Can You Use ChatGPT to Write Real Estate Prospecting Letters?

Somewhere in the last two years, "write me a real estate prospecting letter" became one of the most common things agents type into ChatGPT. Fair enough — the blank page is the single biggest reason prospecting letters don't get sent, and AI kills the blank page in eight seconds.

So here's the honest answer to whether you should use ChatGPT for real estate prospecting letters: yes, for drafting. It's genuinely good at that, and pretending otherwise is snobbery. But there are two problems the draft button doesn't solve — one about the words, one about the envelope — and the agents winning listings with AI-assisted letters are the ones who understand both.

What ChatGPT genuinely fixes

Give AI its due, because it's real:

• The blank page dies. The hardest sentence in any prospecting letter is the first one. A draft — even a mediocre draft — turns "I should do a letter drop" from a someday-task into a this-afternoon task. Campaigns that actually get sent beat perfect campaigns that don't, every time.
• Structure comes free. A decent prompt produces a letter with a warm opening, a reason for writing, a credibility line and a call to action in the right order. That's more structural discipline than plenty of hand-written-from-scratch agent letters manage.
• Variations are instant. Warmer, shorter, more direct, aimed at downsizers instead of investors — each variant used to be twenty minutes of rewriting. Now it's one sentence of instruction. For testing different angles across a farm area, that's a real capability, and it's what makes a proper multi-touch sequence feasible for an agent who'd never write six letters from scratch.

If your alternative was a stiff template read ten thousand times or — worse — no letter at all, an AI draft is an upgrade. Our template library and an AI draft solve the same problem from different ends: templates give you a proven skeleton, AI gives you a fast first pass. The best use is both — a known-good structure, drafted fresh in your situation's specifics.

Where the raw draft quietly fails

Now the problems with sending what the machine gave you, untouched.

Everyone's draft sounds like everyone else's draft. Language models write toward the statistical middle — the most likely next word, every word. That's why AI text has a recognisable flavour: "I hope this letter finds you well", "in today's dynamic market", "I would welcome the opportunity". Homeowners see AI-flavoured writing in their inbox every day now, and their radar is improving fast. A prospecting letter's entire job is to feel like one person writing to another; the house voice of a language model is the exact opposite.

It invents local facts. Ask ChatGPT about your suburb and it will confidently produce median prices, "recent strong sales on your street" and demand claims it has no data for. In a prospecting letter, a made-up number isn't a small glitch — it's the fastest possible way to torch credibility with the one reader who knows the truth: the person who lives there. Every local claim in an AI draft has to be checked or cut. No exceptions.

It doesn't know the physical page. ChatGPT writes for a screen that scrolls. A real letter is a fixed sheet — an A4 page holds roughly 300 written words, a note card half that. AI drafts routinely run long, and a letter that spills awkwardly onto a second page (or gets shrunk to fit) loses the effortless, personal feel that makes it work.

It forgets the response mechanism. Left alone, AI closes with "please don't hesitate to contact me" — a CTA with no offer, no deadline and no way to know the letter worked. The call to action is where prospecting letters are won and lost, and it's precisely the part that needs your offer, your number, your link. The machine can't know it.

All four are fixable in one editing pass. Which brings us to the workflow.

The mailbox problem AI can't touch

First, the bigger issue — the one no prompt can fix.

Say you get everything above right: a genuinely good letter, personal, accurate, sharp CTA. Then you print it on office paper, in a printed envelope with a window face, and drop 500 of them.

You've just produced junk mail with excellent words inside.

Homeowners triage their mail in about two seconds, and the verdict — personal or junk — is decided by the outside of the envelope before a single word is read. Printed and bulk-looking means binned unread; hand-addressed in real ink means opened. We've covered the evidence in handwritten vs printed mail, and the gap isn't subtle: single-digit open rates against 90%+.

Here's the irony worth sitting with: as AI makes words cheap, the premium on proof-of-effort goes up. Every business in the country can now generate polished copy instantly — which means polished copy signals nothing. What can't be faked at scale is a letter that visibly took human effort to produce: real ink, real handwriting, a real stamp. AI didn't weaken the case for handwritten mail. It strengthened it, by devaluing everything else in the letterbox.

So the strategic picture is simple. AI solves the writing problem. It cannot solve the opening problem. Agents who use AI to write and then send its output as ordinary printed mail have automated the cheap half of the job and skipped the half that decides whether anyone reads it.

The workflow that actually works

Draft with AI. Make it yours. Send it in real ink. In practice:

1. Brief it properly (2 minutes). Don't ask for "a prospecting letter". Tell it who you are, who the reader is, why you're writing now, and what you want them to do: "I'm a local agent writing to homeowners in [suburb] after a strong nearby sale. Audience: owners 10+ years in their home, possibly thinking about downsizing. Goal: book no-obligation appraisals. Warm and plain-spoken, under 250 words, no marketing clichés." The quality of the draft is set by the quality of the brief.

2. Do the make-it-yours pass (10 minutes). This is the step that separates letters that work from letters that get made. Four moves:

• Check or cut every local claim. Replace invented numbers with your real ones — the actual sale, the actual street, the actual result.
• Add one sentence only you could write. A named street, a detail from a recent open home, something true about the area this month. One genuinely local sentence does more than ten polished paragraphs.
• Kill the AI tells. "I hope this finds you well", "dynamic market", "don't hesitate" — delete on sight, and read the result aloud. Anything you wouldn't say to a homeowner at their front door, rewrite until you would.
• Rebuild the close. Your offer, one clear action, a reason to act now, and a way to capture the response.

3. Cut to fit the page. Trim to what a single sheet actually holds. Shorter reads more personal anyway — a half-page note from a person beats a full page from a marketer.

4. Send it handwritten. Real pen, hand-addressed envelope, real stamp. This is the step that gets the other three read.

Ten to fifteen minutes per letter template, then the sequence runs across the whole farm. That's the honest cost of doing it right — dramatically less than writing from scratch, and dramatically more effective than pasting raw output into a mail merge.

(It's how we think about it at Scribbly, too: the letter editor in our order form has writing assistance built in — it drafts to the exact paper size, keeps your response link in, and signs off as you, then a real pen writes the result. AI drafts, you make it yours, ink does the convincing.)

The bottom line

Use ChatGPT for your prospecting letters — for the draft. It kills the blank page, it makes multi-touch sequences feasible, and refusing the help is just slower, not nobler. But the draft is the beginning of the job: check the facts, add what only you know, make it sound like you, and put it in an envelope a human being visibly touched.

Because in a letterbox filling up with machine-polished mail, the letter that wins the listing is the one that couldn't have been mass-produced — and now you can produce it in a quarter of the time.

Start with the evidence for why the envelope decides everything: handwritten vs printed mail.