We scan new podcasts and send you the top 5 insights daily.
Paul Meinshausen says marketing still measures aggregate averages for a campaign sent to a million people. Agentic personalization needs counterfactuals and incrementality at the individual user and message level. He compares this to Netflix measuring movies at scene level rather than asking if viewers liked the film.
To prove marketing's value beyond clicks, use brand lift surveys (A/B testing ads) to measure awareness shifts. For sales impact, use causal impact modeling to compare forecasted sales (based on historical data) with actual sales during a campaign, isolating the campaign's "bump."
Human marketers get trapped by averages, even within segments. AI-powered personalization can test countless variations at scale, revealing unexpected "winning" messages that resonate with sub-segments, leading to significant performance lifts and unlocking hidden growth.
To gauge the real impact of a campaign, isolate a small percentage of your audience (a "holdout group") from all marketing. The difference in conversion rates between this group and the targeted audience reveals your actual performance lift, moving beyond simple conversion metrics.
Traditional digital attribution is broken. Marketers should adopt measurement models from the pre-digital era, like those used by ad agencies. This involves running controlled experiments (e.g., billboards in different cities) and measuring for lift and incrementality rather than last-click attribution.
Instead of batching users into lists for A/B tests, AI can analyze each individual's complete behavioral history in real-time. It then deploys a uniquely bespoke message at the optimal moment for that single user, a level of personalization that makes static segmentation primitive by comparison.
Moving beyond basic attribution, LinkedIn's new Conversion Lift Testing tool measures the causal impact of campaigns. It compares conversions between an ad-exposed group and a control group that saw no ads, allowing marketers to determine the true incremental value generated by their advertising.
Don't trust the CAC reported by platforms like Meta, which can be off by 20% or more. Brands must build their own attribution models using incrementality testing (e.g., turning off ads in one geo) to understand the true, causal impact of each channel.
Enterprise ABM cannot be evaluated through linear attribution models that connect a single asset to an opportunity. Success must instead be measured by comparing meeting conversion rates and pipeline lift across targeted accounts against similar accounts that received standard marketing.
Relying solely on ROAS is outdated. A comprehensive strategy requires a three-tiered approach: daily attribution for media buyers, incrementality studies for media planners, and longer-term Marketing Mix Models (MMMs) for CMO-level strategic decisions.
In the privacy-first era, deterministic attribution is declining. Leaders must accept this new reality by getting comfortable with confidence intervals from probabilistic models and focusing on long-term incrementality testing, rather than chasing precise, absolute numbers from legacy models.