During my time at Accenture, the marketing press spent a few years genuinely excited about a single prediction: Gartner’s 2012 forecast that by 2017 chief marketing officers would spend more on technology than chief information officers. It was reported as an inflection point. Marketers, the argument went, were about to matter more, because they would soon control more technology spend than anyone else in the building, IT included.
I found the excitement odd at the time, for two reasons the coverage never seemed especially interested in. First, whether that technology spend was actually adding anything incremental – a genuinely new capability, a genuinely better outcome – or whether a good deal of it was simply funding the maintenance of systems already bought. Second, and more basic, whether the money was supporting marketing at all, or whether it was propping up other domains entirely – data platforms, customer service tooling, enterprise infrastructure – that had ended up filed under “marketing technology” because that happened to be where the budget line sat.
None of this is an argument about whether marketing’s budget is, or should be, large. It’s a recognition of something else: the more central technology becomes to marketing, the more tempting it is to measure marketing by the things technology makes easy to measure – spend, reach, click, channel-level return – and the less attention gets paid to the thing hardest to put a number on and, not coincidentally, the thing that separates good marketing from great marketing. The creative itself.
Marketing Mix Modelling (MMM) has the same blind spot, and the honest position is to say so plainly. A model can tell you, with real precision, that a campaign produced incremental volume – how much, over what period, at what marginal return. It cannot tell you why. Two campaigns can run with identical spend, identical targeting, an identical channel plan, and produce wildly different measured effects, and the model will never explain the gap. The variable most likely doing the explaining – whether the advertising itself was any good – rarely enters the regression unless it’s deliberately split out, and even then the model shows that the creative differed, not why. It sits outside what the technique was built to see.
That creates a temptation in both directions, and a good measurement practice resists both. The first is to let a strong result be read as proof the media plan was right, when the true driver was that the creative simply worked – a channel gets the credit an idea deserves. The second, subtler temptation runs the other way: because creative quality resists quantification, it gets quietly written out of the conversation altogether, treated as a soft input alongside a set of hard, model-backed numbers, when it is very often the largest lever in the room.
Being an honest broker means saying, without hedging, that Marketing Mix Modelling measures what happened and stays silent on why – and that creativity is very likely doing more of the “why” than any coefficient in the model will ever show. It means not dressing up a model’s silence on creative as evidence that creative doesn’t matter, and not pretending a model can score an ad’s quality just because it can score its effect. Where creative quality genuinely needs to be understood, that is a different job entirely – pre-testing, in-market creative experiments, a proper qualitative read – run alongside MMM, not squeezed into it.
A good model will tell you a campaign worked. It was never going to tell you the ad was good. Both are true, and a measurement practice worth trusting keeps them separate.
Knowing what a model can and can’t see is part of a wider readiness question – see The four disciplines of Marketing Mix Modelling.
If your read-outs have started crediting the media plan for what the creative actually did, drop a line to hello@themmmdoctor.com.