Short answer
Do not buy an MMM if you spend under roughly $5–10M a year on media, if your spend has been flat so there is nothing to learn from, if your data cannot support it, if nobody will own it after launch, or if the decision is already obvious. In four of those five, one match-market test is the better purchase.
We sell this work, so a page arguing against buying it needs a reason. Simple: the fastest way to make a client unhappy is to build something they were not ready to use.
You do not spend enough yet
Under about $5–10M a year in media, this is usually a waste of your money. One or two channels does not need a model — the platform’s own reporting gets you most of the way.
Above that, things change. You are probably oversaturated in one or two channels, sitting on easy wins in others, and missing channels you never tried.
Your spend has been flat
A model learns from change. Flat spend for three years contains no evidence about other spend levels. You will still get a curve — it will be an assumption with error bars.
Your data cannot support it yet
- Spend is monthly, not weekly. That halves your data points.
- Channel names changed halfway through, so "paid social" means different things in different years.
- Offline spend lives in an agency spreadsheet nobody checked against the books.
- Your KPI is defined two ways in two systems and nobody picked one.
None of these mean never. They mean spend a month fixing data before spending six figures modeling it.
Nobody will own it
A model is a quarterly tool. Someone has to refresh it, read what changed, and argue for moving the money.
“Ask who runs the quarterly refresh. If the room goes quiet, the project is not ready. No vendor can fix that.”
Is there a hard spend cutoff?
Roughly $5–10M a year is where it starts to make sense, but it depends on how concentrated your spend is and how much is offline.
What if leadership is demanding an MMM?
Run a test first and bring back a real number. That usually satisfies what they actually wanted — credible evidence, not a particular method.
Can we try open source to see if we are ready?
Yes. It surfaces every data problem you have at no license cost. If the attempt stalls on data, that answered the question.
Does this apply to market testing too?
Much less. Testing needs regional sales data and the willingness to hold a channel back. Most companies not ready for a model are ready for a test.
Where we’re different
Click a claim. We’ll show our work.
Marketing mix modeling is a time-series problem. So we hide the last three to six months from the model, then make it predict them. If it cannot get under 10% error on months it never saw, we do not ship it — and you get your money back. Ask anyone else on your shortlist for that number. Most of them do not have it.
We are data scientists, not a software company. We do not sell a tool that competes with the vendors on these pages. That is why we will tell you when one of them — or a free open-source framework — is the better answer for you.
Book the two-week auditFind your wasted third
Want this run on your own numbers?
The two-week audit shows you where each of your channels sits on its own curve — what to cut, what to grow, and how much budget is sitting in the wrong place.
Book the two-week auditKeep reading
- Recast alternatives: what else solves the same problemRecast sits in a specific spot: Bayesian MMM as a managed service. The real alternatives, grouped by what you would actually be choosing between.
- How to find the point where your channel stops paying offWhat data you need, which method fits your budget, and how to check the answer against the real world before you move money.