Is your brand suitable for MMM?

Is your brand suitable for MMM?

Marketing Mix Modelling (MMM) has been gaining a lot of traction recently - but what to expect from it and is it suitable for your business at all? In this brief guide we will cover the main questions that should allow you to make an informed decision on whether to invest into MMM or not (or maybe just not now).

What does it take to Use mMM? Does it make sense for your Brand?

Despite significant progress in automation and democratisation of MMM over the last circa 5-6 years, successful MMM projects are still quite complex and require considerable investment regardless whether you procure it completely from an external party or build it yourself.

A successful MMM project/initiative has several components, some of the most important being:

Good Initial model + Quality input data

You will need reliable input data and a model to start with. While the initial model most likely won't be perfect and it will evolve over time, it still needs to be good enough - if its results are unrealistic or it gets some important channels wrong, you may easily lose the trust of other stakeholders in the organization.

01
Ongoing model development and maintenance

Often overlooked by internal teams trying to build an inhouse MMM. MMM should be viewed as an iterative process, not a one-time model creation exercise. It is very likely that your model will evolve and improve over time.

02
Internal attention of your organisation

Especially with larger organisations it will be necessary to spend time and effort on explaining the model, its results, assumptions and mechanics to other teams (commercial, finance etc) in your company before it can be used in practice to change and optimize budgets.

03
Operationalizing the model results

I.e. making changes (or call it optimization experiments) and gradually incorporating MMM into your planning, reporting and performance review processes and systems.Depending on your organisation complexity, this can be easy or it can be a major change management exercise taking multiple months.

04

All in all you should not expect MMM to be a “quick win” – getting real value out of it will realistically take a few months for most organisations and will require active involvement and resources on your part.

What to expect from MMM?

After 1 year of actively using MMM for optimization, advertisers usually see a 5-25% improvement on their overall media ROI. In any specific case the overall uplift depends e.g. on

  • How quickly your organisation can absorb and use MMM results: are you able to make changes, experiments based on MMM guidance? Or will each change require a lot of internal approvals, back and forth with multiple stakeholders? Or maybe you are currently lacking resources to actually implement the changes and your team is overwhelmed with other projects?
  • Market and brand maturity and market share – a market leader in a mature stable market simply may not have any more opportunities to grow market share 
  • Your starting point – it doesn’t happen often but of course it is possible that your current media mix is close to the optimum already
  • Consistency: As mentioned above MMM is best viewed not as a one-off but as an iterative process, unless you are very lucky, you will not get the full benefits from MMM right after its implementation.
 

It is also important to understand that MMM optimization will impact your overall media/marketing ROI, it will not directly affect your baseline sales in the short term (long term is another story). Of course without MMM you don’t know what your baseline sales is to start with but as a general heuristic you can use that

  • If you are a market leader in a limited market (e.g. regionally – maybe a major telco provider or a bank  in a specific country), it is quite possible that your baseline sales will be 80-90% of your total sales). So in this case MMM optimization can directly affect in the short term only the remaining 10-20% of your sales.
  • If you are a smaller or mid-sized company on a large market, your baseline sales will be smaller – maybe in the 15-40% region. In this case MMM would directly affect the 60-85% remaining “non-baseline” sales.

Suitable brand/business size

Technically there is no minimum size (of revenue or media spend) for MMM to work but in some cases the potential relative improvement just isn’t worth the effort and complexity.

Our experience is that at around 1.5-2m USD of annual marketing spend, MMM starts to make sense – both in terms of potential business results uplift and the ability of the company to absorb and use the results. If you are currently spending significantly less, it may be totally sufficient to use attribution models and incrementality testing.

Suitable marketing mix structure

Another factor is how structured marketing mix you have. Take these 2 examples:

  • Company A: DTC brand in wellness, advertising on FB/IG, TikTok, Youtube, Google, working with influencers across various platforms, working with promotions and pricing a lot. The brand actively generates demand for its products. It sells its products via its own website and several marketplaces.Total marketing spend 15m USD/y.
  • Company B: A dropshipping-type business in a “need”-based category: vast majority of traffic acquired via Google organic and product level Google Ads campaigns (without any deeper structure eg by SKU segments).. This business mostly captures existing demand for 3rd party products and categories it sells on its website.. Total marketing spend 15m USD/y.

In this case both companies have substantial marketing spend but Company A is a much better fit for MMM – there are many more “levers” in how they can structure their media mix and budgets across various channels + added value in MMM from understanding how advertising also impacts their sales on various marketplaces, not just their website..

Typically if you have several (at least 5-6) media channels (here channel may also be a campaign type, eg Google Ads Pmax can be considered a channel, not just Google Ads as a whole) that you can actively manage then it is a good indication that MMM may be useful for you.

If you have to actively generate awareness and demand for your product/service, then it is also a good indicator for MMM being useful for your brand.

Additionally if you have multiple sales channels – your website/app, marketplaces, own retail stores or you sell via other retail partners – then MMM can help you understand the impact of advertising across all these sales channels.

MMM is also very good at mitigating tracking problems – either missing consents or issues with traffic from channels like Facebook, Tiktok. So if it is difficult for you to measure marketing also for these reasons, MMM may be a good fit.

Suitable business type

Examples of industries typically well-suited for MMM:

  • DTC and ecommerce
  • Omni-channel retail
  • Mobile app publishers
  • Game publishers
  • CPG
  • Fashion
  • Travel
  • Some SaaS companies
  • Financial services
  • Consumer technology & electronics
 

This list is not exhaustive – many other industries can use MMM, generally if marketing/advertising & demand generation plays a big role in your business, it is a strong indicator that MMM may be useful for you.On the other hand here are some examples where MMM may be quite challenging or not a good fit:

  • Completely new brands (no historical data to train the model)
  • Business or industry undergoing major disruption or changes – eg changing the business model etc (strong hypothesis that historical data will be useless in predicting and uncovering patterns that will hold in future)
  • Very long sales cycles – eg in industrial engineering or similar verticals where the sales cycle can take many months or even years
  • Very complex distribution channels – where it may be difficult to even track actual sales on weekly (or at least monthly basis)

Other key considerations and success factors

Based on our experience there additional factors that can greatly increase the chances of success with MMM:

  • Some experience with marketing testing / experiments
    • First it helps and speeds up the model development process. Second, using the model recommendations is basically doing such tests / experiments (change investment into channel X and then check it brings the expected result). Marketing experiments also make you aware of all the various limitations of your attribution model and you get practical experience with marketing incrementality – all these usually make the MMM implementation and application process much smoother.

  • Some analytics maturity – this doesn’t mean you need to have an analytics team, or even one marketing analyst or that you need to have some advanced knowledge of statistics. It’s rather abou being able to work with uncertainty – measuring marketing effectiveness (eg campaign/channel incremental effect and ROI) will always have some uncertainty to it – the underlying reality (consumers and their shopping behaviour, competitors with their own activities, macroeconomic effects) is a very complex system that changes in time. MMM does not try to hide this (unlike many attribution models that offer the comforting illusion of precise super-granular results while often being completely wrong)


Being open-minded and ready to challenge previous beliefs – it is quite possible that MMM results will show quite different results than what you were used to from Google Analytics or some attribution model solution.

Share Knowledge
Solutions
Contact

Get In Touch

© 2024 MarketingIntelligence.io

Terms and Conditions