When we talk to folks who are new to Media Mix Modeling (MMM) and tell them there’s no pixel or digital tracking involved, it can be a surprise and they want to learn how the methodology works without tracking.
The idea of tracking people as well as possible has become so ingrained that it can be hard to imagine media measurement working without it.
With digital tracking becoming so institutionalized over the last decade, there is a generation of marketers that have only thought about attribution through multi-touch attribution (MTA), where browser cookies (or similar identifiers) allow advertisers to track the marketing source of their customers.
The truth is that for a while, for ecommerce and digital product companies, digital tracking worked… until it didn’t. With the implementation of GDPR, the impact of iOS14, and the trajectory towards even more privacy regulations, digital tracking has become less effective in measuring marketing’s impact – especially for larger brands.
Smart and adaptive marketers have seen this shift and have looked for alternatives that don’t rely on pixels and tracking people across the internet. That’s where MMM comes in.
What is MMM?
Let’s start with some context: marketing mix modeling is a statistical modeling technique that marketers use to determine which channels in their marketing mix deserve credit for sales, in order to reallocate budget to the highest-performing areas. The best consumer brands already use MMM (here’s a database on who’s talking about MMM in their case studies).
The concept of MMM isn’t new – and that’s why it works. Digital tracking is a product of the internet, but if you zoom out, it’s just a blip in marketing’s history. Marketers had never had the capability to follow people across the buying journey. If you worked for Coca-Cola in the 60s, you had no MTA on your TV, radio, or newspaper ads.
MMM was developed to solve this problem with only aggregated data, and now we’ve circled back to facing the same challenge.
How Does MMM Work without Tracking People?
MMMs are top-down econometric models that attempt to identify patterns in historical data. These data will generally have some amount of variation in them: there will be some days where there are different amounts of marketing activity happening and there are different amounts of new customers that a brand acquires, or revenue that they’re generating, or new marketing qualified leads being brought in.
The statistical algorithm is attempting to answer questions like this: “On days when you spend a little bit more on Facebook relative to other channels, how many additional new customers did you generate? On weeks when the podcast drops, do we tend to see an increase in MQLs?”
Of course, to do this well you need to account for a lot of complexity that impacts marketing performance in the real world. That includes things like time-shifts or adstock effects, diminishing marginal returns of increased investment, seasonality, and many other types of effects.
(Modern) Marketing Mix Modeling
Approaches to MMM have developed meaningfully since their first incarnations, where legacy vendors would deliver static reports every six months and use that plan to help determine what brands should buy in the “upfronts”.
Today, the world is a lot more complex and marketing channels are so much more dynamic. Marketers are dynamically buying media on a daily or weekly basis and ROI in a marketing channel can change dramatically week-to-week.
That once-every-six-month PowerPoint just doesn’t work anymore, but there are still big CPG brands using them to inform their budget allocation decisions.
What we’ve done at Recast is take the magic of modern statistics to fully automate this process. Modern marketers can finally get a real-time read on what’s performing well and what’s not– algorithm changes, new platforms, creative adjustments – all in one place and with the scientific rigor they need to make accurate decisions.
Unlocking the Potential of MMM without Tracking
MMM, especially when powered by Bayesian methods, offers invaluable advantages that traditional tracking cannot provide. First and foremost, it delivers a holistic view of marketing effectiveness. While digital tracking focuses on individual channels or campaigns, MMM takes a big-picture approach. It gives marketers a comprehensive understanding of how different channels interact with each other and contribute to overall marketing goals.
Moreover, MMM can handle the complexity of real-world marketing data with numerous overlapping channels, campaigns, and external factors. It can effectively untangle this web and attribute outcomes to specific activities, thereby revealing the true incremental impact of each channel.
Finally, as a measurement method that doesn’t involve digital tracking, MMM protects consumer privacy and complies with strict data regulations. It achieves this without compromising on the richness and depth of marketing insights. By leveraging machine learning and advanced statistical methods, MMM ensures that your marketing strategy remains robust, effective, and future-proof in the ever-evolving regulatory landscape.
TLDR:
In a world where consumer privacy regulations are becoming stricter, digital tracking is failing to provide the right information to marketers. Multi-touch attribution is still the most common starting place for most brands, but if your brand is looking for an ongoing way to measure incrementality, you should consider marketing mix modeling.
Without having to track people and by focusing on aggregate data, marketing mix modeling circumvents the downfalls of digital tracking and empowers marketers to make data-informed allocation decisions.
