What is a “match rate?”
A match rate refers to the percent of users from a file that an onboarder is able to find and anonymously tag with data. As data onboarding becomes a core part of every marketer’s toolbox, knowing the match rate for your user set is critical for understanding the size of your addressable online audience.
Let’s say your CRM has three million email addresses. If the match rate is 40%, you can expect your online audience—the number of users you can target with online display—to be around 1.2 million users.
How is match rate calculated?
It’s important to clarify what defines “match rate” because it can be calculated in different ways. At LiveRamp, we calculate match rate with the following formula:
# of unique records that matched to at least one RampID solution / total # of unique records.
The numerator, it’s important to note, is not the total number of identifiers, rather, it’s the number of unique records matched to an identifier. Often, there are multiple identifiers (cookies, device IDs, IP addresses) per record because users browse the internet across multiple devices and browsers. So if a company counts those identifiers instead of records matched, they are actually inflating the match rate.
Additionally, the “record” needs to be defined consistently. If you input a list of addresses (so the denominator is measured in households), but the match is being done at the name and postal address (so the numerator is measured in individuals), you are again going to get an inflated match rate because there are multiple individuals per household.
What drives match rates up and down?
The number of users that a data onboarder sees in its match network is what primarily drives match rates. The match partner within the network sends hashed registration information, such as a de-identified email address, which is then used to match back to offline data, such as purchase history and demographics. So the number of users found via the match network directly drives match rates up or down.
The second biggest driver of match rate is the precision of the match. A data onboarder can dial down precision, increasing the number of “false positives” (bad matches) in the audience file. This will lead to a perceived greater match rate and inflated sense of scale. Some ways to do this include loosening match criteria (such as only requiring name and city instead of an exact address match), neighborhood (i.e. zip or zip+4) level matching, or IP address matching. For some marketers, this is perfectly acceptable (imagine a supermarket trying to reach people in a certain neighborhood).
However, for marketers trying to reach their actual customers, being flexible about required inputs to achieve greater scale is a risky trade-off as it dilutes the audience, makes it impossible to accurately measure the success of cross-channel campaigns, and wastes online ad spend.
There are plenty of other ways to achieve scale online, such as lookalike modeling off of exact matches. But without the accurate match, some of the more established use cases of onboarding, such as retargeting and analytics, cannot be relied upon. As they say in the data world: “garbage in, garbage out.”
What can you do to ensure high match rates without compromising precision?
As a marketer, you should understand this trade off as you consider potential onboarding partners. Many marketers have been duped into buying an onboarding solution that promised a “high match rate” only to find that the data was entirely inaccurate or the match rate was being calculated in a wonky way that inflated numbers.
To avoid this type of scenario, ask prospective onboarding partners how they calculate match rate, where they get their online matches (i.e. do they have direct relationships with high-quality publishers?), and what their accuracy standards are.