How to tackle measurement as Google rolls out open-source marketing mix modeling

How to tackle measurement as Google rolls out open-source marketing mix modeling

With the death of third-party cookies underway, the pressure is on marketers to adopt new solutions for targeting, tracking and measurement before long-standing tactics become inoperable by the end of the year. While many of Google’s efforts to lay the groundwork for a cookieless future have centered around its Privacy Sandbox proposals, the company this month announced plans for another tool that could help marketers looking for new ways to measure the effectiveness of their advertising.


Google on March 7 announced the launch of Meridian, an open-source marketing mix model (MMM) that aims to provide marketers with the foundation for comprehensive, privacy-durable measurement. The introduction of Meridian represents a major step in boosting marketers' measurement capabilities, but comes with its own caveats and challenges, according to industry experts.

"The initial impression here is that Google is once again rolling out a tool primarily tailored for its own products. However, what's truly commendable about this release is that Google is recognizing a pivotal shift: the demise of traditional tracking and attribution methods," said Matt Hertig, CEO of analytics company ChannelMix, in emailed comments. "This recognition is another signal to marketers that the walled gardens are getting higher, marking a significant leap forward in adapting to the evolving digital landscape."


MMM, MMM, good (and bad)


While MMMs have been around for the better part of a century, Google has observed a "renaissance" in the technique, with 60% of U.S. advertisers currently using MMMs and 58% of those not using these models considering doing so in the future, per Kantar data cited by Google in a blog post. The tech giant defines MMMs as statistical analyses for measuring the impact of cross-channel marketing. 


Meridian, which is currently offered in limited availability, with plans for general availability to all marketers and data scientists coming soon, is built around four pillars: innovation, transparency, actionability and education.


Meridian will work to make MMMs more accurate and analytically rigorous by incorporating methodology innovations around incrementality, reach and frequency; ensure transparency by being open source; provide rich data inputs across Google and YouTube to enable cross-channel budget optimization; and provide documentation and opportunities for further education.

"MMMs today are not perfect, but are evolving. With Meridian, we look to help your team navigate toward your future North Star, both through innovation, and by sharing our data in conjunction with an open source model," said Harikesh Nair, Google's senior director for data science, in the blog post.


While it is difficult to fully judge Meridian before it is widely available, its four governing pillars and the ability for marketers to modify open-source algorithms to suit their unique business objectives could make it attractive to a variety of marketers, according to Tina Moffett, principal analyst at Forrester.


"Meridian may be enticing for the mid-market brands that have data scientists or people who are savvy with analytics that can use Meridian's algorithmic framework to build a marketing mix model, but may not have a big budget to go with other independent marketing analytics vendors or service providers," Moffett said. "I think that's where the sweet spot will be for Meridian."






“There’s a little confusion in the market of what an MMM does versus what an advertiser actually needs.”




Tina Moffett


Principal analyst, Forrester






However, as with Google’s Privacy Sandbox proposals, Meridian should not be seen as a one-to-one replacement for the type of measurement allowed by third-party cookies. Marketing mix modeling inherently cannot provide the same level of granularity in tactical, day-to-day reporting about ads being served to specific audiences at specific times, for example.


"There needs to be a little bit more clarity around what an MMM can provide and what it cannot," Moffett said. "I think there's a little confusion in the market of what an MMM does versus what an advertiser actually needs."


Despite advertisers' long-term reliance on third-party cookies, the technology has been an unreliable mechanism for tracking online behaviors and targeting media buying amid an increasingly fragmented marketplace. By removing third-party cookies from the board, Google is encouraging marketers to use tactics like MMMs that, while less granular, might be more effective overall.


"You don't necessarily need that type of data to make investment decisions, and I think Google realized this and created this framework for advertisers that are advertising within YouTube or buying media within Google to make smarter investment decisions within the ecosystem," Moffett explained, noting that MMMs can help with larger decisions about budget allocation.


Steps marketers must take


There are several steps that marketers looking to utilize Meridian to build their own MMMs must take, according to Moffett. The main challenges that enterprise clients face when using marketing analytics models, including MMMs, are transparency and selling it to business leaders — despite businesses putting a premium on analytics and technical knowledge — according to a recent Forrester Wave report. And while Meridian is open-source, that does not guarantee full transparency.


"It's transparent for the data scientists and the machine learning engineers because they can see the code," Moffett explained. Forrester's research shows that VPs and CMOs are often buying marketing analytics model tools and services providers but don't necessarily understand how underlying models or algorithms work. 

To ensure true transparency around models like MMMs, organizations must come up with plans to engage not only stakeholders who use the models, like data scientists, but main stakeholders and marketing executives who make purchasing decisions. Transparency levels will differ across those groups, Moffett notes.


That transparency goes hand-in-hand with education around different service providers and not only their competencies around measurement but also their ability to develop tactical strategies that are tailored to brand needs.


"It's important for advertisers to really think about building a holistic strategy," Moffett said. "What that looks like, tactically, is not only looking at marketing mix as an option, but understanding what different methods can help answer certain questions and what their goals are."


For example, MMMs can helps allocate budgets to appropriate programs and channels to drive short-term revenue goals. But marketing executives will likely also need other models and tools to measure incrementality, or identity-based approaches that underpin sophisticated attribution models — two areas where Forrester sees marketers investing in measurement. While it will only help with certain measurement cases, Meridian is still a major step forward for marketers preparing for the death of the cookie.


“Meridian symbolizes one of the most significant changes in marketing over the past 15 years," said ChannelMix CEO Hertig. "Tools like this offer a new opportunity for marketers to rejuvenate the analytics space by leveraging advanced technologies like AI for effective navigation in this new era… This future of marketing measurement is now.”