The 30-second filters operators run before the numbers

Most candidate sites never reach a model. An experienced operator kills them with three questions, each answered in well under a minute: who else is already in the center, whether the physical box can hold the format, and whether the asking rent clears the occupancy-cost ratio the business actually runs at. Every one is pass or fail. None of them requires data you have to buy.

That screen is running whether or not anyone's written it down. The problem with leaving it undocumented isn't that it's wrong. It's usually right. The problem is nobody else on the team can run it, defend it, or check it.

The screen exists because good space moves faster than analysis

A full site evaluation takes hours at minimum, often weeks. The supply of space doesn't wait around that long. In the second quarter of 2026, Cushman & Wakefield put national retail vacancy at 6.0%, well under the 7.4% historical average, with only 2.3 million square feet of new retail delivered nationally in the quarter and an active development pipeline under 0.3% of existing inventory (U.S. Retail MarketBeat, Q2 2026, published July 2026). Colliers read the same quarter at 4.4% vacancy and noted that small-format space in particular stays constrained (U.S. Retail Q2 2026, July 2026).

Two firms, slightly different universes, same story underneath. There isn't much space out there, and there isn't much coming. A team that gives every listing the full treatment will still be modeling the third site on its list once the first two are gone.

So the screen isn't a shortcut around rigor. It's what makes rigor affordable. The filters don't tell you a site is good. They tell you which sites are still worth an opinion.

Filter one: read the tenant roster before the demographics

The fastest read on any site is the list of businesses already in it. The anchor and the tenants around it tell you who drives to that corner on purpose and what they came for. That's a different, more useful question than who lives within three miles of it.

A census tract describes residents. A tenant roster describes behavior. If the center is built around a customer who has no reason to buy what you sell, no demographic profile is going to save it. The people in that profile are shopping somewhere else.

The roster also carries risk demographics can't show you. When an anchor leaves, the tenants who stay feel it, and it shows up in the rent: a study of small and moderately sized centers in Florida and Georgia found anchor tenant loss associated with roughly a 25% decline in rents for the remaining tenants (Gatzlaff, Sirmans and Diskin, Journal of Real Estate Research 9(1), 1994). That study's old and regionally narrow, so treat the number as an order of magnitude, not a national benchmark. The mechanism it describes is why co-tenancy clauses exist at all, and why an inline space next to a wobbly anchor is a different bet from the same space next to a healthy one.

Two things to check in the thirty seconds you've got. First, does the anchor draw your customer or somebody else's. Second, is the roster stable or churning, which you can usually tell by how many spaces are dark and how recently the signage changed. For the longer version of that read, anchor tenant quality and tenant mix is worth understanding properly, and co-tenancy strategy covers how to size up the neighbors before you get anywhere near the lease.

Filter two: the physical box is pass or fail, never a score

The second filter is a list of hard requirements. The space either meets them or it doesn't. Square footage. Ceiling height. Drive-thru feasibility and stacking depth. Parking count. Ingress and egress. Electrical service and venting. Permitted use under current zoning.

Franchise brands publish these as thresholds, and that's the clearest evidence they belong in a screen, not a scoring model. Little Caesars asks for 1,200 to 1,600 square feet, a minimum of five parking spaces, 400-amp three-phase electrical service, and at least 20 feet of frontage (Little Caesars real estate requirements). Dairy Queen's 46-seat prototype runs 2,207 square feet and calls for 33 parking spaces plus five to six cars of drive-thru stacking between the window and the menu board (Dairy Queen building types). ALDI wants roughly 22,000 square feet with a minimum of 95 dedicated parking spaces and at least 103 feet of frontage (ALDI property requirements).

Read those as what they are: floors, not preferences. A space at 900 square feet doesn't score lower for Little Caesars. It's out.

This is the filter teams get wrong most, and it goes wrong the same way every time. A hard requirement gets folded into a weighted site score as one input among twenty, and a strong demographic profile outvotes it. The site clears the model with a good number, and somebody finds out on the walkthrough that there's no venting path and no realistic way to add one. A weighted average will happily average away a wall you can't move.

Zoning belongs in this filter too, and it's the cheapest one to check. A permitted-use problem ends the conversation in seconds, before anyone even drives out.

Filter three: run the rent backwards through your occupancy ratio

The third filter converts the asking rate into the sales number the site would have to produce. The math's short enough to do standing in a parking lot.

Take an inline space of 1,500 square feet asking $38 per square foot per year, with triple-net charges quoted at another $12. Add those together and the all-in rate is $50 per square foot. Multiply by the 1,500 square feet and the space runs you $75,000 a year, or about $6,250 a month. Now divide by the occupancy-cost ratio your format actually survives on. At 5%, that space needs roughly $1.5 million in annual sales. At 8%, about $940,000.

Which of those numbers you use matters more than the rate itself. For a reference point, the National Restaurant Association's 2025 Restaurant Operations Data Abstract, drawing on fiscal 2024 data from a survey of more than 900 operators, put median occupancy costs at 5.2% of sales for limited-service restaurants and 5.7% for full-service (National Restaurant Association, September 2025). Those are medians across a wide field. Fine as a sanity check, useless as your number. Your ratio comes off your own P&L, from the stores you already run, and it has to get measured against that same cost basis you just added up.

Then compare that required volume against a store you own. Not a forecast, not your best location. The median of your comparable stores. If the site needs a volume none of them has ever hit, that's a finding, and you got there before anyone built a model. The rent per square foot math is worth getting exact once, because every site you look at afterward runs through it.

What the 30-second screen cannot see

The filters are good at rejection and useless at selection. Everything that decides whether a surviving site is actually worth signing lives outside them.

They can't see how much of the new store's revenue would come out of a location you already own. That question needs your own sales data and a real trade area, and it regularly changes the answer on sites that pass all three filters clean. Cannibalization analysis is the work a fast screen exists to make room for, not a step it replaces.

They can't see the true trade area either. Drive time, physical barriers, and where customers already come from produce a shape that rarely matches the ring somebody drew on a map. And they can't forecast revenue, which is the number the committee's going to ask about eventually.

That division of labor is the whole point. Cheap filters knock out the sites that were never going to work. The expensive analysis then runs on a short list instead of a long one. Inside the GrowthFactor workflow, that split is explicit: the screen narrows the pipeline, the scoring and trade area work run on what's left, and every input that moved a score stays visible when somebody asks why a site made the cut.

Write the filters down, then audit your reject pile

Here's the part most teams skip. The screen's already running. It lives in one or two people's heads, it's never been written down, and it drifts.

Two things fix that, and neither one takes long.

Write the filters down as explicit thresholds. Not "good co-tenancy" but the specific anchor categories that qualify. Not "adequate parking" but the actual count. Not "rent has to work" but the occupancy ratio and the comparable store you're measuring against. A filter you can state in one sentence is a filter a new hire can run, and one you can hand a broker so they stop sending you sites that were never going to clear.

Then audit the reject pile. Pull the last fifty sites you passed on quickly and write the reason next to each one. If those reasons collapse into a short, repeated list, you've got real filters, so document them. If they scatter, or a lot of them read as "did not feel right," you've got preferences instead of filters. And the ones you can't articulate are the ones nobody can check.

That audit is the same discipline as back-testing a site score or knowing which judgments your instinct has actually been trained on. Fast judgment's worth keeping. It just has to be the kind you can show somebody.

The screen doesn't replace the analysis. It decides what the analysis gets spent on, and on a team that reviews far more sites than it ever opens, that decision moves more than the analysis itself does. Once the filters are written down, the full evaluation runs on a list that's already earned it.

Frequently Asked Questions about fast site screening