Most restaurant location advice is either real-estate folklore or a sales pitch for a six-figure consulting engagement. This is the middle path: the actual factors that predict whether a site works, in the order they matter, written for an independent operator who has to sign a lease with their own money.
The single strongest predictor of a restaurant's survival is how many people actually walk past the front door, at the exact corner, not somewhere "nearby." A block can change character completely in 150 metres — foot traffic drops off a cliff at the edge of a commercial strip.
Two things matter when you look at any foot-traffic figure:
Read the full breakdown in our foot-traffic guide.
The instinct is to find a gap in the market: an area with no other restaurants. The data says the opposite. In Toronto's food-and-beverage record, a new venue with no other food business within 400 m survives two years 68.9% of the time, against 79.5% for a venue with a real peer cluster nearby. The isolated corner is the worst-measured position, not a safe one.
Why: a strip of food businesses creates a destination. People decide to "go out for dinner on X street," then choose a specific spot when they arrive. The restaurant on an empty block has to generate every single visit on its own.
The risk to actually avoid is opening a clone — the same cuisine, same price point, same format as the place next door. Competitors in general are good; direct clones are a coin-flip.
Restaurants rarely fail because no one showed up. They fail because rent was 15% of revenue while food, labour and overhead were another 90%. The ratio that matters is total occupancy cost ÷ projected annual sales.
The trap is forecasting sales off the landlord's pitch deck. Work the math yourself — see the rent-to-sales guide for the full stress-test.
Before you sign, find out what happened at that address for the last five years. A corner that has burned through three restaurants in two years is not "available," it's telling you something structural — bad visibility, a hostile landlord, an access problem you haven't spotted yet. A site where one operator ran for a decade and retired is a completely different asset.
The survival baseline is worth knowing before you judge any individual site. In Toronto's licence register — 12,146 storefront food businesses licensed 2018–2023 — 94.3% survive their first year, 74% reach two years, and 41.8% reach five. The format matters too: sit-down concepts reach two years at 76.5%, take-out at 71.2%. So when a block shows three closures in two years, that's not normal attrition — that's a site problem. Normal attrition is the 26% that don't make two years citywide; a corner that's worse than that is the one to avoid.
Where a city publishes licence records, closure history is directly measurable. Where it doesn't, ask neighbours and the local business improvement association. Ten minutes of questions beats six months of a failing lease.
Count the restaurants already serving the neighbourhood and divide the local population by them. In Hoodly's Canadian data, the spread is wide: Toronto runs about 385 residents per restaurant, Montréal about 300, and Victoria about 161 — three very different markets, three different answers to "is there room for one more." The citywide average is almost meaningless; the neighbourhood number is what decides your lease, and it varies by an order of magnitude across a single city.
The useful comparison is your candidate block against the city baseline, not against some national average that includes suburbs and industrial parks.
A neighbourhood that is adding residents and workers is a rising tide; a static one makes every customer a zero-sum win against a neighbour. Look for approved and under-construction development within 500 m of the site — new residential units are future customers, new office space is future lunch traffic. A pipeline of zero is not a dealbreaker, but it means you are buying today's market, not tomorrow's.