In the last few years, “identity graph” technology has emerged to help advertisers tackle the problem of conceptualizing a single consumer across multiple devices.

For many in connected TV, the focus has settled on household-level identity data.

But, in this video interview with Beet.TV, Scott McKinley, CEO of ad-tech firm Truthset, says today’s data isn’t accurate enough.


“The accuracy of general identity in terms of … understanding which people go into a household, which IPs are associated with that household, which devices are associated with those people… I think we’re right around 50%, sometimes better or worse,” McKinley claims.

“That’s not good enough to overthrow any legacy systems.

“We have to try harder to understand the quality and improve the quality of these fuel sources before we just shove them into these engines and expect a different car to win the race.”

Household focus

Many in CTV are focusing on household-level identity for two reasons:

  1. Subscriptions for internet and TV services tend to be held at the household level.
  2. TV is a device still reckoned to be shared or viewed communally, making the promise of individual-targeted ads more trickier for the time-being.

But Truthset’s McKinsey says improvements in household-level identity graphs have really been about growing awareness of their importance. He says the quality of the graphs has not improved much.

That is a concern because, already, there is growing industry demand for household-level measurement.

Source of truth

San Francisco-based Truthset was founded in 2019 by veterans from Nielsen, Salesforce, LiveRamp and Procter & Gamble.

Its Truthset Inside product is a multi-sourced service to standardize demographic accuracy for audience targeting and media measurement.

It offers Truthscores for 25 demographics based on more than 850 million digital IDs.

Truthscore is Truthset’s attempt to generate a numerical rating corresponding to the accuracy of consumer data.

Expanded offering

Now Truthset is expanding its offering.

“As dollars move from linear into more addressable forms of television, we need to understand who’s in the household and do a better job of ringing data to describe those audiences and predict them,” McKinley says.

“We are now recruiting participants to join a collective that will run the same process we’ve done for attribute data…

“We will actually build confidence scores for the entire census of households to understand ‘which household are you really likely to be in and which household am I really likely to be in?’.

“You need to understand the impact on the entire household and the buying behaviour of the household.”

McKinley spoke at the Advanced Advertising Summit in New York. He was interviewed for Beet.TV by Jon Watts, Editorial Director, Beet.TV Events.