Media mix modelling: the future-proof, silo-busting future of measurement
Dane Buchanan, Global Director of Analytics, Data, and Tech was recently featured in Performance Marketing World. Read the original article here.
The death of the cookie is no somber occasion, but the switch to a new (and also old) form of measurement will require re-education – and a likely restatement of past campaign metrics
Sometime in the second half of 2024, the advertising world will change (unless this deadline gets moved once more). That is, of course, when Google expects it will finally phase out cookies from Chrome, the most prolonged goodbye in history.
This date is on every marketer’s mind because the cookieless world changes everything. So let’s call it what it is: the death of direct tracking on mobile and the web. But this does not need to be a sombre occasion. In fact, it will allow marketers to reassess and improve how they’ve been assessing digital campaign performance by calling upon a measurement blast from the past that will provide a lot of value in the cookieless future.
The biggest change
The death of cookies will require everyone to rethink their strategy and understand that their historic metrics must also change. This change in landscape is best thought of as a move from attribution to incrementality measurement.
An attribution model – powered by Internet-wide cookie tracking and mobile identifiers – considers the universe of people who have seen your ad and whether they took an action afterward. The most common approaches are last click, where the last ad a person clicked on before a conversion gets all the credit, first click, where the first ad clicked in the user journey receives all the credit, or somewhere in between, e.g., linear, time decay, etc. Before the death of tracking, the holy grail for marketers was data-driven attribution.
However, while this certainly trumped all other attribution methodologies, it still tended to neglect your offline media (TV, OOH, Print, etc.) and, in many cases, your owned media (email, organic search, etc.). It also missed one vital piece of the puzzle, which is that attribution doesn’t measure incrementality. Key to the growth of any business is understanding the incremental impact of your marketing efforts and not just redistributing credit for conversions across your paid media channels.
The attribution model only works with cookies/ identifiers, which we’ve discussed are definitely going away. So how will we measure in a cookieless world?
We will move to a measurement framework that measures incrementality, relying on a tried and tested methodology, media mix modeling (MMM). The switch to a new form of measurement will require re-education and likely restatement of past campaign metrics. By switching from attribution to incrementality, your ROI will initially seem less to the untrained eye. But while the cost per engagement will seemingly rise, it’s not that ROI has decreased or that your media has become less efficient; just that you’re now looking at the results through a different lens.
The New MMM Blueprint
What is old is new: advertising’s new savior is MMM, a measurement approach that has always held value but fell to the wayside as other measurement approaches gained steam. The good news is that MMM is evolving and can be a significant component of performance marketing.
Traditional MMM has struggled to keep up with the ever-changing digital landscape and took forever to show results (typically once or twice per year). Now, however, we can model at high speed and a higher granularity than ever before, providing outputs that can impact media plans at speeds performance marketers have come to expect. This increase in speed and granularity is driven, in part, by advances in machine learning and computing power.
So if MMM-powered incrementality measurement is the future, how do companies get started? First, we recommend you try it by measuring a couple of campaigns with MMM. It won’t change your strategy overnight but will give you a different perspective on your performance. And ultimately, you need to get comfortable with cookieless measurement sooner rather than later.
The improved MMM approach is based on live, ever-changing statistical modeling. As with all models, there will be margins of error, which means you may experience some concern and pushback from executives. That’s a positive thing that demonstrates your executive leadership wants to be involved in campaign success. We also believe it’s positive and good to question results. If you have additional budget, you can run a geo-lift test to prove the model is telling you the right thing. And if it is wrong, you can feed it back into the model to make it more accurate.
The best part of the MMM is that, unlike attribution, it does not use user-level data; thus, it is future-proofed. Of course, we’re not saying MMM will be the only way to measure in the future, but it will always work, no matter what decisions Google, Apple, and others take.
An additional benefit of embracing MMM is that it gets everyone across the organization talking about the same data and outcomes and working together to achieve the same goals, thereby breaking down silos.
Ultimately, preparations for the post-cookie world must begin today (if we’re honest, they should have happened a long time ago). MMM is new and improved, and future-proofed. It will work with data clean rooms if those become the standard for capturing and assessing data. Media response curves, a key MMM output, will give a fuller picture of how successful (or unsuccessful) your advertising is; these are the core of your budget setting and optimization engine. Ultimately giving your C-suite a much clearer picture of how much money your team will need to invest to achieve your business goals. So while the competition may mourn the death of cookies, you can instead reach for the future and start leaning into a measurement that will provide a better picture and remain valid no matter how many twists and turns the advertising industry produces in the future