MMMs + Creative Data
How creative data can integrate with MMM to solve attribution challenges
Written by
Michael Murray
September 1, 2022
4
min read
Be the first to receive the latest insights in advertising effectiveness to help you achieve creative excellence and subscribe to Solve for X.
Last month, Google delayed the deprecation of cookies until 2024 to better balance people’s privacy with business needs.
Despite competing economic signals, experts say a recession will likely hit in 2023. Given the $10bn+ ripple effects of Apple’s privacy push (Warc predicts this will rise to $40bn during 2023), this delay gives marketers more time to test and build attribution models to better measure the ROI of their marketing investment. At the same time, WSJ notes, platforms are giving more control to users to determine the types of ads they see (this also has a business benefit: it reduces wasteful ad spend towards people that will never buy a category product).
As Apple takes a bigger piece of the (ad) pie, streaming services proliferate (Walmart is adding streaming on top of their ads business), advertising unbundles (Amazon is trialing TikTok-style ads), and privacy becomes a Congressional priority – it’s clear the ad industry is experiencing a shake-up that puts attribution in a weird place.
So let’s explore it.
Attribution matters. It allows us to measure the impact of our marketing efforts and begin to overwrite the damage done by that apocryphal Wanamaker quote, “half the money I spend on advertising is wasted; the trouble is I don't know which half.”
But attribution is also a double-edged sword because measurement bias can lead marketers astray. It begins with the need to demonstrate immediate or short-term marketing ROI. This leads to an overinvestment in performance marketing & attribution metrics that measure these activities, which results in short-termism that eats long-term profitability. Adidas is a great example of this story.
Measurement bias then kicks in as marketers deem targeting and media optimizations to be the most effective levers at driving ROI (they’re not), as they are easier to optimize and attribute. So marketers invest more in them, leading to a misconception or fallacy that they are more valuable.
Rory Sutherland argues, “the very fact that it’s measurable has made people obsess about what you can measure, on the assumption that what you can measure must be important.” He goes on to say the industry has spent “far too much” time on “targeting optimization” at the expense of “creative optimization.”
Currently, the industry has two attribution models: top-down and bottom-up. “Bottom-up” relies on cookie-based media exposure data and individual-level sales for modeling, while “Top-down” relies on marketing impressions data and aggregate sales for modeling. The first is quick to show measurable marketing impact, while the second is slow.
As the cookies crumble, top-down models, like Market Mix Modeling (MMM), will become more valuable. Despite weaknesses like a lack of granularity or temporal issues like being static (not real-time) and taking too long to complete and inform campaigns, MMM’s provide marketers with a “big picture” view of their marketing investments. This lets marketers build marketing effectiveness measurement programs to better allocate budgets across marketing levers and drive business growth through more effective and efficient marketing investments.
And yet, MMM – the oldest and most established measurement model – is evolving.
One evolution introduces bottom-up, multi-touch attribution data to the model, which provides more granular data to enable in-flight campaign budget allocations. This is an important step toward turning MMM into an always-on measurement system and alleviating some of the short-term measurability challenges marketers currently face.
Another evolution is the introduction of creative data to the mix, information generated from the individual assets that can represent your creative impact. Marketers can incorporate creative data into existing marketing measurement frameworks to better understand how creative decisions impact their brand and sales outcomes. Given creative is the biggest lever of effectiveness (second only to brand size), adding creative data to attribution models is a game changer.
For example, last month, we shared Nestlé's story on how they leveraged creative data in their MMM study to map their creative decisions to sales outcomes – re-visit here.
MMMs are not the only solution. It would be unhelpful to suggest as much. Data requirements (e.g., 2-3 years of longitudinal data) demand a level of readiness. A future of always-on MMMs, combined with creative measurement and creative data, looks promising, particularly with media fragmentation, content proliferation, and privacy all impacting attribution. Further delays to the deprecation of the cookie suggest Google is also intent on building the future of marketing measurement.
When discussing attribution, however, it would be remiss not to mention consumers: the most important person (unfortunately) not in marketing meetings. Ehrenberg-Bass consistently points out the biggest danger to advertising is the failure to brand. Said differently, most ad exposures aren’t noticed or attributed to the correct brand. The rise of attention measurement companies like Amplified Intelligence can help marketers better attribute consumer attention to their ads. Creative data can help marketers codify those lessons and scale them.
Given the need to demonstrate ROI in increasingly shorter time periods, it’s no surprise marketers doubled down on digital attribution, as well as targeting and media optimizations. This isn’t to say digital attribution is bad; it’s to say Rory Sutherland makes a good point when he says, “if you optimize targeting, that’s helping you find your customers. But good creative can actually create them.”
Thanks for reading!
If you enjoyed reading this, consider signing up to receive Solve for X before it's published.
.png)
.png)
.png)