The genius Austrian philosopher Ludwig Wittgenstein wrote eloquently on the concepts of certainty and predictions. He is quoted as saying, ‘I won’t say “see you tomorrow” because that would be predicting the future, and I’m pretty sure I can’t do that.’ Marketers and data scientists alike laugh in the face of such philosophical pessimism. That’s because, with sophisticated marketing mix modeling, predicting the future is what marketing professionals try to do. And, in 2023, with access to vast amounts of data and accessible tools, Marketing Mix Modelling is making a comeback. Let’s dive in.
What is Marketing Mix Modeling – A Quick Recap
Marketing mix models help teams predict what marketing budget will give you the best return on investment (ROI). Let’s explain the concept by understanding the question an MMM strives to answer. By engaging in Marketing Mix Modeling, we ask, “What percentage of my sales are due to marketing efforts?” Even more specifically, you query, “Which marketing channels/strategies were most effective in leading to sales?”
If you get accurate, data-driven answers to these questions, taking external factors into account, you can apply your findings to future marketing strategies to optimize your ROAS. However, the number of channels available today has dramatically increased. Modern marketers need to consider many external factors when answering such questions. Above that, in the modern world, marketing budgets are tight and marketers are under increased pressure to provide evidence for their suggested budget allocation. Therefore, effective MMMs are as close to a crystal ball as marketing managers can get, using statistical analysis to look back at sales over time to determine precisely what caused those sales.
Determining a Baseline
Establishing a baseline is crucial in estimating the impact of different marketing variables on sales. Historical analysis, time-series modeling, control groups, and economic analysis are four common tactics used to determine the baseline and consequences of your various marketing channels.
Historical Analysis: Examining past events and situations to gain insight into what happened and why. For example, you could look at how much you spent on each campaign and the resulting business outcomes, such as sales or customer acquisition.
Time-series Modeling: Forecasting the impact of future marketing campaigns by analyzing the relationship between marketing spend and business outcomes over time. Thus, a company can predict how much it should spend on different marketing activities in the future to achieve its desired business goals.
Control Groups: A fairly well-known concept. Let’s say a company wants to evaluate the effect of a particular marketing activity, such as a social media ad campaign. They can assess the impact of the social ad campaign for a control group against a group that wasn’t exposed to the ad. This can aid the company in determining the campaign’s incremental impact on its business outcomes.
Economic Analysis: Identifying the most effective marketing activities for a given budget. By weighing up the costs and benefits of different marketing activities, a company can use the optimal mix of marketing activities to try and maximize its ROI.
At their core, these analyses provide a picture of what sales would be without any marketing interventions and then assess what marketing activities might achieve the best results. They form part of the equation to assess what leads to sales. After all:
Sales = Sales from Marketing + Sales from Non-Marketing
That’s why determining a baseline and assessing the results of each marketing channel is critical in building the whole picture in an MMM.
What Tools Come Into Play? And Do I Need a Ph.D. in Data Science?
The short answer is you need sophisticated tools, but you don’t need to be a mega data head to use them, and you definitely don’t need a Ph.D. Marketers need to use a variety of tools when building Marketing mix models.
- Statistical software would require programming languages such as R, or Python to build models.
- Data visualization tools like Tableau or Power BI to create visuals that communicate results.
- Marketing automation software such as Hubspot, Marketo, or Eloqua to track marketing activities and performance metrics such as website visits, leads, and sales. This could be replaced with your internal analytics system if available.
- Database management systems and econometric models to manage large volumes of data quantify the impact of different marketing activities on business outcomes.
Depending on your constitution, these tools may intimidate or excite you. If you have the expertise and resources in data science, statistical analysis, and data engineering, there are open-source software packages available for you to use.
Summary: Why are MMMs Back in Vogue?
Reasons for the renaissance:
- Businesses are gathering more information than ever before. As data from third-party cookies and ad platforms have become less accessible, many organizations are prioritizing first-party data instead. Data from offline channels must still be supplemented with digital marketing data. MMM makes sense of all this accessible historical data from all marketing channels.
- Machine learning and cloud computing technologies mean we can easily make more sense of all this data!
- More data + better technology = more weighted predictions to inform marketing decisions, which is incredibly valuable in today’s competitive environment.
That is why MMMs are returning to the fore for marketing teams, set to thrive as a relatively future-proof measurement solution for brands seeking actionable insights into cross-marketing effects. And, of course, this means a healthier bottom line which, after all, is what it’s all about.
Looking to optimize your media spend and maximize your ROI for your app or brand? Look no further! At Admiral Media, we specialize in helping our clients make data-driven marketing decisions. Get in touch and let’s talk about how we can optimize your media mix to achieve your business goals.