Bad Seeds Grow Bad Customers

Retention begins Pre-Acquisition

My acquisition numbers are looking great. It’s just my one & done customers that are killing me.

Then your acquisition numbers aren’t really looking great…

Acquisition isn’t about transactions

We all know this truth to be self-evident…but sometimes a reminder is good. When we are lofting our media out on Meta, our goal isn’t to pick up transactions but to pick up customers. Sure – we like a healthy ROAS (Return on Ad Spend)…even better…some NCCD (net customer contribution dollars) to fill the coffers and secure some accolades… unfortunately, that really isn’t the measure of customer acquisition…and why your one and done rates are still a problem.

Lookalike Seeds (LaLs)

For those unaware – Lookalikes are a “sample” audience (commonly called “seed” audiences) that you send to your advertising partners and say, “Find me people who look like these.” I built my first Lookalike program over a decade ago with my friends from at The FinishLine – and the 4x-12x increase in ROAS over traditional ad targeting pretty much signaled the end of an era in “old school media targeting.”

Lookalike Advertising: Not all wine and ROASes

The problem with Lookalikes (prt1): Off Target

WARNING: REAL WORLD EXAMPLE. Let’s say that you sell smoking jackets – cool smoking jackets…the kind with leather on the elbows. Your core customer base is an aging collection of college professors from Ivy League schools. However, you want to expand into an emerging market of younger, handlebar-mustached, craft-cocktail-downing hipsters (cause that is the future). You cut a seed audience of recent purchasers (aged college professors), but your ad targeting criteria (and creative) is for your mustached hipster. Yeah – you see the issue here…

Conflicting criteria between seeds and targeting will cost
between 2x to 5x incremental ROAS

Aside from making Meta (even more) schizophrenic, you just tanked your program. Conflicting criteria is where most “vendor” based Lookalike (LAL) automation programs fall flat…to the point of (at times) driving negative lift…and it happens frequently.

The problem with Look ALikes (prt2): Bad Seeds (grow bad customers)

Let’s talk about the Lookalikes themselves. What criteria are you using? Are you pulling people who purchased in the last 30 days? People who browsed? What about factoring in customer LTV (Life Time Value)? Well…recent purchasers are reflective of the past…and a lot can happen in 30 days. Browsers….yeah – it’s intent…but not $$$. And LTV – I love the idea, but you are pulling a look similar to LAL for people close to the END of their purchasing journey.

The quality of your Lookalike (LAL) seeds is the
single biggest contributor to your 1 & done rate.

Stop Living in the Past: AI-Assisted Predictive Seeds

What if you could produce a list of Lookalike seeds that will be purchasing smoking jackets in the next 30 – 60 days? What if those customers were also selected by their persona (twitchy moustaches and all). What if those customers all had a high predicted future value? What if that seed wasn’t just reflective of the predictive purchase of the new felt smoking jack – oh, but also the matching slippers?

Predictive seeds acquire customers on a journey, not simply transacting.

My Retail family – this isn’t hyperbole. I’m not an analyst painting pristine pictures of some rose-colored nirvana-esk view of what could be…this is real. In fact – it’s been proven by brands that you know. The simple reality is that –

1. Predictive Seeds will increase ROAS an incremental 7X over existing Lookalikes (LALs)
2. Predictive seeds will more than halve your one-and-done rates
3. Predictive seeds 3x strategic acquisition effectiveness (acquiring your coveted “who”)

Oh…and as a little bonus…predictive seeds work for new customer acquisition for your website…wait for it…waaaaaaait for it….stores and marketplaces! Yeah…and it’s damn effective (especially when coupled with DMP in-market signals).