Fast, flexible MMMs are growing in popularity.  We call this MMM Lite. Let’s look at some of its features and benefits.

Think of MMM Lite as fast, flexible and low-cost econometrics

MMM Lite, is fast, flexible and low cost. These advances are driven by AI and Machine Learning (ML).   MMM Lite is still an advanced form of marketing and media attribution modelling. It includes many factors from both within and beyond the marketing ecosystem. Alongside marketing and media variables, MMM Lite can include seasonality, pricing, distribution, underlying trends, and economic variables like interest rates, inflation and consumer confidence. Unlike MMM, MMM Lite does not examine the impact of all variables in the marketing mix and beyond.

It’s a much better approach than MTA or Rule Based Attribution

Multi-Touch Attribution (MTA) is a form of cookie or tag-based micro measurement. MTA tracks digital touch points – think of them as unique digital fingerprints – at different stages in the path to purchase. Each time a page is visited, a touchpoint is logged. But there are three big problems with MTA. First, MTA needs a tracking mechanic (tag) which creates cookie and privacy problems. Second, because MTA is a digital technique, it is largely restricted to digital channels.  Third, MTA cannot estimate the ROI impact of non-digital investments like cinema, linear radio, out of home, TV and some BTL channels like door drops and DM.

It’s not Rule Based Attribution: Rule-based attribution has been an important part of the digital ecosystem for the last two decades. “Rules” like first click or last click attribute channel performance based on the consumer passing known points in a click-path journey.

Explaining MMM Lite

So, if MMM Lite is not MTA and not rule-based attribution, then what is it?

Well the best way to think of MMM Lite is as a very light and fast form of MMM. It uses the same statistical principles as MMM (different forms of regression modelling, where we measure the influence of multiple independent variables like media spend, distribution and price on a dependent variable like sales), but it focuses mainly locating detectable effects of media spend.  MMM Lite is different to traditional MMM because it tends to slightly under fit models but in doing so retains high integrity with the real world. This approach might sound simplistic  but it is based on the well-established principles of “Occam’s Razor” and statistical parsimony which state that the simplest approach to a problem that produces a similar result is likely to be the more powerful and reliable solution.  This approach is further endorsed by George Box, one of the UK’s leading statisticians (who famously said all models are wrong but some are useful):

“Just as the ability to devise simple but evocative models is the signature of the great scientist so over-elaboration and overparameterization is often the mark of mediocrity.”


At MarketingIQ our focus is on fast, simple, reliable and correctly specified models with strong validation and diagnostics.

What are the benefits of MMM Lite:
  1. Clear insight: MMM Lite gives you clear insight on the channels that deliver incremental business growth. It will find the obvious high ROI channels quickly – but these findings may challenge what you see in MTA and rule-based, last touch attribution.
  2. Being 80% right rather than 100% wrong: Many forms of digital attribution are simply wrong. The IPA author Peter Field has referred to forms of attribution like rules-based as little more than misattribution and he’s right.
  3. More interpretable: MMM Lite models tend to be relatively simple. This aids interpretability and robustness. It makes them easier to explain and understand and this can be a key benefit in marketing applications; if your model is so complex that no-one understands it, interpretability will be low and this will mean reduced confidence in the outputs and recommendations.
  4. Incrementality: MMM Lite will give you a clear view on incrementality. You will see that Chanel’s that deliver genuine incremental growth, rather than those that monetise purchase traffic that you were going to receive anyway.
  5. Defines base: It will give you an estimate of your base – the level of natural traffic you would receive if you don’t do any short terms advertising. This is not incremental growth but it is often counted as such. Focussing on re-buying your natural traffic is one of the main causes of declining media budget effectiveness over the last decade.
  6. It is fast – provided the data is in place models can be built and tested within a week and deployed in real time.
  7. It is low cost – this approach does not require the complexities of MMM or MTA.
  8. More accurate: Despite it being fast and low cost, statistical attention can give you more accuracy than MTA and almost as much as MMM

MMM Lite finds genuine incremental growth. If your business has stopped growing it’s probably because your marketing and media budgets are targeting buyers who were already coming your way. The only way you can get incremental growth is to win genuinely new customers. And the only quick, and cost-effective way to optimise your budgets to deliver incremental growth is by using fast MMM Lite.

PS – a reminder on the limitations of rule-based attribution

Rule-based attribution must be treated with caution of you want to understand how your marketing ecosystem is working. It says that if you passed a certain ‘signpost’, like a specific web page, then that signpost must have had a causal role in the consumer’s journey. If we extend this analogy a bit further, we all pass signposts on our journeys, but can you say that each and every one made a critical contribution to you reaching your destination?