Latest / Franchise Conversations with Fexingo: Buying, Running, and Scaling Franchise Businesses / How Franchisees Use Predictive Analytics for Site Selection
Transcript
- Lucas: So there's this multi-unit franchisee I've been following — operates about 30 quick-service locations across three states. He told me that five years ago, picking a new site meant driving around, counting cars, talking to the local planning department, maybe buying a traffic study. Now he uses a predictive model that crunches 40 variables and spits out a first-year revenue forecast within 5 percent accuracy. Luna: 5 percent? That's tighter than most corporate development teams hit. How many sites did he evaluate with that model last year? Lucas: Forty potential sites. He built a shortlist of eight, opened four, and the first three are already beating projections. The fourth opens next month. He said the model cut his site-search time from about six months to six weeks. Luna: That's a massive edge. And it's not just about speed — it's about avoiding the bad ones. A bad location in quick service can bleed $200,000 a year before you even realize it. Lucas: Exactly. A bad site costs you the build-out, the lease, the training time, the lost opportunity. So this is episode 86 of Franchise Conversations — I'm Lucas. Luna: And I'm Luna. Today we're talking about how franchisees are using predictive analytics for site selection — not as a replacement for boots on the ground, but as a way to focus their human judgment on the best candidates. Lucas: Let's start with the data layers. The model this franchisee uses pulls in foot traffic patterns from mobile location data — think Placer.ai or similar — plus competitor density within a one-mile radius, local income growth over the past three years, even historical weather patterns that affect seasonal traffic. Luna: Weather? That's interesting. So you're not just choosing a location, you're forecasting how many rainy Tuesdays might affect drive-through volume. Lucas: Right. If your concept is heavy on walk-in traffic and you're in a city with 60 inches of rain a year, the model adjusts the revenue forecast downward by a factor. That kind of granularity is what makes the difference between a model that's 50 percent accurate and one that's 95 percent. Luna: And the franchisee community is talking about this a lot. I was at a multi-unit franchisee conference in Dallas last month, and two of the breakout sessions were on data-driven site selection. The audience was full of operators with 10 to 50 units. Lucas: It makes sense. If you have 10 units, you're probably adding one or two a year. A bad site costs you a year of growth. These tools aren't cheap — the software subscriptions can run $10,000 to $50,000 a year — but for a franchisee adding two sites annually, the ROI is almost immediate. Luna: But what about the franchisee with three units? Can they compete? Lucas: That's the tension. Some of the major franchisors — think the big QSR brands — are now offering site-selection analytics as part of their franchisee support package. So you don't have to build your own model. They'll run the numbers on any site you're considering and give you a score. Luna: I've heard that from a few franchisees — they say the corporate score is helpful, but they still want to do their own validation. Because the corporate model might be optimized for system-wide growth, not for your specific unit economics. Lucas: And that's exactly why the best franchisees build their own models or layer their own data on top of the corporate tool. Let me give you a concrete example. One franchisee I spoke with — he runs 18 units — he started tracking his own internal data: average ticket size by location, labor cost as a percentage of sales, even the local high school football schedule. Luna: Football schedule? Tell me more. Lucas: He noticed that on Friday nights in the fall, his stores near high schools saw a 20 percent drop in dinner traffic. So he built that into his model. Now when he evaluates a potential site, the model asks: is there a high school within two miles? If yes, it factors in the Friday night dip and adjusts the annual revenue projection. Luna: That's the kind of niche insight you'd never get from a national data provider. It's local knowledge turned into an algorithm. Lucas: Exactly. And this is where the conversation gets really interesting — because the models are getting more sophisticated, but the best franchisees are using them to ask better questions, not to replace their own judgment. Luna: Before we go deeper, I want to mention something that keeps this show going. We don't run ads, and we don't have a paywall. A couple of dollars a month from listeners who find value in these episodes genuinely makes a difference. If these conversations have helped you think differently about your franchise strategy, you can support us at buy me a coffee dot com slash fexingo. It's that simple. Lucas: Yeah, and it's not about big donations — it's about the collective. Even a few bucks a month helps us keep researching cases like this one. So if you're inclined, thank you. Luna: Alright, back to site selection. So Lucas, what are the most important variables that a franchisee should be looking at if they're starting from scratch? Lucas: I'll give you the short list. First, population density with a growth trend — not just current numbers, but projected change over five years. Second, competitor density — you want to know how many similar concepts are within a mile, but also their sales per square foot if you can get it. Third, traffic patterns — both vehicle and foot, broken down by time of day and day of week. Luna: And how do you get that data without spending a fortune? Lucas: Start with free or low-cost sources. The Census Bureau's American Community Survey gives you demographic and income data. Your local department of transportation has traffic counts. Google Maps can give you competitor density. Then layer on a paid tool like CIVIQ Smartscapes for pedestrian traffic or Placer.ai for mobile location data. You can get a decent model for under $5,000 in your first year. Luna: And what about lease terms? That's not in the data, but it's half the equation. Lucas: The model can't negotiate for you. But it can tell you what rent you can afford. If the model says a site will generate $1.2 million in revenue, and your target rent is 8 percent of sales, you know you can't pay more than $96,000 a year. That's a concrete number you take into the negotiation. Luna: I like that. It turns the lease conversation from 'what's the market rate?' to 'here's what this site is worth to my business.' Lucas: Exactly. And that shift in leverage is why I think predictive analytics for site selection is going to become table stakes for serious franchisees within three years. The ones who are early adopters now are building a data moat that latecomers will struggle to cross. Luna: Let's talk about one more case. You mentioned the quick-service franchisee with 30 units. What's the most surprising insight his model revealed? Lucas: He told me that his model consistently ranks sites near community colleges higher than sites near four-year universities. He thought the four-year schools would be better — more students, higher spending — but the model showed that community college students visit more frequently because they're often working part-time and need quick, affordable meals between classes and jobs. The transaction size is smaller, but the frequency is higher. Luna: That's the kind of insight that changes your whole real estate strategy. Suddenly you're looking at strip malls near community colleges instead of the student union. Lucas: Right. And he's now opened three sites near community colleges, and all three are in his top quartile for revenue per square foot. That's the power of letting data challenge your assumptions. Luna: So what's the next frontier for franchisees in site selection? Are we going to see AI that recommends the exact location? Lucas: We're already seeing that. Some commercial real estate platforms are integrating predictive models that score every available lease space in a metro area and rank them for a specific franchise concept. The franchisor inputs their unit economics, and the system says: these 20 locations have the highest probability of success. Luna: That sounds amazing, but also a little creepy. Is it just for the big players? Lucas: For now, mostly. But the cost of these tools is dropping fast. The same technology that a national real estate fund uses is being repackaged for franchisees. I expect within two years, a franchisee with five units will be able to access a tool like that for a few hundred dollars a month. Luna: That's an exciting democratization of data. But the human element still matters — the franchisee who knows the local market, the relationships with landlords, the intuition about a neighborhood's direction. Lucas: Absolutely. The model is a co-pilot, not the pilot. The best franchisees I've seen use the data to shorten their list, then they walk each site, talk to neighboring businesses, get a feel for the community. The model tells you where to look; your gut tells you which one to pick. Luna: And that combination — data plus human judgment — is the formula we're seeing across the top-performing multi-unit operators. It's not either-or. Lucas: Exactly. And that's the takeaway for today: if you're a franchisee adding locations, don't rely on your gut alone, and don't rely on a black-box model alone. Build a process that uses data to inform your intuition. Start with the free data, add one paid layer, and track your own internal metrics. Over time, you'll build a model that knows your business better than any off-the-shelf tool. Luna: Good advice. And if you want to see an example of that process in action, we'll put a link in the show notes to a spreadsheet template one franchisee shared with us. Lucas: Next episode, we're talking about how franchisees are using revenue-based financing to fund growth without dilution. Until then, keep asking better questions.