persistence of ignorance of offline consumer experiences and actions together ensure that a "single customer view", across a representative subset of customers, which is the granular data really necessary to achieve what the author is talking about, will continue to be unattainable for the foreseeable future. Therefore, I believe that some level of aggregation of groups of consumers is unavoidable. In this situation, when media investments are too constant to be able to measure in marketing mix modelling and other methodologies on aggregate data, then a system of testing and learning can be adopted (periods off-air for each media channel, for example) - essentially creating sufficent variation in media plans for modelling to be able to disentangle the impact of different channels and campaigns.
  2. Bill Pink, October 31, 2018
    Indeed, the constant flow of communications from an analytics perspective equates to an explanatory variable that also lacks variation, and makes it harder to parse out the impact of the communications on attitudes and behaviors.  That is another great reason why we need to leverage more granular data to uncover the effects.  Continuously delivered communications may look like a relatively large and flat trend in aggregate, but if you look at the distribution of exposure to the communications at a person level we will see a range of frequency of exposure for each consumer across touchpoints.  The variation in frequency of exposure allows us to unpack the impact of the communications that would be masked in aggregate, which is how our cross media research works.
  3. Jhon van der Ceelen, October 31, 2018
    Thanks Bill for your thoughts. Variation is indeed mother´s milk in analytics and I do see a development for a long time in which this get´s more limited. Brands (larger) have a constant flow of communication, which means less variations and with that less insights in effect. How do you look at this?

    Leave a comment