Match rates have been one of the hot topics in the ad tech world for some time now and still comes up frequently in conversations with digital marketers. The topic continues to perplex because match rates are a widely misunderstood metric that are not entirely useful on their own.

In some ways, this is understandable. Match rates are an easily digestible metric by which a marketer can determine how well two distinct systems overlap with data, be it a DMP, a DSP, or an identity resolution service. It’s not dissimilar to how marketers used to ask for “click through rates” from ad networks.

However, match rates come with a dirty secret, a decay in their efficiency that threatens to undermine all the hard work. The average user owns 3.64 devices according to GlobalWebIndex, but likely has many more IDs, such as browser cookies, mobile advertising IDs, hashed email addresses and signups for connected TV services such as Roku and Apple TV.

In order to tell consumers a contextual, consistent story across devices, channels, and platforms, marketers must link these identities to a single customer. At that point, the marketer must match the ID across the places where they can find it – the elusive “match rate.”

The Match Rate Challenge

Today, daisy chaining identity solutions to determine a match rate is the marketer’s only option. For instance, if a marketer has a cohort of 1 million people who are addressable by email, it’s likely to only be able to match around 40% of those emails to cookie IDs.

This is because cookies are a changeable measure of identity. They expire as we move across devices, we delete them, and sometimes match pixels simply don’t fire as they should due to technology lapses or ad-blocking. That cohort then needs to be matched across the devices they use, which can account for another 50%. Once a marketer has run segmentation on the cohort in their activation platform, the total addressable audience could be as just 8% of the initial audience. Clearly, daisy chaining identity solutions is not the answer.

Beyond the Match Rate Challenge: First Party Data Matching

For a long time, many in the industry have thrown up their hands and said: “we can’t do anything about those cookies.” Others have been less than transparent and talked in terms of “matchable populations” so that they can get around admitting they can only match against a small subset of cookies.

The key to solving the match rate challenge for marketers lies in applying machine learning algorithms to first-party data sets. This can be accomplished by running a lookup against a cross-device graph, running a customer identity algorithm against the first party data, or a combination of the two. Either of these options will also need constant attention and optimization over time.

It’s vital to get match rates as high as possible, because low match rates make it hard to be sophisticated with data-led marketing strategies. Marketers want to be able to make the most of their first party data sets, and the match rate challenge forces them to ignore over half of that data.

For a long time, the industry has tried to address this through expanding cookie footprints. Many vendors have tried to pitch the “one ID to solve them all” panacea, with the result being simply another competing standard. The way to a real step change progression is through applying machine intelligence to first party data sets, in addition to traditional methods. Only then will marketers truly be able to solve the match rate challenge and maximize the value of their first party data.