Due to signal loss from cookie deprecation, no single model like MTA or MMM is sufficient. The new gold standard is using all available algorithms together in a machine learning framework, allowing them to influence each other for a more accurate ROI picture.
Related Insights
Relying solely on data leads to ineffective marketing. Lasting impact comes from integrating three pillars: behavioral science (the 'why'), creativity (the 'how' to cut through noise), and data (the 'who' to target). Neglecting any one pillar cripples the entire strategy.
With AI enabling precise control over media spend, key performance indicators are changing. Brands now move beyond simple Return on Ad Spend (ROAS) to more sophisticated metrics like incremental ROAS and contribution margin, reflecting a new emphasis on profitable growth rather than just volume.
AI's most significant impact is not just campaign optimization but its ability to break down data silos. By combining loyalty, e-commerce, and in-store interaction data, retailers can create a holistic customer view, enabling truly adaptive and intelligent marketing across all channels.
Cookie deprecation blinds ad platforms like Google and Meta to on-site conversion quality. Marketers can gain a significant performance edge by creating a feedback loop, pushing their attributed first-party data (like lifetime value and margins) back into the platforms' AI systems in near real-time.
New measurement tools are moving beyond probabilistic models (guessing based on IP/device) to deterministic view-through attribution. By using first-party data like platform logins, marketers can now directly match an ad impression to a purchase, solving a major measurement challenge.
Standard attribution models, even multi-touch, fail to credit influential, non-clickable touchpoints like a child watching a Netflix show that inspires a purchase. This "Hot Wheels Problem" highlights the need to account for view-through attribution and the full, often hidden, customer journey.
A common attribution error is assigning all sales to paid marketing activities. In reality, most brands have a strong "baseline"—sales that would occur even without marketing. Accurate measurement requires modeling this baseline first, then attributing only the incremental lift from campaigns.
Shift the mindset from a brand vs. performance dichotomy. All marketing should be measured for performance. For brand initiatives, use metrics like branded search volume per dollar spent to quantify impact and tie "fluffy" activities to tangible growth outcomes.
AI's growth is hampered by a measurement problem, much like early digital advertising. The industry's acceleration won't come from better AI models alone, but from building a 'boring' infrastructure, like Comscore did for ads, to prove the tools actually work.
Solely judging marketing by last-touch attribution creates a false reality. This narrow metric consistently favors predictable channels like search and email, discouraging investment in brand building and creative storytelling that influence buyers throughout their journey. It's a losing battle if it's the only basis for decision-making.
