Marketing Mix Modelling: What is it and what is it for?

Marketing Mix Modelling
Ignasi Fernández 10m of reading

Marketing Mix Modelling (or MMM) is a fundamental tool that helps CMOs and Marketing Directors to optimise their investments and get the maximum return. Am I investing enough in my campaigns? Could I increase the impact by redistributing my budget differently? What could happen if I increased my investment in a particular medium by 15%? All these questions are answered through a marketing mix modelling model. Today we’ll tell you all about how to use these models to optimise your marketing budget.

Advertising effectiveness – how do you measure it?

Achieving maximum advertising effectiveness is one of the main objectives of marketing teams. Brands have limited budgets and need to get the maximum impact out of them. And that’s where media research and analytics help them with different solutions that cover complementary objectives:

  • Predict the success of a campaign before it is aired. This is achieved with advertising pre-tests, market studies that predict the success or failure of a campaign by analysing its creativity. In addition, pre-test advertising evaluates the details of the creative to find ways to improve the final execution.
  • Measuring the effectiveness of a campaign. To measure the impact of a single campaign, brands have all the media analytics tools at their disposal, as well as post-tests. The post-test measures how many people remember seeing the campaign, measures brand attribution and other important variables to build a brand funnel and know if the campaign has met its objectives.
  • Measure the impact of advertising on the brand. When a brand runs several campaigns over time, it makes less sense to measure the impact of each campaign individually, as the different actions overlap in the media and in the consumer’s mind. In these cases, the brand is often equipped with brand tracking, which assesses the impact of all activity on the brand on an ongoing basis. In addition, a significant part of the impact of advertising is long term, so continuous measurement helps to better understand the cumulative effect of campaigns.
  • Measure the impact of advertising on sales and business objectives. There is no doubt that sales impact is the measure of advertising effectiveness of most interest to CEOs and CFOs. The most common ways to measure the impact of advertising on sales are attribution modelling and marketing mix modelling.

Attribution models

‘Half of the money I spend on advertising is wasted; the problem is I don’t know which half’. John Wanamaker’s famous quote is still on the lips of many CMOs today. We have more data than ever before and yet it is difficult to get a clear and complete picture of the sales impact of different advertising budget items.

One of the ways to measure advertising effectiveness is through attribution models. These models assign credit to different channels or campaigns in the customer conversion process. However, despite their usefulness, they have limitations in providing the full picture that CMOs need:

  • Media consumption has fragmented. TV used to be everything, but now there are many media that share the consumer’s attention, and this makes the distribution of credit more complex and therefore less explanatory of reality.
  • They underestimate the long-term effect of advertising. Our purchase decision process is influenced by all the advertising we have seen over the years and any data source does not capture such long periods. This undervalues the role of brand equity built up over the years.
  • It is more difficult to measure the impact of offline media. Tracking where the consumer has clicked in the later stages of the funnel is easy, but it is much more difficult to measure the impact of offline advertising. This means that offline media are often undervalued in attribution models.
  • Data from each medium is often collected in silos. Combining different data sources to get a complete picture requires effort and expertise that takes into account the inherent biases of each data source.

Attribution models are useful, but it is important to recognise their limitations and use them with caution. Critics argue that attribution models have led today’s marketing to measure its impact primarily by immediate results rather than the long-term impact of actions. This ‘short-termism’ may have weakened the impact of marketing actions and the importance of the function in achieving long-term business goals. This is why more and more brands are complementing their attribution models with marketing mix modelling.

What is Marketing Mix Modelling?

Marketing Mix Modelling analysis is a statistical technique used to evaluate the impact of different marketing investments on the progress of a brand or business. MMM does not seek, as attribution models do, to record the impact of each medium on consumers and apportion credit for the sale among them. The MMM looks for correlations between an explained variable – for example sales – with a set of explanatory variables – the different marketing investments. It does this through econometric models that analyse time series.

Combining attribution models with marketing mix modelling allows for a much more complete picture of the impact of marketing investment. MMM integrates all media and other marketing variables into a single analysis, so it gives a much more complete and balanced picture. It is not biased towards online media, as it is based on marketing activity rather than impact, making it easier to get the data. Nor does it underestimate the contribution of brand value, as there is always a ‘base’ sales ‘story’ in any analysis that is not affected by marketing activity.

The benefits of combining attribution and marketing mix modelling have recently been highlighted by advertising effectiveness experts such as Les Binet.

What is the purpose of a Marketing Mix Modelling analysis?

Thanks to a marketing mix modelling model a CMO can:

  • Understand which marketing actions have the greatest impact on business objectives: The MMM explains how advertising and other marketing mix variables influence results. This helps to have more useful conversations in the organisation with marketing and non-marketing experts.
  • Understand the saturation level of each media. Advertising investment in each medium provides diminishing returns. Initially it is very profitable to invest in it, but as investment increases, the marginal contribution decreases to the point where further investment may even be counterproductive. The marketing mix modelling model allows you to know where you are on the saturation curve and therefore whether it is advisable to continue investing in it or to divert the investment to another medium.
  • Optimise the advertising budget. The MMM allows you to determine how to most effectively allocate your marketing budget across different channels and tactics, ensuring that more money is spent on the activities that have the greatest impact on your business objectives.
  • Predict the impact of changes in advertising spend. An MMM allows for ‘what if’ modelling, i.e. what would happen if we shifted investment from medium A to medium B. Would we increase overall advertising effectiveness and ROI? In fact, most MMM models provide a best-case scenario where for each level of investment the results are optimised.
  • Quantify the optimal advertising budget. As long as the ROI is greater than the investment, it is in the company’s interest to continue to allocate resources to advertising. This helps the CMO to have budget conversations based on objective data.

How to carry out a Marketing Mix Modelling analysis?

The process of conducting a Marketing Mix Modelling analysis generally involves the following steps:

  • Decide which variable you want to explain. The response variables, or dependent variables, have a direct relationship with the business objectives. Often they are sales, sales in a given channel – such as ecommerce – or other sales-related variables – registrations, subscriptions, application downloads…
  • Select initial explanatory variables. Explanatory variables are those that represent the different marketing activities and other factors that can influence the commercial performance of a brand or product. Any MMM model takes into account advertising investment and perhaps other advertising activity variables that are complementary to investment. But the MMM model goes far beyond advertising. It must capture everything that influences the variable to be explained. And that includes price changes, promotions, level of distribution or in some cases the size of the sales team. There are also other exogenous factors that can have an influence such as seasonality, competitor activities, the evolution of the economy or even the weather or events of great impact when they influence the demand for the product. The selection of initial variables should include all of them, and as the analysis iterates, those that are less correlated should be discarded.
  • Collect data. Historical data on marketing variables (price, advertising, distribution, etc.) and performance metrics (sales, registrations, etc.) are collected over a relevant period of time and as disaggregated as possible. The more individualised measurement points in the time series, the better.
  • Statistical analysis: A statistical model, such as linear regression, is used to analyse the relationship between marketing variables and performance metrics. Some variables will not be relevant in the analysis and can therefore be discarded to simplify the model.
  • Model validation: The validity of the model is tested by comparing the estimates with new measurements not used in the training of the model. It should be borne in mind that it is advisable to adjust the model over time to ensure that there are no changes in the market that degrade its explanatory capacity.

Once its validity has been checked, we are ready to start using the model to make decisions.

Criticisms of Marketing Mix Modelling models

Although Marketing Mix Modelling (MMM) is a valuable tool for understanding the impact of marketing activities on business, it has also been criticised.

  • Limitations in accuracy: for an MMM model to be operational, it must be relatively simple. Therefore, its definition eliminates variables that are not explanatory, or that cannot be managed, so its detractors argue that it is not completely accurate.
  • Sensitive to changes in the market. If new operators enter the market, or if there are technological changes, the marketing mix modelling model may degrade quickly, so it must be reviewed regularly to ensure its validity.
  • Difficulty in obtaining time series data for some data. Sometimes we know what we invest in an advertising tactic, but we don’t know when that investment has impacted the market. As media becomes fragmented, it is difficult to get a sufficiently granular view of execution.
  • Correlation does not imply causation. Critics argue that some variables may turn out to be correlated, but without the explanatory variable being the one that determines sales performance (perhaps both are affected by a third variable). This is where the expertise of the econometric modelling expert makes a difference.
  • Costs and resources: Implementing and maintaining an MMM model can be costly and require resources and technical expertise. This can be a disincentive to some companies, who may not have the resources or the means to obtain the data they need.

Despite these criticisms, Marketing Mix Modelling is still perceived by many advertising effectiveness experts as a fundamental tool, especially when complemented with attribution models, as it greatly helps practitioners and companies to make decisions based on objective data.

Improve your advertising effectiveness with We are testers

At We are testers we can help you evaluate the impact of your advertising activity in the way that best suits you:

  • We can work with you to ensure the success of your creative by running regular pre-tests. Using the same questionnaire, you can compare data between campaigns to identify the best performers (and the below average performers of your creative).
  • And if you’re already running your advertising, we can help you measure the impact of your campaigns through advertising post-tests or brand and ad tracking.

Improve your advertising effectiveness today. Contact our experts to find out all the details.

Update date 26 April, 2024

Get in touch with our experts and discover how to take your research further.