AI Agents for Real Estate Leads: Speed Wins the Deal

A buyer submits a form on a listing and keeps scrolling, often the same search on two other sites, while your agent is still in a showing across town. An AI agent for real estate leads can text or call back within minutes, ask the questions that separate a browser from a buyer, and book a showing or a callback before that scroll ends. Oldroyd, Elkington, and McElheran, in the Lead Response Management study from MIT and the Kellogg School, measured a callback that slips from 5 minutes to 30: contact odds fall 100 times, and qualification odds fall 21 times. Kellogg surveyed 495 companies between June and September 2007, and a separate three-year file followed six companies through at least 15,000 leads and at least 100,000 call attempts. The first-reply job is one we build for home service companies and clinics, where missing the call means missing the booking. A real estate team lives on that same clock. The note below is about that job on a team. Running a brokerage is outside what we do.

Why does response speed decide who wins a real estate lead?

A buyer who fills out a form is not waiting quietly. They are still browsing, sometimes texting a friend about the kitchen. The agent who reaches them while that attention is open gets the conversation. The agent who calls two hours later gets someone who already booked a showing, or who stopped looking at that house. The MIT and Kellogg study did not measure real estate. The mechanism it documents, interest that fades and a late callback that rarely recovers it, matches how a home search actually happens: short bursts of attention in the evening and on weekends.

What happens when nobody answers for hours?

Zillow's 2025 Consumer Housing Trends Report for Agents, a nationally representative survey of more than 5,000 people including more than 2,200 successful buyers, fielded March through July 2024, found that 47% of buyers hired the first agent they contacted, and 71% approached only one or two agents. Buyers are not scoring a dozen agents on a spreadsheet. Most pick whoever shows up and seems competent. A slow reply hands the lead, the timeline, and the later referral to whoever answered while your phone sat quiet.

Among repeat buyers, who already know the process, 52% still hired the first agent they spoke with. Familiarity does not make people more patient. A repeat buyer who has done this before moves on faster from an agent who takes hours, because they remember what a quick reply felt like last time.

How fast do real estate teams respond to leads today?

The gap is not a lack of interest in AI. It is AI pointed at everything except the minute that decides whether the lead becomes a conversation. The National Association of REALTORS 2025 Technology Survey (1,241 respondents, July 2025, margin of error 2.78 percentage points) found that 41% of REALTORS use AI or generative AI, and one in five use it daily. Only 7% use a chatbot for lead capture or client communication. Social media (39%), the CRM (23%), and the local MLS (17%) remain the top lead-generating technologies, with no instant reply sitting behind them.

Where the gap actually sits.

Agents use AI to write listing descriptions, draft posts, and summarize contracts. Those tasks can slip an hour and nobody loses the house. The first reply to a live inquiry is still mostly manual. A notification buzzes, the agent is finishing a showing or driving, and the callback happens when the schedule allows, often 30 minutes to several hours later. The distance between 41% using AI somewhere and 7% using a chatbot on the lead is the minute this kind of agent is for.

What does an AI agent for real estate leads actually do?

It is not an autoresponder that says thanks, an agent will be in touch. It reads the inquiry, replies in the buyer's channel within a minute or two, asks a short set of questions matched to buyer or seller, and either books a showing or flags a person with the summary and the transcript. The autoresponder only admits the form arrived.

A buyer and a seller need different opening questions. A buyer on a specific listing gets timeline, financing, and whether that house or something similar is the target. A seller, usually from a valuation form, gets timeline to list, whether they have already spoken with another agent, and what is driving the move. The right question set in the first message is what separates a working agent from a generic widget on the form.

Which channel matters most, text or call?

Zillow's same 2025 report found that 50% of buyers who used an agent prefer texting or messaging apps, versus 33% who prefer phone calls. Most teams still call first. A missed call leaves a notification. A text sits there until the buyer has a moment, in the channel they already said they prefer, instead of a cold call landing in the middle of someone else's showing.

A five-question opener for a new lead text.

1. "Thanks for asking about [address]. Are you still looking at this one?" 2. "What is your target move-in window?" 3. "Are you already talking with a lender, or would an introduction help?" 4. "Is this size and area right, or are you open to nearby listings?" 5. "Do you want to see it this week, or would a video walkthrough come first?" Two answers are enough to route the lead.

What questions should it ask before handing a lead to an agent?

Ask too little and the human walks into a showing to discover there is no pre-approval, or the seller listed with someone else last month. Ask too much and the buyer stops before question three, annoyed that a form they already submitted is being repeated one line at a time. The goal is enough to prioritize, not a full intake disguised as a text.

For a buyer, three points do most of the sorting: timeline (this week, this quarter, or browsing), financing (pre-approved, working with a lender, or not started), and whether the listing still fits. A lead with a timeline under 30 days and financing in place should not sit behind a "just browsing" lead that arrived earlier.

When does it stop and route to a person?

Hand off when a buyer names a showing time, mentions financing already in place, or asks about price, terms, or negotiation. Also hand off immediately if the timeline is under two weeks or they mention a competing offer on another property. The scoring pattern this uses is in how lead qualification logic works. A buyer with financing, a clear timeline, and a specific address should reach a person within minutes, not after a long question set.

How does it cover nights, weekends, and open houses?

Inquiry volume clusters when the office is closed. A buyer touring on Sunday afternoon, or scrolling after the kids are in bed, sends the form at the peak of interest. One agent cannot staff every evening and weekend without giving up the family time that made a flexible job attractive. Coverage on those hours catches the moment. A Monday callback list usually means the buyer has already walked three houses with someone else.

Open houses add a second squeeze. A dozen visitors sign in over two hours, and the host cannot text all of them before the next person walks in. A sign-in sheet entered into the CRM two days later has lost most of its value. Working that sheet the same day, while interest is still warm, turns more visits into a second showing without pulling the host away from the door.

What does an AI agent cost compared to hiring an ISA?

This is a round example, not a team we work with. A part-time inside sales agent covering evenings and weekends might be budgeted at $2,000 to $2,500 a month for 20 to 25 hours, plus several weeks to learn the scripts, the CRM, and which questions matter for which lead source. Turnover resets that ramp. A new inside sales agent every six to nine months, which is common in the role, means the team is often paying for training rather than coverage. An agent on the same window runs continuously, does not take vacation, and does not forget the questions on a Saturday when eleven leads arrive during one open house.

The breakeven math in one closed deal.

One missed lead who would have closed is the cost to set against either option. At a commission of a few thousand dollars a side, one buyer who would otherwise have gone to a faster reply covers months of either an inside sales agent's pay or the running cost of the software. The decision is how many leads go quiet every month, and what one of them is worth. Most teams have never run that number, because the lost leads never appear in a report. They stop replying, and the team assumes the lead cooled off on its own.

How do you know the agent is actually working?

Three numbers: response time, qualification rate, and showing-booked rate. Response time is minutes from the inquiry to the first reply. Qualification rate is the share who answer enough to be prioritized. Showing-booked rate is the share of qualified leads who take a time or a callback. Lead source and rough deal size wait until those three are trusted.

In week one, watch response time and qualification rate. Showing-booked rate depends on the team's calendar, which needs a week against real volume. If replies go out in under two minutes and qualification is low, the opening question is too broad, too long, or reads like a script. Read the transcripts where buyers stopped, and shorten one question at a time. If you want to see which lead source loses the most to a slow reply, get your free Systems Plan and bring the missed-lead log. The plan arrives by email a few minutes after you send it, and it covers the first job, the same first-reply work we build for a clinic or a home service company.

Common questions.

How fast should a real estate agent respond to a new lead?

As close to immediately as you can, ideally within minutes. Inside the Lead Response Management study from MIT and the Kellogg School, a callback that waits 30 minutes instead of 5 cuts contact odds by 100 times and qualification odds by 21 times. The team that replies first usually gets the conversation, whatever the later pitch would have been.

Can an AI agent actually respond to real estate leads?

Yes, for the first reply and the qualifying questions. It can read a new inquiry, send an email within a minute or two, ask a short set matched to buyer or seller, and either book a showing or hand a person the transcript. It should not negotiate price or terms, and it should hand off as soon as financing, a specific timeline, or a competing offer comes up.

Is an AI agent cheaper than hiring an ISA for real estate?

Count training time and turnover before you compare the monthly bill. A part-time inside sales agent for evenings and weekends is commonly discussed at an illustrative $2,000 to $2,500 a month plus ramp time, and turnover every six to nine months resets that ramp in many shops. Software covering the same window does not need vacation coverage or a new hire. Weigh that bill against what one lead is worth when nobody answers.

What should an AI agent ask a new real estate lead?

For a buyer: timeline, financing status, and whether the listing they asked about still fits. For a seller: timeline to list, whether they have spoken with another agent, and what is driving the move. Three to five questions are enough to prioritize. Longer sets see buyers stop before you have enough to route them.

Do buyers actually prefer texting over a phone call from an agent?

For the first contact, yes. Zillow's 2025 Consumer Housing Trends Report for Agents found that 50% of buyers who used an agent prefer texting or messaging apps, compared with 33% who prefer phone calls. A missed call leaves a notification. A text waits in the inbox until they have a moment to answer.

Where these figures come from.

Contact odds falling 100 times, and qualification odds falling 21 times, between a 5-minute callback and a 30-minute one, are reported in the Lead Response Management study from MIT and Kellogg (Oldroyd, Elkington, McElheran, 2007). The Kellogg portion surveyed 495 companies. The behavioral portion covered six companies, 15,000 leads or more, and 100,000 call attempts or more. Adoption figures (41%, one in five daily, 7% chatbots, 39% social, 23% CRM, 17% MLS, n=1,241, margin of error 2.78 percentage points, July 2025) are from the National Association of REALTORS 2025 Technology Survey.

The finding that 74% of buyers interviewed only one agent (67% of first-time buyers, 76% of repeat buyers) is from the NAR 2025 Profile of Home Buyers and Sellers (n=6,103 primary-residence buyers, mailed to 173,250 recent buyers in July 2025, margin of error 1.25 percentage points). Buyer channel preference (50% text, 33% phone), the 47% who hired the first agent, the 71% who approached only one or two, and the 52% of repeat buyers who hired the first agent they spoke with, are from the Zillow 2025 Consumer Housing Trends Report for Agents (more than 5,000 respondents, including more than 2,200 successful buyers, fielded March through July 2024). The inside sales pay range is a round example, not a client result.