LAS VEGAS: Telecoms group AT&T is enhancing its digital measurement capabilities to ensure it has truly granular insights into upper-funnel metrics, as well as into lower-funnel indicators such as last-touch attribution.

Fiona Carter, the Chief Brand Officer at AT&T, told delegates at CES 2017 that while the firm's traditional strategy tapped "mass-marketing" and "mass-tonnage", its increasing activity on digital requires different measurement capabilities.

More specifically, she asserted that the company had deep knowledge of "lower-funnel" digital metrics – like clickthroughs and conversions – but not always of the upper-funnel indicators.

"We have a very rigorous econometric media-mix model, which is fantastic. But it was [focused on] the traditional channels, and it is last-touch on digital," Carter said. (For more, read Warc's exclusive report: AT&T builds its digital marketing muscle with OTT.)

"What we're trying to do is expand that to be a much more sophisticated look at multi-touch attribution. We really want to understand where each part of the funnel plays and how important it is."

In response, the company has doubled down on its efforts to understand the "brand-building benefits" of digital in order to supplement the granular knowledge that can be gleaned through last-touch attribution and similar sources of insight.

"I have a very strong belief in the value of the brand and driving a business outcome," Carter told the CES assembly gathered in Las Vegas.

"And when you're in a very retail-competitive business, you tend to sacrifice that. So we have a real focus on multi-touch [attribution] throughout the funnel.

"We're also trying to look at a single view of our target, so that we can see them across the formats, across the platforms, and understand this magical alchemy of how this customer journey now happens so that we can be smarter, and better, next time."

Looking ahead, emerging technologies such as artificial intelligence could have extremely powerful implications for these kinds of measurement.

In this context, Carter suggested it is intriguing to think of the possibilities arising from "applying machine-learning capabilities to econometric models that are telling us which parts of the funnel work, and how they work together, and when – and the importance of the upper funnel along with the lower funnel."

Data sourced from Warc