Ask an organisation what’s stopping it from doing Marketing Mix Modelling (MMM) well, and the answer is almost always “data” or “the right tool”. Those two matter, but they’re rarely where a programme actually goes wrong. I think about MMM readiness across four disciplines: people, process, data and tooling. Get the first two wrong and the best data and the best tool in the world won’t save you.
People
Before anything else, three questions need honest answers inside the organisation. Do we understand what MMM actually is? Do we understand what we want to get out of it? Do we understand what it can do, and, just as importantly, what it can’t? If the honest answer to any of those is no, that’s not a reason to hire a better vendor. It’s a reason to hold off, because no amount of technical delivery fixes a gap in understanding at the level the model’s outputs are meant to be used.
Process
This is the discipline I see overlooked most often. Everyone plans for the data and the tool. Almost nobody plans for where the resulting insight is actually meant to sit inside the business. You’ll generally land in one of two positions. Either your existing decision processes need to change, because MMM is about to hand you insight nobody currently has a slot for, or, just as often, the MMM programme itself needs shaping to deliver the right insight at the right moment in a process that already exists and isn’t moving.
The exercise that actually solves this is straightforward to describe and rarely done: map your marketing decision processes end to end, then note what input is needed at each point in that map – what question is being asked, and by when. Design the MMM programme working backwards from that map, not forwards from whatever data happens to be easiest to collect. A model that produces excellent insight nobody asked for, delivered after the decision it would have informed has already been made, has failed, however good the statistics. See Start small, then get bigger for what this looks like in practice.
Data
If you haven’t started collecting the data yet, the best time to start is today. I see this constantly: an organisation wants to dip a toe into MMM and discovers it doesn’t have enough historical data, not because it hasn’t been operating, but because nobody has been retaining the data that operating produces. There’s no organisational memory to draw on.
That is not a reason to delay your first MMM project. In fact, your first project is usually where these gaps get found – which questions can’t currently be answered because the data to answer them doesn’t exist yet. The job is to name those gaps precisely, and then start closing them, understanding that this is generally a gradual process, not a one-off fix.
Tooling
There is no shortage of ways to build a Marketing Mix Model. At one end, there are strong open-source packages that most competent data scientists can pick up and use to build something statistically robust. At the other, there is a growing set of dedicated MMM platforms – I’ve built tools in this space myself over the years – alongside consultancy-run portals, of the kind firms like Ipsos offer their clients, where you log in to review outputs someone else has produced for you.
The tooling is a continuum, not a single choice, and the right point on it depends far more on your answers in the first three disciplines than on which package has the best marketing. Understanding which tool suits your organisation takes real judgement, considerably more than picking up the package itself does – the same defensibility test that applies to a vendor selection applies here too. It’s a question we’re happy to help organisations work through, without a stake in which answer they land on.
None of the four disciplines is optional, and none substitutes for the others. An organisation with brilliant data and the wrong tool will get somewhere. An organisation with the right tool and no idea where the insight is meant to land in its decision-making will not.
If you’re weighing up where your organisation actually stands across the four, drop a line to hello@themmmdoctor.com.