Sam’s Club Connect, the membership-based retailer’s ad business, launched a suite of new targeting and measurement products on Wednesday. Brands can now create custom predictive modeling audiences and segments for churn prevention or waning brand loyalty, as well as options for targeting shoppers who may be in market for a higher-end item with a longer replacement cycle (like a TV or fridge). The company is also expanding its Omni-Impact offering, an AI-backed measurement solution, to include in-store retail and Meta ad campaigns.
The traditional model is to retarget and attribute campaigns based on recent purchases. But the new suite of targeting products extends Sam’s Club’s data in more forward-looking ways, said Harvey Ma, VP and general manager of Sam’s Club Connect (the “Connect” branding was added in June, following the line of parent company Walmart).
One new product, called Predictive Precision Targeting, creates a segment of households that have not previously bought the product being advertised. Procter & Gamble’s tested the product with their Cascade Platinum dishwashing detergent. Sam’s Club “pre-identified” some 800,000 households with a propensity to upgrade to the detergent pods. Ma said there was a 94% accuracy rate on those households, and the total those shoppers spent on Cascade Platinum Plus was up 71% compared to their initial campaign.
One major advantage of being a retailer, and a membership-based warehouse retailer in particular, is the longitudinal data Sam’s Club has accumulated on its members, he said. For one thing, even when people pay using non-traceable, non-attributable cash, Sam’s Club still ties that purchase to a person and household-based account. And then there’s the retention of shopping behaviors and purchases over time.
The historical data is used to model out audience segments for churn prevention and brand loyalty, and that data is integrated into another new tool released on Wednesday. Some brand marketers refer to this audience as “lapsed or staggered buyers,” Ma said. But the idea is to identify people who have been customers, and now seem to be waning or perhaps stopped buying the item.
The new predictive targeting and churn prevention metrics also follow a few months behind the launch of a brand and attribute targeting product that allows for conquesting campaigns, when marketers deliberately target a rival brand’s loyal customers, or in-category buyers who don’t buy their product.
Ma said the brand targeting product comes with guardrails, so as not to simply retarget a rival’s recent customers, but which still allow for one brand to target buyers within their category or with particular attributes (like, say, shoppers for other dish detergent pods, or buyers of other grain-free, sugar-free cereals).
Conquesting is touchy subject, and requires more controls. Which is why products like conquesting and the new predictive targeting are available by managed service.
Ma said that for Sam’s Club connect has a big advantage in that the retailer sees across so many brands and understands how shopping for one product or category might indicate different types of purchases down the road.
Another product released today creates segments that Sam’s Club predicts will be in the market for rarer, high-end purchases. This service is being tested this holiday season with a focus on driving TV sales. The retailer has been tracking customer activity in adjacent categories like TV audio soundbars, gaming consoles and streaming devices that might indicate someone will buy a new television.
“But you can see how the model can be applied across a number of different items,” Ma said. Aside from TVs, for instance, he hypothesized that someone’s general laundry product purchases might indicate that their washing machine is starting to fail or may be due for an upgrade.
Sam’s Club’s long-term and fully addressable archive of shopping data, plus the capabilities of LLMs and AI models to forecast buyers, is opening up new targeting metrics and ways to confidently assert ROAS on sales that haven’t yet occurred, he said.
“What we’re trying to demonstrate is that much more than historical purchase behavior, this machine learning model is an indication of what might come 12 months into a member’s life cycle,” Ma said.
