Real estate brokerage call routing: 41% of callers are other realtors

In short: Real estate brokerage call routing is most of the job on a brokerage phone line, and a complete log of 1,190 answered calls to one Toronto brokerage line over 102 days shows it plainly. 41.0% of the callers were other realtors, and 92.8% of them left a callback number. 41.7% of all callers asked for a specific person by name. Members of the public with a question and nobody in particular in mind were 29.2% of the line, and junk was 12.2%. Route a cooperating agent's call to the listing agent with the listing and the ask attached, send a named request straight to that person, and judge any answering service or AI receptionist on where the message goes once the call ends.

Key takeaways

• Four callers in ten were cooperating agents, 488 of 1,190, and they left a callback number 92.8% of the time. Capture is already solved for them. What breaks is getting their message to the listing agent with the listing and the ask attached.
• 496 callers, 41.7%, asked for a named person. That request belongs in front of that person, with the number attached, and every extra hop adds time to the callback.
• 145 calls, 12.2%, were wrong numbers, robocalls or silence. On a contract that bills every answered call, that is 145 calls a quarter you pay for and get nothing from.
• The three busiest weeks of the summer were the last three full weeks of August, at 98, 107 and 115 calls against an average of about 79. Do not change how your phones are answered in the run into the 1st.
• About one percent of calls on this line carried an urgent flag, so the emergency pitch matters far less here than on a property management line.

Who actually calls a brokerage line

Search for an answering service or an AI receptionist for a brokerage and most of what comes back leads with the same list: capture the lead, answer common questions, book the appointment. That pitch pictures a buyer with a question on the other end. We run the voice agent on a Toronto brokerage line and log every call, so we can check who actually rings.

The log splits into four groups. Each call sits in exactly one, assigned in this order: junk first, then realtors, then members of the public who asked for someone by name, then everyone else.

| Caller | Calls | Share | Left a callback number | Share after hours |

|---|---|---|---|---|

| Another realtor | 488 | 41.0% | 92.8% | 26.4% |

| Member of the public asking for a named person | 210 | 17.6% | 72.9% | 27.6% |

| Member of the public with a question, no named person | 347 | 29.2% | 71.8% | 38.0% |

| Junk: wrong number, robocall or silence | 145 | 12.2% | 0.7% | 36.6% |

The window is 1 June to 10 September 2026, and after hours means outside Monday to Friday, 9am to 6pm. The realtor flag, the named-person flag and the callback field are logged on every call; the junk flag is a judgment made call by call.

The row the answering-service pitch is built around is the third one, 347 calls and under a third of the line. The 698 callers above it were professionals or people who already knew who they wanted, and what they needed was a clean message in the right hands.

That third row matters most in the evening. Members of the public with a question were the group most likely to call after hours, 38.0% of them against 26.4% of realtors, so if a vendor's question answering is good, that is where it earns its fee. The after-hours picture for the line as a whole, where 31.3% of calls came outside the office week, is in what 3,332 answered calls say about after hours.

Cooperating agents already leave a number

A realtor calling another brokerage is a professional caller. They know what they want, they are used to leaving messages, and on this line they left a number 92.8% of the time. For them the failure to fix is delivery, and capture takes care of itself.

What they called about, as a judgment made on each call:

| What the realtor wanted | Calls |

|---|---|

| A callback from a named agent | 178 |

| A showing | 103 |

| Whether the listing is still available | 73 |

| Something else | 56 |

| An offer or an application | 40 |

| Maintenance | 25 |

| Lease administration | 11 |

| Payment | 2 |

Separately, 285 of the 488, or 58%, named the person they wanted. That logged flag counts every call with a name in it, whatever the reason, which is why it runs higher than 178.

Every row is a message that has to land somewhere. A showing request needs a time and an address, and an availability check needs the listing identified. The 40 calls about an offer or an application are where a slow handoff has a real price, because an agent with a client ready to write does not wait while a general inbox gets cleared.

The standard for a cooperating-agent call is a message that reaches the listing agent quickly and says which listing, what the caller wants, and the time or question they can act on without calling back. "An agent called, please call back" costs everyone a second call.

A name tells you where the message goes

496 of 1,190 callers asked for a named person, and the share barely moved with the clock: 42.9% during business hours, 39.0% after. At 7pm on weekdays it was 53.2%, 25 of 47 calls.

Callers who named someone left a number 86.3% of the time (428 of 496), against 61.7% (428 of 694) for callers who did not. We first read that as the name doing the work, as if a caller with a person in mind were more committed. Most of the gap is who is asking. 285 of the 496 named requests came from realtors, who leave a number almost every time regardless, and among members of the public the difference nearly disappears: 72.9% with a name, 71.8% without.

So a name told us less about intent than we assumed, and a great deal about where the message has to go. The caller has told you whose phone the message belongs on, and every hop through a front desk or a shared inbox that has to work that out again adds time to a callback the caller thinks is already on its way. When that agent is away, the message should go to whoever covers for them, by a rule written down in advance.

An eighth of the line is junk

145 calls over the summer were wrong numbers, robocalls or silence, a judgment made on each call rather than a logged field. 0.7% of them left a callback number.

36.6% of the junk arrived after hours, a little above the line as a whole, so most of it rang during the working day. It clusters by hour: on weekdays the worst was 6pm, when 14 of 59 calls (23.7%) were junk, followed by noon at 16.2% and 2pm at 14.5%.

If you pay per call or per minute, those calls are on your invoice unless the contract says otherwise. We have not surveyed how each vendor treats them, and a sales call is the wrong place to find out. Get it in writing.

The busiest weeks are the run into the 1st

The summer averaged about 79 calls a week. The three busiest were the last three full weeks of August: 98 calls in the week of August 10, 107 in the week of August 17 and 115 in the week of August 24, the run into the September 1 move-in date. The mix shifted toward routing as well: across those three weeks realtors were 43.4% of callers against 40.1% for the rest of the summer, and named-person requests were 44.7% against 40.6%.

Weekdays were flat, from 12.3 calls a day on Thursday to 14.0 on Friday, with Monday at 13.8. Saturday ran 9.1 and Sunday 5.8. Unlike the property management line at the same operation, this line has no Monday spike and a real Saturday.

We would not change how a brokerage's phones are answered in the last two weeks of a month, and in Toronto we would not do it in August at all. Switch early in the month, when this line ran quieter (the first five days of a month averaged 9.8 calls a day, against 13.1 from the 18th to month end), and give the new setup a couple of weeks of ordinary traffic before the rush.

Four questions for any vendor

They apply equally to a live answering service and an AI receptionist.

How do you handle a cooperating agent? Ask to see the exact message a listing agent receives after another realtor calls about a showing. It should name the listing, say what the caller wants, and carry a time or a question. If it is a transcript in an email, someone in your office is still routing by hand.

Where does a named-agent request go, and who covers when that agent is out? It should land on that agent's phone or CRM record, with a written fallback. If your agents work out of Follow Up Boss, that is where the message belongs, and how that connection is built is on our Follow Up Boss automation page.

How is junk billed? Ask how a wrong number, a robocall and a silent line are each billed, and ask for the answer in the contract. On this line it was 145 calls in one summer.

What happens in the last two weeks of the month? A normal week of about 79 calls became 115 in the week of August 24. Ask what changes when that happens, and whether any setup or migration work would land inside it.

Emergencies come after all four. About one percent of calls here carried an urgent flag. They still need a written path to a person. On a property management line that path decides the purchase; on this one it is the fifth question.

Method, and what this is not

This is one operation in one market over one summer. The data is every answered and timestamped call to a single brokerage line at a Toronto property management and brokerage operation, 1 June to 10 September 2026, 102 days and 1,190 calls. Every call was answered by a voice agent we built and run. Its job on this line is narrow: it answers, logs the call, captures a callback number, and posts a structured message to the team that names who the caller asked for. How the agents are built is on AI voice agents, and the brokerage version of the intake work is on AI lead intake for brokerages.

Records were deduplicated on timestamp and reduced to derived fields, with no caller name, number, email or address kept.

Some fields are logged and some are judged. The timestamp, the realtor flag, the named-person flag and the callback field are logged on every call. The junk flag, the reason category and the urgent flag are judgments made per call, and they are the softest numbers here. The four caller groups are mutually exclusive, assigned junk first, and no call was flagged as both junk and realtor.

It is not a sample of Canadian brokerages. A brokerage that sells more than it leases, or a line measured in February, will look different. Every figure on our proof page carries its window and method the same way, and the wider picture of AI in a brokerage is on AI for real estate brokerages.

Frequently asked questions

What is real estate brokerage call routing?

It is getting each call on a brokerage line to the person who should act on it: a cooperating agent's showing request to the listing agent, a named request to that agent, a general question to whoever handles enquiries, junk nowhere. On the Toronto line we measured, 41.0% of callers were other realtors and 41.7% asked for someone by name. For most calls the answering was the easy part, and the value was in a message with a callback number reaching the right person.

How many calls to a real estate brokerage come from other realtors?

On one Toronto brokerage line over 102 days, 488 of 1,190 answered calls, 41.0%, came from other realtors. They left a callback number 92.8% of the time, and 58% of them asked for a specific person. In the last three full weeks of August their share rose to 43.4%. Your own share will depend on how often other brokerages show your listings.

Should a brokerage use an answering service or an AI receptionist?

Either can work. What matters is how it routes. Ask each to show you the message a listing agent receives after a cooperating agent calls and where a named-agent request goes, then ask how junk is billed. Emergencies were about one percent of calls on the line we measured, so round-the-clock emergency handling weighs less for a brokerage than for a property manager.

Do answering services charge for wrong numbers and robocalls?

It depends on the vendor and the contract, so ask for the policy in writing before you sign. On the brokerage line we run, 145 of 1,190 calls, 12.2%, were wrong numbers, robocalls or silence, and on weekdays the 6pm hour was the worst, with 23.7% of its calls junk. A contract that bills every answered call pays for all of them.