Losealikes

On January 31, 1969 The Lompoc Record (a newspaper from Lompoc, California) carried an uncredited article from the United Press International Wire service with the Nipsey Russell-esque headline “Nets, Mets even play same way.” The author goes on to poke fun at the Nets by calling them a “welcome visitor” to any town they go to since they are likely to show up and lose to the home team.

The author then (and given the year, we have to presume it was a guy, and probably a guy that wore a fedora, and smoked cigarettes) went on to call the Nets and Mets “sound-alikes and also lose-alikes as the Nets stagger around the ABA like men in a fog.”

The reason for this history lesson is that this week I came up with the term losealike and since the only other use of the term (according to the all knowing Google) is this 1969 article (where it was spelled lose-alike – i.e. TOTALLY DIFFERENT), consider this me planting the flag on behalf of Trust Insights as bringing the concept of losealike to marketing.

If you’ve stuck with my absurdity thus far you’re probably asking “Enough you prat! What does losealike mean?” Well, ok, nobody says ‘prat‘ but you get the idea. The short version is that over the past few years we’ve seen the growth of marketers using data to generate lists that look like their customers – lookalikes. But what if you generated a list of prospects not like your customers?

Lookalikes are people you don’t know about yet (not on any of your lists), who are, in some traits, similar to your customers. In theory they should be prospects a lot more qualified than the general public (Everything riding on the “in theory” here.) It’s very simple for Facebook or Twitter, you take a customer list, run a campaign against a lookalike list generated from the customer list and hopefully it’s your best hit rate ever and you are basking in the glory of your most successful ad campaign.

Working on a data project this week we had a client that’s segemented their customers and prospects. They are doing lookalike marketing and it’s going well so we were brainstorming where to hunt next. As we were reviewing the segments there was one we came to where they said “this list is TERRIBLE, these people never buy anything EVERRRRR.”

Queue the hokey metaphor for new idea here – lightning strike, chocolate bar landing in PB jar (see the call back to rhyming there? you’re welcome), person saying “Why… that’s just crazy enough to work!”

You take a big list that you don’t know anything about yet. Use your favorite statistical analysis / machine learning package to find the people on the big list most like the ones on the list that your team knows is TERRIBLE. Split the new list and suppress the names like the terrible ones from the campaign. Test to confirm the “terrible” list is truly worse. If it validates, from now on every time you run it you’re creating the losealike list! And you don’t have to advertise to, call, email them! The time you are not spending is money in your pocket!

Have fun creating your losealike lists (TM Trust insights 2019)!

Updated: Nipsey Russell corrected to two “L”s