Google has introduced “Meridian,” an open-source tool that utilizes marketing mix modeling (MMM) to assist marketers in finding better user measurement alternatives to the soon-to-be-discontinued Chrome tracking cookie.
The tool is currently accessible to a select group of users, but Google plans to make it available to all marketers and data scientists in the near future.
MMM takes a mathematical approach to measuring true impact, using calculations to measure sales impact based on reach, response, and more to connect marketing metrics to broader business goals like sales or market value.
A Google study conducted with Kantar revealed that 60% of US advertisers are currently using MMMs, and 58% of those not using these models are considering doing so in the future. This move also puts Google in direct competition with Meta’s own MMM, Robyn, launched in 2022.
Discovering your North Star metrics
“MMMs today are not perfect but are evolving,” said Harikesh Nair, Senior Director of Data Science at Google. “With Meridian, we aim to help your team navigate towards your future North Star, both through innovation and by sharing our data along with an open-source model. We invite everyone to join us in this endeavor.”
Unlike many existing MMM solutions, Meridian will include YouTube reach and frequency metrics alongside indexed Google query-volume data.
Users will have the ability to conduct scenario and budget planning alongside future cross-channel forecasting for budget and contribution. Google will also offer educational resources and technical FAQs for troubleshooting.
Nair also emphasized that Meridian is “privacy-durable,” adhering to relevant privacy laws and regulations across regions.
‘Modelling clay’ for further development
As an open-source tool, Google assures users that the Meridian methodology will be transparent: “Meridian is designed to give the user control as any modeller is fully empowered to change the code and model parameters to meet their unique business needs,” Google stated in a launch blog. “Think of Meridian as modelling clay: a starting point that allows users to iterate and develop further.”
Typically, MMMs are seen as costly, time-consuming, and requiring specific data analysis skills compared to other analytics methods like GA4. Google hopes that Meridian will provide a more accessible and adaptable way to measure campaign performance than existing models, while meeting the need for more privacy-conscious tracking methods.
‘The inclusion of both reach and frequency introduces an exciting concept’
Niall Oulton, Director of Marketing Analytics at PyMC Labs, a consulting firm specializing in data analysis and predictive modeling, believes that the launch of Meridian could bring new life to an industry that has been dominated by expensive consultancies offering opaque ‘black-box’ solutions.
“Marketers should consider moving away from largely inaccurate attribution-based measurement approaches and embrace MMM,” he added.
Oulton sees the inclusion of reach and frequency data points as a significant enhancement to Meridian’s offering that sets it apart from other solutions. “Marketing Mix Models have been somewhat limited, focusing on reach through the estimation of diminishing returns. [Meridian’s] inclusion of both reach and frequency introduces an exciting concept, allowing planners to use their marketing strategies to determine not only how much to invest but also how often to expose their target audience to multiple exposures.”
One area where Oulton believes Meridian falls short is in not accounting for time-varying impacts of marketing and the extent of its openness as an open-source solution. He stated, “The effectiveness of marketing can vary based on the time of year, seasonal influences, or changing dynamics of marketing impacts.
“In addition, I am curious about its level of openness and how much Meridian will evolve into truly open-source software. Currently, it appears to be accessible only to a select, approved few (for now).”
Google’s Meridian MMM is currently in limited availability. You can apply for access here.
Loading PerspectiveSplit analysis...
