August 13, 2022

Digital Marketing Education

Latest news and events

Cookie death threatens multi-touch attribution, study finds

Today, mobile is a much bigger part of most marketer budgets, so marketing mix modeling can measure it better. But the growing role of digital platforms segmenting their data within “walled gardens,” makes it harder to track consumers even across digital media, much less other media. Take away cookies that track people from site to site and the job of attribution gets even harder.

‘Too damn complicated’

“Marketing has become too damn complicated,” said MMA CEO Greg Stuart. “The persistent and pervasive challenge of attribution,” he said, is one of the biggest challenges facing marketing today. “Our members are not only telling us that this is one of the most frustrating aspects of marketing today,” he said. “They are also rolling up their sleeves and coming together to take it on by sharing their best practices and innovative solutions.”

The MMA sees hope in recent tech workarounds that track marketing outcomes without tracking individuals. For example, the report cites recent research by Neustar (a partner in the MMA study) showing an MTA simulation using cohorts of 10,000 people across a million records produced an attribution bias (or error) of only 1%. That suggests workarounds such as Google’s Federated Learning of Cohorts could be highly accurate, said Lou Paskalis, the former Bank of America media executive who recently became president and chief operating officer of MMA Global.

Even before digital identifiers made it harder to track people, attribution had problems. The MMA report acknowledges the percentage of marketers who said they intended to adopt MTA never came close to what was projected in its prior-year surveys since 2016. Questions about data quality (ranging from inaccuracy to data-sharing resistance by walled gardens), disputes about methodology among different providers, organizational silos with different parts of the marketing organization supporting different solutions, and lack of leadership support all contribute to adoption lagging intentions, the MMA finds.

Overcoming cynicism

The MMA report also finds hope in overcoming one of those hurdles—cynicism spawned by providers of MTA, marketing mix modeling or other analytics solutions favoring only their own methodologies. A growing number of providers handle both MTA and marketing mix modeling, the MMA finds.

It’s getting more common for marketers to blend MTA with modeling, Paskalis said, for example by using attribution on media where it’s possible and modeling for the rest, which in turn is likely to make the modeling more accurate.

Moreover, while loss of cookies presents data challenges in digital, growth of connected and addressable TV is increasing the pool of media where it’s possible to attribute ad exposure directly to an individual or household’s purchase behavior, the MMA notes.

Better TV data is one reason General Motors has for hoping its third try at getting MTA right will work better than the prior two, said Hardy Faison, marketing track associate for GM, in the MMA report. “TV is such a big part of how we do marketing,” he said, including telling brand stories, introducing new models and drawing attention to dealers.

GM’s new approach to MTA also includes unifying analytics approaches with marketing-mix modeling, in part by having them both under the marketing department. In the past, a forecasting and analytics group ran attribution and the marketing department ran modeling at GM.

While MTA can live on, it’s going to have to adapt, Eddie Drake, senior VP of marketing data strategy, attribution and partnerships at Bank of America, said in a statement. “With data deprecation and constant change on the privacy horizon, MTA will not survive in its traditional form,” he said. “It takes real commitment to lead the charge, and it’s not for the faint of heart.”