How we count

The Index is meant to be quoted, so the way it's built is published in full — including the parts that are weaker than we'd like. If you find an error, we want to hear about it: hello@agentroster.ca.

From every licence in Canada to the number we publish

Each stage is a gate in the code that builds the Index. Two are counts; two are things nobody can count, and the diagram says so rather than guessing.

What's counted

Every agent listed on the public "find an agent" directory of one of the 33 brokerage rosters below (national networks and independents), read directly from the directory each network publishes. An agent is counted once, at the office their own network attributes them to. Duplicates across networks are removed. The most recent scan ran on 2026-11-02.

What's not counted

This is the most important limit of the Index. The 79,898 agents counted here are about 50.6% of the ~158,000 REALTORS® CREA reported in early 2026. Not counted:

• Independent brokerages whose roster we don't read yet (we read a couple of dozen independents).
• eXp Realty, Real Broker and HomeLife, and the parts of Keller Williams and Coldwell Banker we don't reach yet — their directories need work we haven't done yet, not a judgement about their size.
• Agents a brokerage doesn't publish (new hires, private roster entries, anyone who opted out of their own directory).

So "largest in Toronto" on these pages always means largest among the networks we track. We'd rather say that on every page than imply a completeness we don't have. As more networks are added, the rankings will change, and we'll say so when they do.

How offices are ranked

• Headcount — agents on the roster. That's the whole ranking. No sales volume, no transaction sides, no reviews.
• City attribution — an office is placed in the city most of its agents are attributed to. Agents who serve a city from an office elsewhere count toward that other office, which is why a busy local team can be missing from a city's table.
• Share of city — the office's agents as a percentage of all tracked agents in that city.
• Page thresholds — an office needs 10+ agents for its own page; a city needs 20+ agents and 3+ offices for a ranking table. Below that the numbers are too noisy to rank.
• Broker share — the percentage whose roster title indicates a broker rather than a salesperson. Titles are free text set by each brokerage, so treat small differences between offices as noise.

Known limits

• Directories lag reality. A brokerage that updates its site slowly will show a stale count. We count what's published, not what's true on the day.
• One network reports branches, not legal entities. Right at Home is a single brokerage with ten branches; we attribute its agents to their branch so city-level comparison works. Other networks are counted at the office they list.
• Some directories return a rotating subset, not a complete list. RE/MAX pages unevenly by location, and Royal LePage caps a city's listing at about a thousand agents — so in the largest cities each scan sees a different slice. Measured directly: between two scans 19 days apart, 555 individual Royal LePage agents disappeared from Toronto while the city's total grew from 835 to 882. Offices in those networks' biggest markets are therefore more likely to be undercounted than over, and we do not treat an agent's absence from one scan as evidence they left.
• Counts are the live roster, not everything we've ever seen. Our database keeps agents after they stop appearing, which is right for a contact list and wrong for a headcount — left unchecked it had inflated one network by 17% against another at 0%. Every number here is what that office had at the single most recent scan, so all five networks are measured at the same moment.
• No movement data yet. Growth, churn and agent movement need several scans before they mean anything. We're not publishing trend numbers until they do.