I’ve taken to ignoring most of the text in trade industry publications until I’ve looked at the numbers. Often, the math doesn’t match the narrative or tells another story completely.

Case in point: a couple of days ago, this article about an identity management platform showed up in AdExchanger with a brief case study about how the platform helped clean up a restaurant chain’s data. From the article:

After going through [vendor’s] pre-onboarding vetting process, one client, a casual dining restaurant popular with an older demographic, found that 13% of its file [emphasis mine] – between 13 million and 14 million records – were no longer alive.

The 13% had me fire up the calculator. That means the restaurant chain had 100-108MM records in their database. There are 51.3MM Americans aged 65+, which means “older” to me. Assuming the restaurant chain’s marketers are the typical Millennial-obsessed frauds populating many marketing departments, “older” might mean the 135.2MM 45+ US adults.

In the former case you’ve got maybe 2 records per “older” U.S. adult and in the latter it implies you’ve got 80% of all “older” adults on file. Nobody has 80% penetration, particularly in a business where lots of people still pay with cash and offer no PII in exchange for lunch.

Stop right there, do not pass go, do not collect $200. There is no way either of those numbers are right. You don’t need a vendor to tell you that about your file, if you’re competent. And it gets worse when we look at the deceased rate.

The mortality rate for 55-64 year olds is .878% (NIH stats, 2007), 2.01% for 65-77 year olds, 5.01% for 75-84 year olds and 13% for 85+.

In a good file, per industry standards, perhaps .5% to 1.0% of your records will be deceased. It’s a little lower for younger customers in a well maintained file and a little higher for older customers in an inadequately managed file.

Assuming the business is growing at the rate of GDP and adding records at roughly that rate, they’re probably losing ~0.8%-ish of their customers annually to the Grim Reaper.

A 13% deceased rate is FUBAR. And all the identity in the world isn’t going to fix it. They’ve got a gigantic process problem there. Either:

• They’ve got a database as old as the hills, or;
• They’re pouring in lots of duplicate records of dead people, or;
• Their logic to check records on input is lousy or non-existent, or;
• They never, ever purge a record or; (most likely)
• They don’t really have a customer database

The real story here isn’t identity matching. The real story here is the process is TARFU, maybe even FUBAR.

Takeaway: Before you spend money on cleaning up your data, check your process for the the collection, cleansing, appending and disposal (yes, disposal) of old or inaccurate data. And win.