Opens in a new tab

Marketing Mix Modelling Glossary: Key Terms Explained

Published-
November 11, 2026
IMS graphic for the Marketing Mix Modelling Glossary, showing a stack of glossary cards

Marketing Mix Modelling has its own vocabulary, and much of it gets thrown around in meetings without explanation. This is a working reference for the terms that come up most, written in plain English instead of statistical notation. Where a fuller explanation exists, we’ve linked to it.

The basics

Marketing Mix Modelling (MMM)

A statistical method that estimates how much of a business’s sales came from marketing and how much came from everything else (price, seasonality, distribution, the economy) by analysing aggregated data over time. What Is Marketing Mix Modelling? has the full explanation.

Base sales

The sales a business would expect with no marketing at all, driven by existing brand strength, repeat purchase, distribution and underlying demand. MMM separates base sales from incremental sales so marketing isn’t credited with demand that would have existed anyway.

Incremental sales

The sales caused by a marketing activity, on top of base sales. This is usually the number a business actually cares about, because it reflects what marketing is adding, not just what happens to be correlated with it.

Media mix (or channel mix)

The combination of channels a business uses, such as TV, paid search, paid social and out-of-home, and how budget is split between them. “Optimising the media mix” means adjusting that split to improve overall return.

Statistical concepts

Adstock (carryover effect)

Advertising doesn’t stop working the moment it stops airing. A TV campaign that ran last month can still influence sales this month, at a fading rate. Adstock is the term for modelling that lingering effect.

Saturation curve (diminishing returns)

A curve showing how the effectiveness of spend in a channel changes as investment grows. Most channels are efficient at low spend and become less efficient as you add more, which is the classic shape of diminishing returns. Knowing where a channel sits on its curve is central to deciding whether to spend more or less there.

Response curve

A broader term for the relationship between marketing input (spend) and business output (sales). The saturation curve is one common shape. MMM uses response curves to answer “what would happen if we spent more, or less, in this channel?”

Elasticity

How sensitive sales are to a change in a given input, usually expressed as a percentage. Price elasticity describes how much sales change for a given percentage change in price. Marketing elasticity works the same way for spend.

Multicollinearity

A statistical problem that arises when two or more variables move together so closely that a model can’t separate their effects, for example if TV and paid social spend always rise and fall in the same weeks. It’s one of the main reasons MMM is hard to do well, and one reason Bayesian methods have become popular. See Bayesian Marketing Mix Modelling Explained.

Bayesian terms

Prior

A starting assumption about how a variable is likely to behave, based on past experience or general knowledge, before the model has looked at this dataset.

Likelihood

What the data itself suggests about a variable, taken on its own, without reference to any prior belief.

Posterior

The updated estimate you get once a prior and the observed data are combined. In Bayesian MMM, the posterior is usually shown as a range, not a single number.

Alternative and complementary measurement

Attribution

A measurement approach that assigns credit for a conversion to one or more touchpoints in an individual customer’s journey, such as a paid search click or a social ad view. MMM vs Attribution vs Incrementality explains how it differs from MMM.

Incrementality

A measurement approach focused on proving what additional outcome a piece of marketing caused, usually through a controlled experiment instead of observational data.

Holdout test

An experiment where marketing is deliberately withheld from part of the audience, or from a set of regions, so the difference in outcomes between exposed and unexposed groups can be measured directly.

Geo experiment (geo-test)

A type of incrementality test that varies marketing spend across geographic regions, then compares results between them to estimate the causal effect of that spend. We run these as part of our geo testing service.

Triangulation

Combining MMM, incrementality testing and attribution, instead of relying on a single method, so each approach’s evidence can check and sharpen the others. It’s now the widely accepted best practice in marketing measurement.

Making the model useful

Marginal ROI vs average ROI

Average ROI is the return across all spend in a channel to date. Marginal ROI is the return on the next pound spent, which is usually lower than the average once a channel starts to saturate. Budget decisions should generally rest on marginal ROI, not average.

Scenario planning

Using a model to estimate the likely outcome of a hypothetical decision, such as raising one channel’s budget by 20%, before you make it.

Budget optimiser

A tool, often built on an MMM’s response curves, that recommends how to reallocate a fixed budget across channels to maximise a chosen outcome, such as revenue or profit.

Model validation

Checking whether a model’s results hold up, by comparing predictions against real outcomes, testing on data the model hasn’t seen, or cross-referencing against experiments. A model that hasn’t been properly validated shouldn’t be trusted, however impressive it looks. How to Evaluate an MMM Vendor covers what to ask.

Frequently asked questions

Is “Marketing Mix Modelling” the same as “Media Mix Modelling”?

In practice, yes. The two terms are used interchangeably and both are shortened to MMM. Some practitioners use “media mix” to mean the channel-spend side of the model specifically, but the underlying method is the same.

Do all MMM vendors use these terms in the same way?

Mostly, but not always. A few terms, particularly around validation and optimisation, are used a little differently between vendors and tools. If a term in a proposal or report isn’t clear, ask the vendor to define it in the context of your own model.

Where should I start if I’m new to this?

What Is Marketing Mix Modelling? covers the fundamentals without assuming any background. From there, Bayesian Marketing Mix Modelling Explained and MMM vs Attribution vs Incrementality go deeper on two areas that come up constantly.

If a term you’ve heard isn’t listed here, or you want it explained in the context of your own data, get in touch with IMS. You can also read about our marketing mix modelling service.

Background
Magda Pietrzykowska
Associate Analyst
I bring expertise in data analytics with a background in design engineering, venture capital, and data science. I enjoy presenting my work to clients and engaging with industry peers at conferences. My multi-disciplinary background and passion for data analytics provide unique insights into the latest trends and developments in data analytics.

Transform Your Data Into Your Most Valuable Asset
Connect with us and discover how our tailored analytics solutions can drive business success.

Contact IMS
Background