Benefits of marketing mix modelling: Why is MMM so popular right now?

Introduction

The concept of marketing mix modelling (often referred to as just ‘MMM’) has been around for a while – as early as the 1960s in fact – which should be no surprise, as the business challenge of what marketing channels to use and where best to spend your money has always been the essence of good marketing, at least if somebody is holding you accountable for that spend and performance!

Marketing mix modelling has its foundations in statistical techniques and econometric modelling, which still holds largely true today. However, the mix of channels and advancements in end-to-end analytics create new challenges to be tackled, not least the expectations of what MMM is and what it can deliver.
In reality, there are various analytics techniques that can be undertaken to answer the overall business question: ‘how do my channels actually impact sales?’. In this blog we will answer some common questions about MMM, address some common (comparable) techniques, and share how and when you might look to choose one method over another.

What is marketing mix modelling?

MMM is a statistical technique, with its roots in regression, that aims to analyse the impact of various marketing tactics on sales over time (other KPIs are also available!). Marketing mix modelling will consider all aspects of marketing to do this, such as foundational frameworks like ‘The 4 Ps of Marketing’ (Product, Price, Place and Promotion).

MMM is similar to econometric modelling in terms of techniques used, however there are some key differences. On the whole, econometrics is broader in its considerations and applications, often encompassing aspects of general economic factors in relation to politics, international trade, public policy and more. MMM, on the other hand, focusses more specifically on marketing activities and their impact on business outcomes.

You might also come across the term ‘media mix modelling’ (with the same unhelpful acronym, ‘MMM’). Much like econometrics, media mix modelling tends to differ from marketing mix modelling due to its scope and general objective. Media mix modelling tends to have an even narrower focus than marketing mix modelling; As the name implies, it’s aimed more specifically on optimising a mix of media channels, focussing on optimising advertising spend.

Whether its marketing mix modelling or media mix modelling you are looking at, the key is to consider the business question you are looking to answer and ensure your model is trained using the best input variables to answer that question – Nothing new in the world of a good analytics project!

Why is MMM seeming to gain traction recently?

In recent years, the general trend has been to measure everything, integrate everything, and to link all of your data together, leaving no doubt about who did what, when, and to what end. However increasing concerns (or at least considerations) around data privacy and ethics has caused marketers to take a second look at how they collect and utilise their data.
There is a growing need to adapt to new privacy regulations, but also a greater desire to respect an individual’s privacy and find better ways to understand what marketing activities drive positive or negative outcomes.

With limitations on the ability to track 3rd party cookies, approaches such as marketing attribution may become more difficult to implement, although the effectiveness of these data sources is in itself doubtful. And with consent management becoming increasingly granular, even 1st party measurement can leave gaps in your data collections.
However, the power that marketing attribution gave marketers is well recognised now and the desire to continue to be data-led is only increasing. Machine learning has become a commonplace tool in beginning to fill the gaps that are creeping back in to the tracking of user behaviour. Organisations are also increasingly eager to build on the power of what they have learnt with these joined-up customer journeys, and there is that need again to look across the whole of marketing, not just these digital touchpoints, and replicate that approach in a more holistic way.

So in summary, while marketing mix modelling has never gone away, it is now seeing a revival as an essential tool in a marketer’s toolbelt.

The benefits of MMM: Why should organisations consider using marketing mix modelling services?

MMM is a great tool for any organisation looking to be more data led in their approach to planning and analysis of marketing activities. Key benefits of MMM include:

Ability to measure and optimise the effectiveness of marketing and advertising campaigns:
The purpose of MMM is to measure the impact of your marketing activities on your business outcomes. A well-built marketing mix model will enable you to quantify ROI by channel and make better data-led decisions on the mix of marketing activities that will lead to more optimised campaigns.

Natural adeptness at cross-channel insights:
With increasing limitations on tracking users across multiple channels the methodology for MMM neatly side steps these restrictions by using data at an aggregated level. By its very nature it doesn’t require linking user identities across different devices or tracking individuals using offline channels.

Enables more strategic planning and budgeting:
MMM provides data-driven insight to inform budget planning processes. Its outputs are transparent, allowing organisations to understand the impact each of their channels have on business outcomes and how those channels influence each other within the mix. By incorporating MMM with other tools for scenario planning, spend optimisation and forecasting, organisations can better understand what happened in the past to plan more effectively for the future.

Can be used when granular level data is not available:
As mentioned earlier, MMM works with data at an aggregated level. This offers more flexibility when looking to integrate data inputs into your decision making such as:

• Linking offline activity with online sales
• Linking online activity with offline sales
• Understanding impact of external influences such as macroeconomic factors, seasonality, competitor activities etc

Has a longer-term focus:
MMM is a powerful technique for longer term planning and assessing the impact of campaigns that don’t necessarily provide immediate impact (e.g. brand awareness campaigns, TV, and display advertising etc). By incorporating MMM into a measurement strategy, businesses can ensure longer-term activity is appropriately considered.

Marketing mix modelling vs. marketing attribution modelling: How do they differ? What are the pros and cons?

Earlier in this blog we looked at how marketing mixed modelling compares to econometrics and media mixed modelling. Another very important modelling approach to consider when looking at marketing effectiveness is marketing attribution.
Marketing attribution differs from marketing mix modelling in a number of important ways – most importantly by relying on a more granular approach. It looks to assign weightings to each individual touchpoint on the customer journey, incorporating each user’s journey and determining whether that journey leads to a successful conversion or not.
This very detailed understanding of how each customer interacts with your channels can be very powerful, but also very complex and time consuming to both collect and analyse; In addition with the increasing limitations on tracking individuals without their consent, you may end up having to rely on only a partial picture of the user journey.
While of course it is possible to model on a subset of data, you would need to be careful that the user journey you are looking to understand is not unfairly weighted to those channels (or individuals) that are easier to track.
Marketing attribution also uses a wider range of modelling algorithms, from the simple (linear, time-decay) to the more complex (Markov Chains, Game Theory, ML models). This range of models to select from can be both a benefit and a hindrance, with difficulties arising when you’re not sure what marketing attribution model will suit your business needs best.

Marketing mix modelling does have its own drawbacks to consider too. The biggest consideration when determining if MMM is suitable for you is to understand how much historical data you have.
While a marketing attribution model can work on just a few months of data, so long as it has decent volume and is fairly representative of your typical user journeys, MMM relies on trends over longer periods of time – typically a minimum of 2 years’ worth of data is advised before undertaking an MMM project. MMM also works best when looking at the broader impacts marketing has on your goals. Therefore, if you need to analyse specific campaign performance or delve deeper into specific channels, then marketing attribution will be the better bet.

Can MMM and marketing attribution complement each other?

In a previous blog, we discussed the merits of using both marketing attribution and MMM side by side to provide a more powerful and comprehensive understanding of marketing effectiveness.
While a marketing attribution model will focus on individual touchpoints and their contributions, MMM will take a holistic view, considering the overall impact of marketing inputs. By combining these two approaches, marketers can gain a more complete picture of how different marketing elements work together to drive business outcomes and the demystify the balance they needed across marketing activity for maximum business performance.

Summary

Marketing mix modelling is very a powerful and well-established statistical technique. Most marketers should be at least exploring the benefits and insight it provides into the relationship between marketing activity and business performance to optimise planning and decision making.

Some barriers to entry in starting an MMM project can be navigating what may appear to be a complex set of approaches and techniques. While variations of MMM do exist – econometrics, marketing mix modelling, and media mix modelling – the key difference lies in the scope and objective of the business question you aim to answer. Successfully choosing and developing a model depends on fully understanding your business needs and the data available to you. Investing time upfront to determine what you are looking to achieve is essential in getting the right outcomes.

MMM is best used for strategic planning and determining longer term impacts of your marketing activities. Therefore, if you require more in-depth campaign and channel analysis, then marketing attribution may be more suitable for your business needs. However, it’s important to note that MMM and marketing attribution can work side by side to develop a more complete picture of your marketing activities. While MMM allows greater flexibility when working with a mix of channels that are both tracked and not tracked, the ability of marketing attribution to provide a more granular analysis of your marketing journeys, channels, and campaigns allows for day-to-day optimisation of your marketing activities alongside the longer-term strategy set out by your MMM insights.

If you are ready to explore MMM, marketing attribution, or anything in between, we’d be delighted to discuss your needs in more detail.