• by Dave Morgan , Featured Contributor, March 2, 2023

I’m on a plane returning to New York City from several days in the Bay Area, where I attended LiveRamp’s RampUp 2023. The event is always one of the most important if you care about where the world of advertising and marketing meets data. Since virtually every company in adtech, martech and digital publishing and advertising leverages LiveRamp’s identity technology in one way or another, folks from pretty much every company in the business attend, so it’s not an event I can miss.

The theme this year was “data collaboration,” as the company tries to cement and extend its position as plumbing between most of the publishers, platforms and cloud services trying to leverage first-party audience and customer data in ad targeting, measurement, attribution and analytics.

First-party data has really come to the forefront in our industry the past few years with the explosive rise of retail media networks, connected TV advertising and clean-room technologies that can enable parties to match and map proprietary and private data with each other without having to either move the data or expose it improperly. Being able to truly leverage real customer data is exciting for an industry that has lived way too long pinning its data-driven efforts on look-alike audience models, which have become so diluted that everyone is in virtually every look-alike segment.

It’s great that first-party data plumbing is coming together in a market that is increasingly fragmented and complex. However, one of the things that I worry about a lot -- and given conversations I was part of these past days, I’m not alone here -- is data quality. Great plumbing can sometimes add a luster of provenance to the data moving through it. But, as all in the data world well know, garbage in, garbage out.

With better, faster, more ubiquitous plumbing to match and leverage data in our campaigns and analysis, we can also better apply the essential steps of data hygiene, verification and validation that what is claimed in a data set is in fact true and represented in that data set. Techniques to do this work are well-known and well established in the industry.

Of course, not everyone wants to do this. Testing and validating data add additional steps -- and cost! -- to the campaign process. And, most critically, it can expose data and segments that are not what they claim to be, or what the buyers or sellers hope they are.

All too often, friction and fear raise their ugly heads here, and folks in our ecosystem choose to skip the data quality checking steps. This is short-sighted, dangerous, and wrong, and will only hurt our industry’s development.

There are several areas where I am particularly worried. Here are three:

CTV audiences defined by mis-fit indenting graphs. How most people watch television is quite different from how they use their mobile or web devices. Too many of the identity graphs and data segmentations being used today in connected TV ad campaigns are “reused” from mobile, PC and tablet browsing and usage, which poorly fit into the lean-back, multiviewer world that is TV. This creates a “faux precision” in campaign targeting and reporting that makes people feel good during the initial purchase, it, but is full of empty calories when it comes to delivering any kind of real effectiveness.

Fake data. Audience and purchase data is money for its owners, suppliers and enablers. It’s also an intangible asset that can be created from whole cloth. When people can make money from things that they can make and trade on for virtually nothing, many people do, particularly in a digital programmatic world where most parties in transactions don’t -- or can’t -- know all  the parties in their transactions. Just as we have seen plenty of digital inventory fraud, we are only going to see more data fraud.

Impenetrable, perverting, co-mingled business models. The audience and purchase data supply, provisioning and storage ecosystem is massive, fragmented and growing fast. Companies often create economic incentives to work with their cloud, service, or data source and not their competitors.

These incentives were designed to steer business and money in certain ways, and they are very effective at accomplishing that goal. In a world where incentive payments -- not data quality or campaign results -- drive allocations, quality and clients’ best interests rarely win out; money does. Without full and complete transparency on these deals to all of the parties involved -- something our industry is terrible at -- we will end up in bad places.

Data ubiquity is a powerful and amazing thing. I am really excited about how it will impact our industry. But we can’t lose sight of the critical importance of data quality -- and the reason for its being -- in that excitement.