Nobody publishes “the average Canadian agent closes N deals a year.” What’s actually published is national scale — membership, sales volume — and the office-level ratio has to be built from a brokerage’s own numbers, the same way an internal benchmark works in any professional-services business.
Key takeaways
Ask what a “good” brokerage office looks like in transactions per agent and the honest answer is that no regulator or industry body in Canada publishes that number. What is published is scale, not productivity — and the two are easy to conflate.
CREA represents more than 155,000 real estate brokers, agents and salespeople working through 61 real estate boards and associations across Canada — a real, current figure, and useful context for the size of the profession. It says nothing about how many transaction sides an average member closes in a year, because CREA counts membership and reports national and local sales activity through its MLS® System statistics, not per-agent output. Individual real estate boards publish local sales and price data on the same basis: aggregate activity, not a per-registrant ratio. Treat CREA’s membership figure as the denominator you could theoretically build a national average from, and the numerator — total transaction sides in a given period — as something no single public source ties back to it cleanly enough to produce a defensible national average.
Two ratios cover most of what a brokerage actually wants to know, and both are computable entirely from records the office already keeps for commission reconciliation: transaction sides per agent over a trailing 12 months, and gross commission income per full-time-equivalent (agents plus salaried staff). Sides per agent flags a roster with too many registrants doing too little volume — a recruiting or attrition question. Gross commission per FTE flags whether overhead is scaling with headcount — an operating-cost question, and a different problem with a different fix.
A low sides-per-agent number by itself doesn’t say whether the problem is too many registrants, too little support, or a market slowdown affecting every office — it says where to look next. Compared against the same office’s own trailing-year numbers, a falling ratio with a stable roster points at market conditions or lead flow; a falling ratio with a growing roster points at recruiting outpacing production capacity, which is a different fix entirely. Neither reading is available from a single snapshot — the benchmark only means something tracked consistently, period over period, against the office’s own history.
A single sides-per-agent number quietly breaks down the moment an office has agent teams, where one registered lead agent’s licence covers the production of several unlicensed or licensed team members working under them. Counting a team as “one agent” understates true production capacity; counting each team member as a separate agent overstates it if they don’t carry an independent book of business. The fix is to track sides per production unit — an individual registrant or a team, whichever actually holds the client relationship — rather than per licensed head, and to note in the benchmark itself which counting method was used, so a comparison against a future period isn’t silently comparing two different definitions.
Splitting sides_per_agent into listing-side sides and buyer-side sides separately surfaces a different kind of gap than the blended number does. An office heavy on buyer-side production and light on listings is more exposed to inventory conditions it doesn’t control; an office with a strong listing pipeline has more control over its own forward volume. Two offices can post an identical blended sides-per-agent number with opposite risk profiles underneath it, which is the argument for keeping the split rather than only tracking the combined figure.
A 14-agent office closed 168 transaction sides and $2,940,000 in gross commission income over the trailing 12 months, with 2 salaried support staff:
The math
agents = 14fte = 14 agents + 2 staff = 16
sides_per_agent = 168 / 14 = 12.0
gci_per_fte = $2,940,000 / 16 = $183,750
Neither number means anything against a published national average, because there isn’t one to compare it to. What it means is a baseline: if next year the roster grows to 18 agents but sides stay at 168, sides-per-agent falls to 9.3 — the office added headcount without adding production, which the ratio surfaces immediately even though total transaction volume looks unchanged. That comparison, this office against itself, is the entire value of building the benchmark — not a comparison to a figure nobody publishes.
Run the same 168-side, $2,940,000 office forward a second way: gross commission income holds flat but agent count drops from 14 to 11 through attrition. Sides-per-agent rises to 15.3 and GCI-per-FTE (now 13 FTE) rises to $226,154 — a genuine productivity improvement on a smaller, more concentrated roster, and a very different management story from the headcount-growth scenario above even though both start from the same base year. The ratio doesn’t tell a brokerage which scenario is happening; it tells the office exactly where to look once it has. A roster that grows through hiring rather than attrition is worth tracing back to how those registrants were brought on — a productivity dip that coincides with a wave of new registrants is a ramp-time question, not necessarily a performance one.
No fetched CREA page states a reason. CREA’s own statistics portal publishes MLS® System sales activity and membership counts, both aggregate measures, rather than individual or office-level production data, which real estate boards generally treat as commercially sensitive to their own members.
No — sides count each side of a transaction (the listing side, the buying side, or both if the same brokerage represented both parties) rather than dollar volume. A brokerage tracking both should keep them as separate ratios, since a high-dollar-volume, low-side-count office and a high-side-count, lower-dollar office face different operational questions.
Only with caution. A new office has no trailing-year history of its own to benchmark against, and a young roster typically has a different sides-per-agent profile than an established one purely from ramp time — the ratio is most useful once an office has at least one full trailing year to compare against itself.
Quarterly, on a trailing-12-month basis, is enough to catch a roster or production trend without over-reacting to a single slow month — a monthly recalculation on a short window tends to produce noisy swings that don’t reflect a real shift in the office, particularly in a market with seasonal transaction patterns.
A 30-minute call is enough to tell you whether AI pays for itself here.
