Given the analytical prowess needed to do it, outsourcing MMM insights is no longer a cost-effective exercise for many businesses. Blame longer time to insights or low confidence in results due to limited visibility into the data and underlying analysis techniques, or the obvious costs in a climate where every pound counts.

While MMM insights can be gained through a variety of sources (including marketing analytics software and external agencies) building a capability in-house now offers organisations a range of benefits:

  • Control over the data and the resulting insights: insights coming from outsourced MMM are often ‘black box’, reducing their actionability
  • Quicker insights: new, advanced analytical techniques (like the Bayesian approach – more on that later) have improved the accuracy and frequency of insight generation, allowing Marketers to make quicker decisions to stay one step ahead
  • Cost effective: good data scientists are more readily available than they were 20 years ago, and building a strong in-house data analytics capability will allow you to maximise your data’s value across all business functions

The resulting benefit is a significant impact on the most important KPI: insight to action. You may be investing heavily in a campaign that is delivering little incremental value above a certain spend and need to act quickly after understanding the optimum investment point at which to drive maximum ROI. There may be a campaign that will drive significantly high revenue and return via a different channel, or time of year that you need to change quickly to drive maximum commercial benefit.

A higher degree of ownership, flexibility and customisation enables CMOs to adapt rapidly, make quicker decisions with more confidence to stay relevant and commercially effective in an increasingly competitive world.

The advancements in open-source AI and availability of high-end computing engines at a much cheaper rate means there is no better opportunity to take advantage, run experiments quicker, fail faster and adapt in an agile fashion to improve speed to action. We are living in an age when efficient engineering pipelines, AI models, and computing power are more accessible than ever, and first party data is getting richer & richer. If you aren’t utilising the wide variety of data sitting in the lake that you invested millions to build, you are missing out!

Advancements in MMM approaches

56% of UK firms remain dissatisfied with how they are currently measuring the ROI of their marketing strategies (Gartner, 2022). This is where MMM comes into play.

Memory refresh – MMM is an analytical approach used to quantify the impact of marketing activities on the top line. It involves analysing historical data on marketing spend as well as metrics that drive base sales to isolate the incremental revenue of marketing channels. This enables organisations to identify the optimum marketing mix that will deliver the highest ROI.

Machine learning algorithms are commonly used to analyse the correlation between marketing inputs and business outputs such as revenue or profit, while also considering factors that affect base sales, such as seasonality and competition. Linear regression is the most used technique, with two main approaches: frequentist regression and Bayesian regression. Recent years have seen remarkable advances in linear regression, particularly in the Bayesian field, which is often viewed as more sophisticated.

The below summary of Capgemini’s approach to helping a luxury fashion house based in UK shape its MMM capability demonstrates that combining Frequentist and Bayesian regression approaches can lead to powerful results.

We began by using frequentist regression to model revenue variations, and then moved on to a more advanced Bayesian modelling approach. By incorporating prior knowledge such as industry experience, previous marketing mix insights, and randomized experiments, Bayesian modelling offers a higher level of confidence in MMO results. This helps the machine learning model to return more precise attribution metrics.

As a result of re-allocating budgets from channels that had reached ROI saturation to other channels that had not, we were able to see a 5% improvement in ROI for FY2022 while maintaining the same overall budget as FY2021. This success was made possible by utilizing the insights gained from the advanced Bayesian modelling approach.

The Road Ahead

So, for those who want to build a best-in-class in-house MMM capability, we’ve identified 3 winning principles:

  1. Be friends with new age tech & AI – using advanced ML techniques combined with high end computing power means more accurate, faster and smarter insights from data
  2. Test, learn, re-learn – Leverage the insights to run frequent experimentation and identify the most effective actions that will optimise overall marketing ROI
  3. Make insights accessible to all – Siloed usage of insights without business collaboration is a risky road to take. Make insights accessible to not just marketing, but brand, content and CX to make more connected decisions

The in-house revolution is driving further control and innovation. Do you want to be left behind while your competitors harness the power of data, smart/fast analytics & insights? The call is yours; we hope it’s not a hard one!