Multi-Touch Attribution – We Hardly Knew Ye
Marketing mix modeling (MMM) was not designed to measure ROI for small-scale, targeted digital campaigns. In 2014, multi-touch attribution (MTA) was the dream solution, tracking person-level media impressions, tying them to sales, and precisely measuring the value of every touchpoint. Then the walls went up. In January 2015, concerned about data (revenue) leakage, Google restricted tracking pixels on their display ad network. In 2016, Cambridge Analytica improperly used Facebook data, sparking a deep dive into digital data privacy. Today, from what began as self-interest and expanded to privacy concerns, the two largest walled gardens – Google and Facebook – representing over half of US digital media spend, have largely shut off third party access to their person-level impression data.
Multi-Touch Attribution with Randomized Controlled Trials delivers on MTA’s promise. Like walled gardens’ RCT, MTA-RCT uses a random non-exposed control group for each addressable campaign. But by selecting it before it goes to the publisher, it enables a comprehensive database of hundreds of RCT results, integrated at the person level. MTA-RCT uses RCT to provide accurate, unbiased measurement for all campaigns in all walled gardens plus most other addressable media. Non-addressable media goes in the same model but without control groups. MTA-RCT addresses most of the major challenges we encountered with MTA:
- Measures ROI for all walled garden campaigns
- Minimal privacy issues
- Accurately measures small digital campaigns
- All touchpoints in one model, including TV
- Feeds standard reporting and planning tools
MTA-RCT is GDPR / CCPA compliant and does not need person-level impression data, just person-level sales and person-level control groups. No tracking pixel is required.
Highly Accurate With No Targeting Bias
MTA-RCT has lower error than MMM, MTA, and every other measurement approach we’ve ever tested except walled garden RCT. Like all RCT, by randomizing control group selection, MTA-RCT avoids targeting bias, a serious technical issue that affects MMM and MTA and misattributes differences between test and control groups to the treatment.
Based on Established Science
RCT with a known control group but unknown subset of the treatment group who actually got the treatment is a common form of RCT called “intention to treat” (ITT), dating back to the 1960s. It addresses the fact that in clinical trials, the treatment group doesn’t always follow the treatment. The same principle applies when we don’t know exactly who saw an ad impression, just a control group who didn’t. We can still accurately measure an average effect between the two groups and calculate ROI. And it’s not just us saying that. Researchers from Google, Netflix, and Yahoo agree with our reported error difference between MTA-RCT (ITT) and walled garden RCT, reporting a 31% difference between the two1 versus our 33% (from 15% error to 10%).
Connected to Actionable Planning Tools
Our MTA-RCT analytics can interface directly to our industry-leading reporting, planning, and budget optimization tools. RCTs rarely do that as each experiment is considered a separate, isolated analysis.
MTA-RCT is a new approach to a recent problem and will continue to evolve. But in a dynamic regulatory environment regarding consumer privacy, one thing we can probably count on is the continued ability to block ads. We see a solid, stable future for MTA-RCT and similar approaches, such as the ongoing RCT21 proof-of-concept study championed by the Advertising Research Foundation and led by researchers with deep expertise in RCT and walled garden RCT. Based on our own research and ongoing pilot, we expect good things from MTA-RCT and other intention-to-treat based approaches to marketing measurement.