Buyer segmentation serving both brand strategy & media activation.
Segmentation methodology fusing first-party purchase data with attitudinal and behavioral layers across ZX Ventures' 17 craft beer brands. Served both brand strategy and media activation without compromise.
Most customer segmentations live in one of two worlds. Either they’re built from behavioral and transactional data (great for media targeting, weak for brand positioning) or they’re built from attitudinal research (great for positioning, weak for activation). The methodology we designed for ZX Ventures' craft beer venture portfolio needed to do both, across 17 brands, in a way the brand teams could actually use.
The methodology integrated three data streams.
- first-party purchase data from ZX’s owned e-commerce platforms (Emporio, Ze, BeerHawk): transactional behavior, recency and frequency, brand affinity, discount preference, customer lifetime value.
- behavioral data on interests (entertainment, shopping, travel, fitness), media and tech consumption, and shopper behavior.
- attitudinal data: craft beer category attitudes plus demographic and psychographic layers from first-party survey work.
It produced two layers of segmentation. A global value-based segmentation that worked at the portfolio level (Brand Loyalists, Cross Shoppers, Steady Shoppers, Deal Hunters). And a layer of brand-specific personas grounded in the attitudinal and behavioral data.
The methodology mapped to four downstream applications the brand teams used directly: informing the Brand Muse, refining positioning and passion points, building digital audiences for media targeting, and constructing targeted comms plans. Brand strategy and media tactics, served by the same segmentation, without forcing either team to compromise on the methodology.
The portfolio context made this exercise a little more complex than usual. Doing it once for one brand is reasonably straightforward. Doing it once in a way that holds across 17 brands operating in different geographies, with different positioning, different competitive sets, and different first-party data maturity, is a different kind of design problem.
One caveat we named openly at the time: the e-commerce data layer biased the sample toward customers who already shopped that way. IMO it’s better to know it and account for it than pretend the methodology was perfectly clean.