Latest / Franchise Conversations with Fexingo: Buying, Running, and Scaling Franchise Businesses / How Franchise Brands Are Using AI to Choose Locations
Transcript
- Lucas: There's a question that every franchisee and every franchisor has to answer at some point: where do I put the next location? Luna: And historically the answer has been: drive up and down the main strip, count cars, maybe look at a population density map from the local chamber of commerce. Lucas: Right. Gut feel plus a few data points. But over the last three or four years, a growing number of franchise brands have been quietly replacing that approach with something far more quantitative. They are using ai powered location intelligence platforms to predict revenue for a potential site within about 10 percent accuracy before they even sign a lease. Luna: Ten percent is tight. That is a huge improvement over the old method where you basically guessed and hoped. Lucas: Exactly. And I want to ground this in a specific example. There is a mid-size quick-service brand — about 400 units, mostly in the Southeast United States — that started using a platform called Placer.ai about two years ago. Before that, their new-store failure rate was around 30 percent within the first three years. After adopting the ai driven site selection process, that failure rate dropped to under 8 percent. Luna: That is a massive swing. And it is not just about avoiding bad sites — it is about finding the good ones that you might have overlooked. Lucas: Right. The model ingests something like 50 different variables. Things like daytime population versus residential population, drive-time polygons, competitor locations and their sales density, even satellite imagery to assess the quality of adjacent retail. It is essentially a custom revenue forecast for every parcel of land in a given metro area. Luna: So it is not just a heat map of where people live. It is modeling the actual behavior patterns of customers. Lucas: Yes. One of the more interesting inputs is mobile location data aggregated from apps. That tells the model where people actually go during the day, not just where they live. For a franchise brand that relies on lunch traffic, for example, a site near a business park with 10,000 daytime employees might be far more valuable than a site near a residential subdivision with 20,000 residents. Luna: And that kind of granularity is something a human analyst could never replicate manually across an entire metro area. Lucas: No. And this is not just for the big national chains. I have seen multi-unit franchisees with 15 or 20 units using tools like Spatial.ai or even a simpler option like CELUS. The cost has come down a lot. You can get a pretty robust site scoring report for a few hundred dollars per location. Luna: I want to push back slightly on the notion that this is a silver bullet. The models are only as good as the data they are trained on. If a franchise brand has only ever opened in suburban strip malls, the AI might struggle to evaluate an urban storefront opportunity because it lacks comparable data. Lucas: That is a really important caveat. Data bias is a real risk. If the training set is all drive-to locations with ample parking, the model might systematically undervalue walk-up urban sites. Some brands have actually made that mistake — they relied on the AI too heavily and ended up passing on what turned out to be great urban locations. Luna: And there is also the risk of over-optimizing for short-term sales. The AI might rank a site highly because it predicts strong first-year revenue, but that same site could have high rent escalation or a demographic that is projected to shift over five years. Lucas: That is where human judgment still matters. The best practice I have seen among sophisticated franchise operators is using the AI as a screening tool — it eliminates the obviously bad sites and surfaces the top candidates. Then a human with local market knowledge visits those top sites and makes the final call. The model is not the decision-maker; it is a very good assistant. Luna: So it raises the floor but does not necessarily raise the ceiling? Lucas: That is a fair way to put it. The worst decisions get eliminated, but the best decisions still require a human who understands the nuances of a specific trade area. Luna: There is another dimension here that I think is worth exploring. Some of these platforms are now incorporating competitive intelligence. They can track foot traffic to competitor locations and estimate their sales volume, which helps a franchisee decide if there is enough demand to support another player. Lucas: Right. And that is a game changer for categories that are already saturated. Take the donut segment, for example. In many mid-sized cities, there might already be three or four strong brands. An AI model can tell you whether the market is truly saturated or whether there is an underserved pocket that a new entrant could capture. Luna: And it can also help with timing. If a new housing development is planned but not yet built, the model can project what the trade area will look like in two or three years, not just today. Lucas: Exactly. That forward-looking capability is probably the single biggest upgrade over old-school site selection. Traditional methods look at the existing population and traffic. AI can layer in projected growth from building permits, school enrollment trends, and even changes in commuting patterns. Luna: So if these conversations have moved your work forward in some small way — maybe you are a franchisee thinking about your next location, or a franchisor evaluating your expansion strategy — I will say what you probably already sense. Podcasts like this one run on listener support, not on ad revenue or sponsorship deals. Lucas: That is true. A small group of listeners already chip in monthly through buy me a coffee dot com slash fexingo, and that is literally what funds us producing these episodes. No ads, no corporate overlords — just people who find value in the conversation. Luna: Yeah, and we keep it that way. If that resonates, you know where to find us. If not, no worries — we are glad you are here either way. Lucas: So back to the AI tools — one thing I find fascinating is how some franchisors are now using these platforms to optimize their entire real estate portfolio, not just new locations. Luna: Meaning they go back and evaluate existing stores to see if the trade area has changed enough to warrant a relocation? Lucas: Exactly. A brand might discover that a store that was a great site in 2015 is now in a trade area that has shifted demographically. The model can identify underperforming stores that are actually in strong trade areas — meaning the problem is operations, not location — versus stores in declining trade areas where closure or relocation is the better move. Luna: That is a much more surgical approach than the typical 'close the bottom 10 percent of stores every year' strategy. Lucas: Right. It separates location risk from execution risk. And for franchisees who are deciding whether to renew a lease, that insight is extremely valuable. A multi-unit franchisee I spoke with recently told me that using a location intelligence platform saved him from renewing a lease on a store that was projected to see a 15 percent drop in daytime population over the next three years due to a major employer relocating. Luna: That is the kind of thing you would never catch just by looking at your own sales numbers. Lucas: No. And I think the adoption curve is accelerating. A survey from a franchise consulting firm earlier this year found that roughly 35 percent of franchise brands with more than 100 units now use some form of ai driven site selection. That is up from maybe 10 percent just three years ago. Luna: So we are past the early adopter phase and into early majority. Lucas: I think so. And the interesting question is what happens next. The platforms are getting more sophisticated — some are now incorporating natural language processing to analyze local news and social media sentiment about a trade area. If a neighborhood is getting a reputation as unsafe, that might show up in the sentiment data before it shows up in crime statistics. Luna: That is a bit dystopian, but also probably accurate. Lucas: It is. And it raises the same questions we saw with credit scoring and hiring algorithms. If the model is trained on historical data that reflects systemic biases — like underinvestment in certain neighborhoods — then the AI might systematically undervalue sites in those areas. That is a real concern, especially for franchise brands that want to expand into underserved communities. Luna: So the tool is powerful, but it needs oversight. And probably a diverse team of humans looking at the outputs. Lucas: Exactly. I think the brands that do this best are the ones that treat the AI as one input among many, not as an oracle. They use it to ask better questions, not to get final answers. Luna: So if you are a franchisee with two or three units and you are thinking about a fourth, what is the practical takeaway? Should you go buy an AI platform? Lucas: I would not recommend buying a full platform for a small operator. The licenses can run into the thousands per year. But I would recommend using the free or low-cost tiers that some of these companies offer — Placer.ai has a free version with limited data, and there are also consulting firms that will run a site report for a few hundred bucks. That is a small price to pay compared to the cost of making a bad location decision. Luna: One more thing — I read recently that some franchisors are starting to require franchisees to use an approved location intelligence tool as part of their development agreement. That seems like a big shift. Lucas: It is. And it makes sense from the franchisor's perspective. They want to protect the brand's real estate footprint and avoid the reputational damage of failed locations. But it does centralize more control, which some franchisees might resist. I expect we will see some tension there as the practice spreads. Luna: So the AI is not just picking sites — it is reshaping the power dynamic between franchisor and franchisee. Lucas: It is. And that is a conversation worth coming back to in a future episode.