AI Is Turning Retail Leasing Into an Underwriting Business

Retail real estate has never lacked data. It has lacked a reliable way to connect that data to the health of an individual store.

A rent roll shows what a tenant is obligated to pay. Demographics describe the households surrounding a property. Traffic counts estimate the number of cars passing a site. None of those measures, on its own, answers the questions that increasingly matter to shopping center owners and brokers: Is this store productive? Does its customer actually shop this trade area? How does it compare with other locations in the chain? And, if the tenant leaves, who is most likely to replace it successfully?

That gap is narrowing. Retail landlords now have access to combinations of mobile location data, credit and debit card transactions, point-of-sale feeds, tenant credit information, and artificial intelligence. The important shift is not simply that more data exists. It is that information, once scattered across spreadsheets, reports and separate platforms, can now increasingly be analyzed together quickly enough to influence a lease, acquisition or merchandising decision.

Placer.ai helped push location intelligence into the mainstream by allowing owners and brokers to analyze visits, trade areas and cross-shopping. CenterCheck approaches the market from a different angle, using anonymized card transactions to estimate store-level sales. Guesst connects directly with tenant point-of-sale systems to automate sales reporting, while RetailStat combines retailer financial health, credit risk, store locations and market activity. SiteZeus applies AI and predictive modeling to site selection and sales forecasting.

Individually, each answers a different question. Together, they are beginning to resemble a retail underwriting stack.

For a neighborhood or grocery-anchored shopping center, that changes how a rent roll can be evaluated. An owner can look beyond the presence of a national anchor and ask how that particular store performs, where its customers come from, which neighboring tenants share those customers, and whether a retailer’s broader financial position supports the durability of its lease. 

A strong corporate name does not necessarily mean every location is strong. Conversely, a productive store may be more valuable to a center than its corporate credit alone suggests.

The implications for leasing may be even greater. Brokers have traditionally marketed vacancies with square footage, asking rent, traffic counts, demographics and a site plan. Those remain important, but they can now be supplemented with a tenant-specific argument. A leasing team can identify the consumers already visiting a center, examine where else they shop, and target retailers whose customer base overlaps with the property.

That is a fundamentally different sales process. Instead of telling a retailer that a space is available, the broker can make the case for why that retailer should perform there.

NewMark Merrill CEO Sandy Sigal described that process to Commercial Observer earlier this year. His team used AI to help rank prospective tenants and construct a more detailed leasing pitch for a retailer that had previously rejected a location. After the leasing team checked and refined the AI-generated analysis, the retailer ultimately agreed to the deal. The technology did not replace the broker’s judgment. Rather, it gave the team another way to test and communicate its thesis.

Large landlords are building similar capabilities internally. Brixmor has described combining proprietary leasing information with demographic and geospatial data to analyze co-tenancy and help its teams identify better-fit retailer opportunities. That points to where AI may matter most: not as another stand-alone dashboard, but as an interface across a landlord’s own operating data and outside market intelligence.

For owners and investors, this also changes acquisition underwriting. Losing an anchor or junior anchor has traditionally required an analyst to make assumptions about downtime, replacement rent and leasing costs. Better store-level and consumer data can add another layer: how much traffic the tenant generates, which other tenants depend on that traffic, how the store performs relative to its chain, and which replacement concepts fit the existing customer base.

None of these datasets should be treated as definitive. Mobile data estimates visits. Card data represents observed transactions and requires extrapolation. Point-of-sale reporting depends on tenant participation. AI can accelerate analysis, but it can also make weak assumptions sound more convincing. Human judgment remains essential.

The advantage, therefore, will not belong to the retail owner with the most data. It will belong to the owner who can connect sales, traffic, credit, customer behavior and internal property information to better decisions.

Retail has always been a merchandising business disguised as real estate. AI and better data are making that increasingly measurable.

Ashkán Zandieh is the founder and managing director at the Center for Real Estate Technology & Innovation (CRETI).