Marketing Mix Modelling (MMM) means different things to different people, and a lot of badly fitted engagements start with nobody having agreed what it actually is.
Take a business with one sales channel – everything happens online – running some activity on Meta and a pay-per-click (PPC) plan on Google. For that business, a straightforward attribution analysis is not a compromise. Attribution, sense-checked with platform lift tests, is usually enough. Building a full Marketing Mix Model for that kind of setup is solving a problem the business doesn’t have, with more data, more time and more uncertainty than the question warrants.
MMM starts to earn its keep somewhere else entirely. I find it useful to think of marketing decisions as sitting on a pyramid, with three layers: strategic, tactical and operational. Strategic sits at the top – the once-a-year reset, where budget is set across brands, categories and major channels for the year ahead. Tactical sits below it – the quarterly update, refining that allocation as the year plays out. Operational sits at the bottom – the week-to-week, sometimes day-to-day decisions about which specific ad, bid or creative variant is working right now.
Marketing Mix Modelling is most valuable at the top two layers of that pyramid, and largely useless at the bottom one. It is built to answer meta-level questions: where should the next slice of budget go, across products, brands, channels and tactics, when you have real prioritisation decisions to make and genuine trade-offs to weigh. That is exactly the kind of question a business running true multi-channel campaigns, possibly across several categories, needs answered once or twice a year, properly, with the full picture in view.
It is not built to tell you which of the ten display ads you’re running should be turned up this week and which should be turned down. That is an operational question, and it doesn’t need a strategic instrument – it needs the kind of fast, granular optimisation that platforms already do automatically, in something close to real time. AI-driven bidding and creative tools are automating exactly this layer as things stand, and they are, frankly, better suited to it than any MMM ever will be.
So the honest scoping question, before anyone commissions anything, is not “do we need better marketing measurement?” It is “which layer of the pyramid is the actual decision sitting on?” If the answer is a single channel with a straightforward funnel, attribution will serve you well and MMM will be overkill. If the answer is an operational, week-to-week optimisation question, the platforms already have you covered. Marketing Mix Modelling is for the business asking a harder, slower, more consequential question: given everything we know, where should serious money move next, and why.
Once you know MMM is the right layer, the next question is how often to re-run it – see Marketing Mix Modelling is not an operational tool.
If you’re not sure which layer your question sits on, that’s a reasonable thing to find out before commissioning anything. Drop a line to hello@themmmdoctor.com.