Season 4 Episode 9
Redefining the Future of Segmentation with Lÿden Foust, CEO at Spatial.ai

Episode Summary

In this episode of The Marketing Rapport, host Tim Finnigan sits down with Lÿden Foust, CEO of Spatial.ai, a company that has spent 10 years building behavioral segmentation that goes far beyond demographics. The central argument of the episode is simple and disruptive: the assumptions brands make about their customers based on age, income, and zip code are the things everyone already knows.

Competitive advantage comes from understanding what people actually do, where they spend, what they follow, where their phones go, and that requires a fundamentally different approach to segmentation.

Lÿden walks through how Spatial.ai classifies every household in the United States into one of 80 behavioral segments using four data inputs: credit card spend, social media follows, mobile device movement, and household-level demographics powered by InfutorData.

He shares vivid case studies — a furniture retailer using harbinger customer segments to predict taste trends before the Fourth of July, an aquarium experience that discovered young professional singles as a surprise audience and turned them into a date night revenue stream overnight, and Uniqlo capturing four percent market share in Texas within months by identifying a first- and second-generation immigrant segment that e-commerce data had already revealed was buying. He also unpacks the barbell economy, the role of culture in segmentation, and why social media following is an affinity signal, not a purchase signal.

The episode closes with Lÿden explaining how InfutorData makes household-level segmentation possible — because birds of a feather don’t always flock together, and the person three miles away is a completely different customer than the one next door. His framework is built for action: append your data, analyze which segments drive revenue, and activate marketing or site selection in under 30 minutes. The takeaway for any brand: a PowerPoint persona is not a strategy. Actionable segmentation is.

Guest-at-a-Glance

Key Insights

• Demographics tell you who someone is, not what they do
  Most segmentation systems start and end with demographics: age, income, marital status, zip code. These inputs are useful, but they have a fundamental flaw — people frequently behave in ways their demographics don’t predict. A household making $200,000 a year might shop exclusively at Costco and spend heavily on private school tuition and Sephora, while ignoring the luxury retail that demographic targeting would send their way. A lower-income segment might out-index dramatically on high-end footwear. The census tells you what a neighborhood looked like years ago. It does not tell you what those people are doing today. Spatial.ai’s approach layers four signals together: where people spend their money via credit card data, what they aspire to via social media follows, where their devices actually go via mobile movement data, and household-level demographic context via InfutorData. The result is a classification that explains behavior, not just identity — and that behavioral signal is what makes the difference between a campaign that converts and one that misses entirely.

• First-party data is a goldmine, but it can’t see what it can’t see
  First-party data is always the right place to start. If you don’t have a track record, you’re shooting in the dark on who your best customers actually are. But first-party data has a hard ceiling: it only shows you the customers you already have. It confirms what you suspect. It doesn’t reveal what you’re missing. The SeaQuest case study makes this vivid — an aquarium experience in malls that believed its customers were lower-income moms discovered through third-party behavioral segmentation that young professional singles were its second-largest audience. That discovery unlocked a date night campaign that became the number one selling ticket package across all 14 locations overnight. No amount of first-party data analysis would have surfaced that segment, because those customers weren’t in the database. They were in the mall. Third-party behavioral data lets brands find the customers they don’t know they have, reach audiences their competitors haven’t identified, and make decisions based on what’s true in the market — not just what’s true in their CRM.

• Social media following is an affinity signal, not a purchase signal
  Brands consistently misread social media data in two ways. The first is tracking sentiment and hashtags — which is noise, because people are talking about themselves, not about brands. The second is treating social following as a proxy for buying intent. Following Gucci on Instagram is not the same as spending at Gucci. Lÿden Foust is explicit about this: there are entire segments in Spatial.ai’s data that follow luxury fashion brands at high rates but whose credit card data shows zero spend at those locations. What that following does signal is affinity — a real relationship between those people and that brand’s identity, aesthetic, or values. That affinity is actionable, but only if you understand it correctly. It means there is an audience worth reaching who has the right orientation toward the brand. It doesn’t mean they’re already buying. The right approach treats social data as one input among many — meaningful for what it reveals about interest and aspiration, but incomplete without behavioral and spend data to validate whether that interest translates into action.

• Segmentation is only useful if it’s actionable
  The insight that drove Lÿden Foust to build Spatial.ai was watching brands spend significant money on personas — Busy Barbara, with her likes and dislikes — and walk away with a PowerPoint deck that couldn’t answer four basic questions: Where do I find more Busy Barbaras? How many of them are there? How much do they spend at my locations? How do I actually reach them through Facebook or direct mail? A persona that can’t answer those questions isn’t a segmentation. It’s a hypothesis. Spatial.ai’s three-step framework — append, analyze, activate — is designed to close that gap. Append your customer data or location set to the behavioral segments. Analyze which segments drive the most revenue and understand who they are. Activate by running targeted marketing or placing your next site directly adjacent to those clusters. The goal, as Lÿden describes it, is to get a brand from zero to a fully ranked, executable segmentation in under 30 minutes. The value of segmentation isn’t in the insight. It’s in what you do with it.

Episode Highlights

The Origin of Spatial.ai — Ethnography, Gentrification, and a Gun
~00:02:16
Lÿden Foust didn’t start in data. He started in ethnographic research — in-home studies, focus groups, living with communities over time to understand how they actually behave. A pivotal assignment for Zaxby’s Chicken in Nashville sent him to a neighborhood that looked low-income on census data, but had gentrified significantly by the time he arrived. The demographic data was years out of date. The people on the ground were not the customers he was sent to study. On the last day of the project, he had a gun pulled on him. That moment — disorienting and dangerous — crystallized a question he couldn’t let go: how do we understand communities in real time, in ways that go beyond demographics? That question became Spatial.ai. It’s a founding story worth knowing because it explains why the company’s approach is so different. This wasn’t built by someone who started in a spreadsheet. It was built by someone who had been in the field and understood why static data fails.

“If there’s anybody that should understand communities, it should be me, but I don’t have the data to do it — the census was taken six years ago and the data had changed.” — Lÿden Foust

The Harbinger Customer and the Fourth of July Furniture Story
~00:22:00
One of the most memorable moments in the episode is Lÿden’s description of the harbinger customer — a concept he credits to a major furniture retailer who discovered it in their own data. The harbinger is a small segment that doesn’t represent huge sales volume, but is culturally influential in ways that ripple across other segments. For this retailer, it works like this: if they can get the right colors and furniture styles into harbinger customers’ homes before the Fourth of July, the other segments follow. Because those people are hosting the cookouts. Their neighbors and friends sit on the furniture, ask where it came from, and buy it. It’s a segmentation insight that becomes a product placement strategy and a cultural influencer program all at once. The implication for any brand is significant: your most important customers may not be your highest-volume customers. They may be the ones whose choices other people copy.

“These segments are particularly influential, and so if they want to push a new product, they will get it to that segment in particular — and that’s a cultural thing.” — Lÿden Foust

The SeaQuest Date Night Discovery
~00:15:21
SeaQuest is a 14-location aquatic experience in malls — think petting otters and interactive aquariums. They believed their core customer was lower-income moms with kids, and their first-party data confirmed it. When they uploaded their customer data to Spatial.ai’s segmentation, the moms showed up as expected. But the second-largest segment was something nobody predicted: young professional singles. There was no obvious reason for it. But it was real and it was significant. SeaQuest ran a date night campaign — the SeaQuest Duo — targeted directly at that segment. It became their number one selling ticket package across all 14 locations almost immediately. And the discovery kept compounding: they also found dads bringing their daughters, often right after a Sephora visit in the same mall. None of this was visible in first-party data. All of it became visible when behavioral segmentation revealed who was actually showing up.

“You can’t see everything in your own data. The things that you assume about your customers probably are right, but those are the things everyone knows. You need to know things that your competitors don’t know.” — Lÿden Foust

Uniqlo in Texas and the $200K Fusion Family
~00:12:04
One of Lÿden’s most powerful examples involves a segment he calls Fusion Families — households earning over $200,000 a year who shop like they make far less. They over-index on Costco. They don’t buy expensive retail. But they spend heavily on two things: private school tuition and Sephora. The explanation, when you look at the behavioral and demographic data together, is that this segment over-indexes for first- and second-generation immigrants whose value set doesn’t map to typical American high-income consumerism. Brands that placed premium concepts near this population in Texas kept wondering why their demographic criteria were met but their locations weren’t performing. Uniqlo — a brand with Asian cultural roots — figured this out through e-commerce data, saw that this segment was buying even without nearby stores, and opened five Texas locations within five months. They captured four percent market share almost immediately. The lesson is blunt: knowing your customer’s income doesn’t tell you what they’ll buy.

“If you don’t know your customer — really know your customer beyond demographics — you’re gonna make mistakes. But if you do know your customer, you can make some huge headway with very minimal effort comparatively.” — Lÿden Foust

How InfutorData Powers Household-Level Segmentation
~00:32:21
Lÿden closes the episode by explaining the structural role InfutorData plays in making Spatial.ai’s segmentation work at a household and individual level. The segmentation methodology itself — K-means clustering — has existed since 1970. What’s new is the data feeding it. Real-time signals replace stale census data. And household-level classification replaces census block group averages, where everyone in the same geographic unit was treated identically. That approach ignores the reality that neighbors are often completely different customers. Lÿden uses himself as the example: his neighbor three miles away is an 85-year-old woman; his neighbor directly to the south is Amish. The census block group would have lumped them together. InfutorData’s individual household data allows Spatial.ai to differentiate one household from the next — which is what makes the segmentation actionable at the scale brands actually need.

Top Quotes

Lÿden Foust [~00:00:00]
“You can’t see everything in your own data. The things that you assume about your customers probably are right, but those are the things everyone knows. You need to know things that your competitors don’t know — and that’s where third-party data becomes important.”

Lÿden Foust [~00:01:43]
“We believe humans are incredible, and we’re living in the first era where we can mathematically understand people beyond their demographics so that we can actually action on that.”

Lÿden Foust [~00:09:29]
“How can we make segmentation the fastest path to ROI? If you understand your customers at a fundamental level, you’ll be able to make better decisions.”

Lÿden Foust [~00:10:00]
“Busy Barbara is not actionable. If you wanted to find more Busy Barbaras, where geographically would you find her? How many are there? How much does she spend? How do I actually reach her? You couldn’t do any of those things.”

Lÿden Foust [~00:14:47]
“If you don’t know your customer — really know your customer beyond demographics — you’re gonna make mistakes. But if you do know your customer, you can make some huge headway with some very minimal effort comparatively.”

Lÿden Foust [~00:20:00]
“Persona Live is a segmentation of people based on what you do, not based on your demographics. We wanna know what you do.”

Lÿden Foust [~00:25:23]
“Sentiment and topics are noise on social media. Hashtags are noise. People are not talking about your brand — they’re thinking about themselves and talking about themselves.”

Lÿden Foust [~00:30:31]
“People are a lot less fluid than what they might seem. You have kids, you move, you get married — those are the three things. But what your segment does might change, and what you do within those segments, that can flex.”

Tim Finnigan [~00:17:32]
“Without someone like you, and you’re just relying on your first-party data, you’ve got a name, an email, maybe a cell phone number. What you do is scientific, but it’s really like this tremendous value for brands.”