Every year the National Association of Realtors (NAR) in the U.S. asks its members how they use technology. I asked for this year’s full report, released last week, because it’s the closest thing our industry has to a scorecard on AI. Canada doesn’t keep one. I looked for a CREA, board or regulator survey on how registrants here use AI and couldn’t find one. So the American numbers are what we have, and they deserve a close read, because the headline and the detail tell different stories.

The headline is adoption. In June, 23 per cent of Realtors said they use AI daily and another 25 per cent weekly. A year earlier, nearly a third had never tried it. Now only 12 per cent say they have no plans to. Fifty-five per cent say AI has had a positive effect on their business, up from 50 per cent.

The detail is the payoff. Last year, 17 per cent said AI had a significant positive impact on their business. This year it’s 16 per cent. Nearly half the industry now uses AI every week, and the share getting a big result didn’t move. Almost all the growth landed in “moderately positive.”

One number went the wrong way. Last year, 4 per cent of Realtors said AI had a negative impact on their business. This year it’s 11 per cent. Roughly one in nine agents now say the tool is costing them something.

Surveys like this have limits. NAR invited more than 73,000 members and got 1,165 usable responses, a response rate of 1.5 per cent, and last year’s respondents were different people. A one-point move from 17 to 16 is noise. A jump from 4 to 11 is not.

 Where the time goes

 

The report also lists what agents who use AI actually do with it. Writing listing descriptions leads at 75 per cent. Social media posts, 56 per cent. Emails and follow-up, 52 per cent.

Then look at the work closest to the money. Pricing insights and comps support: 23 per cent. Improving lead generation and prioritization: 11 per cent. Sending clients automatic status updates: 9 per cent. Filling in forms from information the client already provided: 7 per cent.

The industry has adopted AI as a writing tool, and writing is the cheapest part of this job. A listing description that took 10 minutes now takes 30 seconds. That’s more than nine minutes back. No seller has ever hired an agent because of the description.

Agents know where the money is. Asked why they adopt new technology, 81 per cent said to save time and 57 per cent said to close more deals. Asked what they want to try next, 48 per cent named AI-powered lead generation and follow-up. The intent is there. What stops them is the learning curve, which 63 per cent named as their biggest obstacle, ahead of cost at 59 per cent.

 

From a broker’s desk

 

The report doesn’t say why 11 per cent feel AI has hurt them, and I won’t pretend to know. What I can describe is what AI mistakes usually look like when they cross my desk. They are rarely dramatic. A description that promises a feature the home doesn’t have. A market summary built on last year’s numbers. A client email that reads like every other agent’s email, so the client stops opening them. Each one is small. Advertising that is false or misleading is a regulatory problem in every province, no matter who, or what, typed it.

 

What to do with this

 

Rank your AI uses by where your income comes from. Write down everything you used AI for last month. Beside each one, note whether it ever brought in a client or protected a deal. If the list is all writing, you’ve found your gap.

Build one system, not one more prompt. Pick a single money task from the short end of NAR’s list, like a pricing brief before every listing appointment or a Friday status update to every active client, and set it up so it runs the same way every time. One system you trust is worth more than a hundred prompts you retype.

Read everything before it goes out. If one agent in nine reports harm, the fix starts with a person reading every word that goes to a client. Your name is on it either way.

Keep your own score. Nobody in Canada is measuring this, so measure yourself. Track the hours a system saves and whether it touches a lead or a deal. Brokerages can ask their own agents the questions NAR asked. It takes a week and a free survey form, and it would tell us more about Canadian agents than any American report can.

 

The same tool

 

Last year the question was whether agents would use AI. That one’s been answered. The question now is what they use it for, and so far the answer is the part of the job clients notice least.

One more number from the report stays with me. Ninety-one per cent of Realtors who use AI use ChatGPT. So the agents in that 16 per cent are mostly holding the same tool as everyone else. My bet is they’re pointing it at a different part of the job.