Google unveiled Meridian, an open-source marketing mix modeling (MMM) tool aimed at tackling crucial measurement challenges. MMM tools gauge the impact of marketing and media investments on vital performance indicators, such as sales or revenue, while also forecasting the revenue potential of marketing endeavors. According to Forrester’s Marketing Survey, 2023, about 30% of B2C marketers use MMM tools to better understand how marketing drives value for the business.

Unpacking Google Meridian’s Core Pillars

Google’s foray into providing an open-source MMM tool reflects an understanding that restricted access to crucial data points within closed ecosystems impedes advertisers’ ability to effectively measure digital ads. Meridian aims to “[empower] teams to build best-in-class [MMM models] and drive better business outcomes.” The tech giant emphasizes Meridian’s “privacy-durable” approach grounded in innovation, transparency, actionability, and education. But B2C marketers must review Meridian’s claims with a critical eye:

  • “Privacy-durable” is redundant. MMM inherently maintains privacy durability, since it operates without individual customer behavior data. Inputs focus on aggregated media cost and revenue data, steering clear of clicks, views, or individual sales conversions.
  • “Innovation” claims are overstated. Most marketing analytics vendors already use the same methods as Meridian. Meridian’s MMM approach leans on solid but common techniques, such as a geo-level Bayesian hierarchical model incorporating seasonality, frequency, and reach data. The inclusion of nonmarketing indicators also remains ambiguous.
  • “Open source” does not equal complete transparency. While open source allows users to view and modify the MMM algorithm, complete transparency remains elusive, particularly for nontechnical stakeholders such as, for example, a VP of marketing. For many buyers, understanding ML approaches presents a significant challenge, necessitating considerable support and hand-holding.
  • “Actionability” may be limited to marketing use cases. Meridian’s scenario planning aligns with MMM providers’ offerings, aiding marketers in forecasting future marketing impacts. It’s unclear, however, whether Meridian offers scenario planning encompassing nonfinancial KPIs, constraint functionalities, frequent optimizations, or multiple optimization objectives.
  • “Education” skews technical, not practical. While Meridian’s technical specifications may satisfy ML engineers and data scientists, marketing executives require more comprehensive education and support. In The Forrester Wave™: Marketing Measurement And Optimization, Q3 2023, client references prioritize insights-supporting services. Marketing executives seek guidance on ML-driven models, while data scientists using Meridian need best practices for stakeholder engagement and model adoption.

Unlock Marketing Success Beyond MMM And Build A Comprehensive Measurement Strategy

Google’s announcement underscores the significance of ML-driven marketing analytics for measuring incremental marketing effectiveness and guiding budget allocations. Here are some key considerations before you build an MMM model:

  1. Adopt a layered measurement approach. MMM alone will not provide ad tactic or spot level performance. It offers CMOs a guide on overall marketing mix performance and budget allocation. Incorporate marketing mix modeling as part of an overarching layered measurement strategy, utilizing various techniques to gauge marketing efficacy.
  2. Choose vendors with diverse analytics capabilities. Evaluate vendors based on their industry experience, data normalization processes, measurement methodologies, and supporting services. Our recent evaluation, The Forrester Wave™: Marketing Measurement And Optimization, Q3 2023, evaluates the most significant players in the market.
  3. Manage expectations with MMM analysis. Expect to see MMM analyses such as channel halo effects, ad decay, optimal frequency of channels, and large-scale programs. But MMM may not offer granular insights into ad tactic performance — the models don’t process clicks and conversion level data as part of their analysis.
  4. Implement incrementality testing. If you need help with an ad spot, or a particular display ad performance, testing approaches can help measure marketing uplift. Your data science teams, data-driven agencies, or independent vendors can help set up a rigorous marketing or media test to analyze the incremental impact of different marketing tactics.
  5. Educate your organization on marketing analytics. Forrester believes that AI suffers from a trust problem, and explainable AI — the techniques and software capabilities for ensuring that people understand how AI systems arrive at their outputs — is a critical transparency mechanism. Prioritize transparency in MMM, and consider tools with explainable AI capabilities to enhance understanding and trust in AI-driven insights.

Need assistance with marketing mix modeling or broader marketing analytics initiatives? Request a guidance session today!