For retailers, the brand stays fixed while the search focuses on finding space. Developers and landlords start with a parcel and look for a use and tenant that can carry the rent. Both searches rely on the same demographics and foot traffic. But retailers need a score for one address, while developers and landlords need a filter they can apply across many parcels.
Site selection involves two distinct assignments that move in opposite directions. An operator planning to open a format begins with the brand: its prototype footprint, the audience the concept serves, and the occupancy cost the unit economics must support. Those parameters are fixed, so the search moves across parcels until one meets them. An owner developing land or filling space takes the property as the starting point: a parcel with an established shape, access point, permitted use, and cost basis already paid or committed. Those physical conditions are fixed, and the search then considers uses and tenants until it finds one that can generate the earnings the asset requires.
Both groups use the same demographic profiles, drive-time bands, and foot-traffic data. Their deliverables differ because each is addressing a different set of unknowns. From the developers and landlords we speak with, the request we hear most often isn't an address score for a site they've already selected. They want a filter instead: a way to narrow a long list of parcels to the few that can support a viable use, then identify a short list of concepts that would plausibly sign for them.
The distinction matters because analysis designed for one side can be passed to the other and answer a question no one asked. A weighted score for one address is exactly right for an expansion committee approving one store. For an owner deciding which of forty parcels to put under contract, that score is the wrong tool.
Tenant-first and space-first are two different searches
Tenant-first and space-first indicate which side of the site decision remains fixed. Tenant-first holds the brand constant and considers different parcels, ranking addresses for one concept. Space-first holds the parcel constant and varies the use, ranking concepts for a single piece of ground.
A single distinction carries through every later decision: which variables count as disqualifiers, what belongs in the comparison set, how far the time horizon extends, and how the completed analysis must be presented for someone to act on it.
Shared terminology can conceal the misunderstanding. Landlords and retailers both refer to trade areas, demographics, and foot traffic. A retail void analysis shows the difference most clearly: each side runs one, but the owner receives a category call sheet while the operator gets a ranked market list. Their inputs are nearly identical. The results are not.
What a retailer holds constant
A retailer keeps its format requirements fixed: prototype square footage, parking count, venting and power requirements, and the occupancy cost its unit economics must support. These aren't preferences to balance against a strong demographic profile. They set the search boundary, so a parcel either falls within it or doesn't.
Within that boundary, the retailer's own portfolio is the relevant comparison set. A candidate address is useful to the extent that it resembles stores that already perform, so analog matching and cannibalization checks carry significant weight. Owning several good stores means the operator knows what one looks like. What remains to determine is whether this address will behave the same way.
That framing defines the planning horizon as well. Once options are included, an inline retail lease typically lasts about a decade, so one box requires the retailer to underwrite years of trade-area behavior. The 30-second screen most operators run before the numbers addresses the physical filters that can end the discussion early, while the underlying boundary question is how the trade area gets drawn in the first place.
What a developer or landlord holds constant
Instead, a developer or landlord treats the physical site as fixed. Its parcel size, frontage, curb cuts and access, utilities, stormwater detention, and permitted use under current zoning are established before any demographic data is pulled. The search first determines what can legally and physically be built there, then which operators would sign a lease for that space.
The current market makes that sequence more pronounced than usual. Marcus & Millichap put the US retail construction pipeline at 0.3% of existing inventory as of mid-2026. In the first half of the year, 16.4 million square feet was delivered, the second-lowest two-quarter total on record. The firm also noted that most active construction is for built-to-suit convenience stores and supermarkets, not speculative development (Marcus & Millichap, August 2026). Built-to-suit means the tenant comes before the parcel, and the construction schedule follows that order. The lease is signed first; then the building is constructed around it.
This process nearly reverses the retailer's. Physical and legal screening comes first: it's inexpensive, binary and fatal. If a parcel can't take a curb cut on the side from which traffic approaches, it's rejected regardless of the household income around it. Demand analysis begins only after those screens are clear, asking which category the trade area lacks rather than how a single brand would perform.
Tenant selection comes next. A pad site positioned along a parking field is not the same offering as inline space behind it. It attracts a different group of users and commands a different rent. In effect, the owner is comparing a catalog of concepts for one parcel. The useful output is a short list of brands whose published site criteria the parcel actually satisfies.
Why a per-site score answers only one of the two questions
A per-site score reduces multiple variables for a single address to one number. That suits a tenant-first search, but it offers limited value in a space-first search. A developer with two hundred candidate parcels doesn't need a grade for every parcel. Most should be screened out through tests that are binary, not weighted.
Weighted scoring and screening fail differently, and the distinction is substantive. In a weighted model, a hard constraint gets blended into the overall result. A parcel with no legal access can still produce an acceptable score if the surrounding demographics are strong enough. Screening has no arithmetic to offset that issue. Each test is either a pass or a fail, one at a time.
A retailer shouldn't treat a non-negotiable requirement as one more factor in a weighted score; that's an avoidable mistake. For a developer, it's nearly disqualifying, since more of the constraints are legal and physical than commercial. Zoning, easements, setbacks and access are conditions of the site, not assessments of market potential, and demand upside can't make a prohibited use permitted.
Rent sets the land price, which sets the tenant
Residual land value is the calculation that makes the developer's ordering non-negotiable. Achievable rent first becomes stabilized income. Dividing that income by the exit cap rate gives the value on completion. Subtract construction cost, soft costs, and the developer's required return from that figure, and the balance is what the land can be bought for.
Following the chain from left to right makes the sequence clear: first, determine the rent the use can support, which puts the tenant's identity ahead of the land price. The CCIM Institute frames the entire calculation around one question: what can a buyer pay for the land and still keep the project feasible (CCIM Institute, Summer 2022).
Tenant credit directly changes the exit assumption. In the Boulder Group's second-quarter 2026 net lease report, quick-service restaurant properties with a corporate entity as tenant had a national asking cap rate of 5.85%. The same category, when leased to a franchisee, was at 6.85% (The Boulder Group, Q2 2026). The one-point difference applies to the same building type and reflects the covenant supporting the rent. For a given income stream, that cap-rate change creates a large difference in value on completion, which also changes what the developer can pay for the land. The report also shows that long-term leases with investment-grade tenants account for less than 10% of total retail supply. That scarcity is why the committed credit tenant, rather than the parcel, is usually the limiting input in the chain.
That is also why financing depends on the tenant, not the market study. Construction lenders underwrite a signed lease, so a developer with no committed anchor use isn't yet shopping for parcels. The project feasibility work determines whether any use meets that threshold, while the letter of intent is where the rent assumption becomes established.
Tenant mix is the landlord's version of cannibalization
A landlord faces a tenant-mix constraint that a retailer does not. Retailers protect sales across their own stores. Landlords must underwrite how a dozen leases interact within one asset over a hold measured in decades. Place too many leases in one category, and a complementary roster becomes internal competition for the same parking field.
For landlords, this is a named discipline. The ICSC guide to improving tenant mix lays out the work in order: assess consumer demand, find opportunity sites, determine which retail categories the trade area lacks, and recruit against that list only afterward (ICSC). The roster functions as a system, with second-order effects built into the leases. Co-tenancy provisions make one tenant's rent dependent on another tenant continuing to operate, so one vacancy can produce rent relief across several leases at the same time. The anchor that draws the trade area is the tenant against which the rest of the roster is priced, making anchor durability a leasing input rather than a market observation.
Lease duration is another source of divergence. An inline lease typically covers roughly a decade, while a ground lease or merchant-build hold can run longer or shorter for reasons unrelated to the tenant's business plan. If an owner expects to sell at stabilization, the priority is a rent roll that a buyer will underwrite. A long-term owner instead wants a tenant mix that remains workable after two lease cycles turn over. Given the same data and vacancy, those owners will select different tenants.
Running both searches on one dataset
GrowthFactor uses the same layers for both searches; they aren't separate products. A retailer's site score is built from demographics, foot traffic, vehicle counts, zoning and business points, and developers use those same layers to filter parcels. Because zoning appears on the map, the legal screen can happen before anyone commissions a market study.
The key distinction is where the criteria are applied. With Custom Evaluators, the requester defines the test. An operator can encode its prototype requirements once and score every candidate address against them. An owner instead defines what a specific asset can physically and legally support, then runs concepts against those limits. The variables and weights remain visible in both cases. That matters even more on the supply side, where the analysis is often read by a capital partner rather than a committee.
Because brokers sit between the two searches and conduct both in the same week, the practical choice is one workspace for each client rather than one for each brand. Deal records keep lease terms, the site score and pipeline stage together, so the same file provides the answer to "what fits here" and the answer to "does this work for us". The platform overview handles the layers, while deal pipeline tracks what happens after a parcel clears.
Frequently Asked Questions about how developers and retailers approach site selection
How do developers approach site selection differently than retailers?
Retailers hold the brand constant and search for space, asking whether a given address supports their format and volume. Developers and landlords hold the parcel constant and search for a use and a tenant, asking which concepts can legally occupy the ground and pay the rent the asset needs. Both use the same demographics and foot traffic; the deliverable each one needs is different.
What is residual land value in retail development?
Residual land value is what a developer can pay for a parcel after everything else in the deal is accounted for. Achievable rent produces stabilized income, income divided by an exit cap rate gives a value on completion, and construction cost, soft costs and the required developer return are subtracted from that value. What remains is the land budget, which is why the tenant assumption sets the land price rather than the other way around.
Why do developers ask which parcels fit a brand instead of scoring one site?
Because the developer's unknown is the use, not the address. A score grades one parcel the owner has already chosen, while a developer is usually holding dozens of candidates and needs most of them eliminated on binary tests such as permitted use, frontage and access. Screening removes; scoring ranks. The supply side needs the removing done first.
What does a landlord look at that a retailer does not?
Tenant mix across the whole asset, and the lease interactions that come with it. A landlord underwrites how a roster of tenants performs together over a hold measured in decades, including co-tenancy provisions that tie one tenant's rent to another staying open. A retailer is underwriting one box for one lease term and protecting its own sales from its own nearby stores.
How does GrowthFactor support a developer or landlord rather than a retailer?
The same data layers serve both, and the criteria are what change. Custom Evaluators let an owner write what a specific parcel can support and score concepts against it, rather than scoring addresses against one brand's prototype. Zoning on the map handles the legal screen early, and deal records keep the lease terms, the score and the pipeline stage on one file per asset.