Marketing Measurement
Founder of Middlegame Marketing Sciences, turning marketing data into action
Todd argues that siloed retail media ROAS misleads in two ways: it claims credit for demand built elsewhere and ignores halo effects on other channels. He recommends modeling channels together and validating incrementality with geo-lift experiments and holdouts.
Todd argues that the buzz around Meta's Seller app misses the point. New platforms should be treated as customer signals that feed a decision support system, validating which investments drive immediate incrementality and longer-term organic growth, rather than as more reporting.
Todd argues that the search for one proof-of-success metric is outdated. Citing the IAB's updated podcast measurement guidelines, he says measurement is shifting from counting events to understanding how varied customer responses combine to create business value.
Todd argues that the brand versus performance debate is really investments versus customers. Channels, price, distribution and promotions interact along the buying journey, so AI should help optimize them as one connected system rather than speed up media optimization alone.
Todd argues that unused rewards are often an early sign of disappearing engagement, not happy savers. He says loyalty, media, pricing and other investments trigger linked customer responses, so teams should study the whole system instead of optimizing single metrics.
Todd argues that dashboards report what happened instead of guiding decisions. Predictive metrics like Google's are a step forward, but they must work at the level of investment choices and model interactions across the whole marketing portfolio to guide budget allocation.
Todd argues that most measurement systems optimize for one customer response, though marketing triggers many. He presents Agile Insights Modeling as a middle-out approach between top-down MMM and bottom-up MTA, separating incremental response from organic demand across the customer journey.
Todd argues that the "in-game advertising versus CTV" debate frames the choice wrongly. A new channel can amplify, complement, or cannibalize existing ones, so marketers should ask which combination of investments produces the greatest incremental business outcome.
Todd argues the K-shaped economy shows there is no one winning channel, not that Connected TV is the answer. Models that average responses across diverse customer segments mislead, so measurement should treat investments, customer responses, and outcomes as one system.
Todd uses a MediaPost piece on the K-shaped ad economy to argue that marketing measurement leans too heavily on averages. Segments respond to price, merchandising, assortment and media differently, so models should capture that variation instead of compressing it.
Todd argues that CMOs must move from measuring individual marketing activities to understanding the full commercial system. Models should cover multiple investments, responses, and outcomes together, and separate truly incremental sales from demand that would have happened or merely shifted.
Todd argues that analysts too quickly blame predictive models when the real fault lies upstream. Inconsistent definitions, broken processes, stale data and weak governance compound over time, so fixing the surrounding data system should come before replacing the model.
Todd argues that Walmart's acquisition of Vibe signals media, commerce and analytics merging into one system. He says channels measured separately hide how investments reinforce or compete with each other, and that modeling the whole system changed a retail client's decisions.
Todd argues that retail media data should ground marketing decisions in observed shopper behavior, not demographic assumptions. Its real value is showing how content, media, promotions, and merchandising interact, so brands stop optimizing channels in isolation.
Todd argues that evaluating retail media in isolation hides whether it creates new demand. Using a World Cup example, he describes modeling TV, retail media, digital, social and promotions together to find which investments amplified response and which merely shifted existing demand.
Todd argues that brands and agencies waste the World Cup by running isolated promotions during key matches. He says using search data to adjust strategy in real time keeps engagement going beyond the event, even for brands that are not FIFA sponsors.
Todd argues that judging AI one tool at a time yields what Melissa Reeves calls random acts of AI. He says the unit of analysis should be the connected system, and links this to Provalytics' Agile Impact Modeling approach to measuring how investments interact.
Todd argues that impressions and attention are not rival metrics. Impressions create the opportunity for engagement and attention raises the odds of action, so both should sit in one predictive framework judged by whether spend produces incremental growth.
Todd argues that AI in retail, such as Walmart's real-time recommendations, changes how customers buy across channels in ways traditional measurement overlooks. He recommends AI-powered approaches like Provalytics' Agile Impact Modeling to connect marketing activity, customer behavior, and financial results.
Todd argues that brand managers who stick to traditional campaigns miss how demand works as a network of variables that affect each other. Adaptive AI advertising surfaces those feedback loops and supports continuous learning over one-variable-at-a-time seasonal optimization.
Todd argues that high view counts can look impressive while contributing little to revenue. In a CPG measurement update, optimizing for conversions instead of impressions showed that more relevant, lower-view videos produced more purchase actions than widely viewed ones.
Todd argues that marketing leaders who let procurement delay agency payments risk their brand's agility and reputation. Drawing on a CPG client that moved to prompt payment, he says faster payment brought more responsive, more creative agency work.
Todd argues that putting most of a budget into Google Ads creates blind spots, because unclear pricing and conversion tracking make returns hard to verify. In a retail case, social and email outperformed expectations, and diversifying channels made ROI clearer.
Todd argues that marketers relying on demographic targeting miss the preferences of America's growing soccer audience. In a pilot study with a sports apparel brand, adding AI to media planning appeared to improve incrementality, though Todd notes results are far from perfect.
Todd argues that unified AI routing services like OpenRouter are essential rather than experimental for marketers. Using the right model for each task, instead of relying on historical data alone, should sharpen targeting and allow real-time strategy adjustments.
Todd argues that marketers who stick to familiar workflows out of fear of AI hold back their own progress. In his experience, automating data parsing and routine analysis lets agency teams spend their time on strategy and creative work.
Todd argues that marketing measurement built on static, siloed models of past behavior is outdated as AI-referred traffic grows. He says near real-time AI engagement data can turn those models into dynamic insights and support better retention campaigns.
Todd argues that marketers who adopt AI ad tools without validating their impact create blind spots in customer insights. He says human oversight is still needed to interpret AI outputs, and that he and Jeff Greenfield have concrete examples to share.
Todd argues that advertisers still lean on last-click attribution while channels like connected TV, podcasts and influencers drive impact without clicks. He says measurement should model how spend, timing, audiences and creative change outcomes, rather than track individual users.
Todd argues that marketers who rely on clickstream attribution are misreading what drives conversions. Last-click channels get credit while top-of-funnel work is defunded, and a probability-based modeling approach that needs no cookies or clicks gives a fuller picture.
Todd argues that piling up data leads to indecision and clouds the KPIs that matter. Citing Publicis's planned LiveRamp acquisition as a sign of data's importance, Todd says clarity, focus and action beat volume and sophisticated tooling.
Todd argues that sophisticated AI ad tools are stalling because of algorithmic complexity and steep learning curves. Over-reliance on AI risks campaigns that lack human touch, so data must be paired with creativity and strategic integration.
Todd argues that programmatic TV's targeting promise only pays off with measurement at a granular level. MMM that stops at channel or platform performance risks paying premiums for noise, so he says marketers should ditch aggregate measurement.
Todd argues that CTV's growth and precision targeting are wasted if marketing mix models stop at channel and platform level. Measurement needs granular detail on audiences, creatives, publishers and placements, or AI ends up optimizing on averages.
Todd argues that LinkedIn and Amazon are changing B2B advertising through CTV targeting of professionals. Aggregate marketing mix models will leave advertisers behind, so Todd favors granular measurement across campaigns, audiences, creatives, publishers, formats, geographies and placements.
Todd argues that mastering data is not enough in analytics. Insights need context from the audience's perspective to lead to action, and AI works best when paired with human intuition, advocate feedback and a feel for the unknown.
Todd argues that AI should enhance creativity rather than replace it. Automated insights lack impact without narrative, so Todd pairs AI with proprietary algorithms to build meaningful measurement strategies instead of generating more data, keeping human intuition central.
Todd argues that CEOs treat marketing as overhead because measurement stops at clicks and impressions. He says analytics should connect campaign details like platforms, audiences, creative and copy to business outcomes, turning numbers into narratives that show marketing's value.
Todd argues that gathering data is the easy part, and value comes from acting on it. Using TikTok analytics as an example, he says insights need creativity and intuition to become strategies that drive engagement and conversion.
Todd argues that organizations should check whether their strategies are ready before adopting AI. AI can produce options, but data scientists must weigh them, verify results rigorously, and keep human judgment central as data infrastructure and analytics frameworks adapt.
Todd, a longtime skeptic of AI hype, says GPT5.5 handles large datasets fast enough to free the data science team for better algorithm design and testing. He still insists automation must not displace human insight in how findings are interpreted.
Todd argues that TikTok analytics work best when granular, creative-level ROAS is paired with human intuition. He says a hesitant brand began investing with confidence once the data showed which sounds and challenges drove virality, instead of relying on platform-wide totals.
Todd argues that marketing data should be treated as an investment rather than a cost. Value comes from pairing analytics with creative intuition and empathy, since relying only on AI risks missing the nuances that connect with customers.
Todd argues that as AI takes over most tasks, people's value lies in intuition, creativity and strategic foresight. He says leaning too heavily on automation sidelines the human touch, and that the best strategies come when imagination meets data.
Todd points to Netflix's 16% ad revenue rise as evidence that brands are investing in digital advertising innovation. He warns that leaning only on algorithms risks losing the human touch, and says AI should enhance strategy rather than replace it.
Todd uses the example of AI predicting coffee orders to ask whether algorithms can capture the nuance of human-driven strategy. He argues that data models are approximations, and that AI should enhance marketers' decisions rather than replace empathy and intuition.
Todd argues that eMarketer's projection of Meta passing Google in ad revenue reflects a real market shift driven by AI personalization. Marketers should align their ad budgets with AI advances, and Todd's company is retooling its platform accordingly.
Todd argues that marketers confuse data volume with value, and that dashboards which don't guide decisions create an illusion of progress. He calls for reflection over reaction and for AI that works with established marketing know-how.
Todd argues that AI has become its own media channel, pulling budget from performance channels. With a third of ad spend already concentrated on about ten platforms, Todd warns against chasing hype and urges optimizing existing platforms and prioritizing actionable metrics.
Todd questions whether excitement around TSMC's record AI chip surge is pulling marketers away from fundamentals. Todd argues for using the surge to recalibrate, pairing new advances with established practice and favoring deliberate integration over rushed adoption.
Todd argues that the TikTok and HubSpot integration changes how marketing insights are gathered and used, combining social reach with CRM precision. He warns that teams who delay risk irrelevant campaigns as competitors move ahead and ad spending grows.