Off-market listings are a larger slice of the Australian market than most buyers realise. In parts of Sydney and Melbourne, agents will tell you that a meaningful share of premium sales never reach realestate.com.au or Domain at all. The property changes hands quietly, matched to a waiting buyer before a single photo goes online. For the agent, the value is not in the advertising. It is in the research: knowing which owners might sell, which buyers are ready, and how to line the two up first.
That research is slow, repetitive and easy to let slide when listings and open homes fill the week. It is also exactly the kind of work Claude handles well when you point it at your own data rather than the open web.
What off-market research actually involves
Behind every off-market match sits hours of quiet groundwork. An agent building a private pipeline is usually juggling several streams of information at once, most of it scattered across a CRM, a spreadsheet, old emails and their own memory.
The recurring tasks tend to look like this:
Reviewing ownership tenure and last-sale data to spot owners approaching a typical sell cycle
Matching a buyer's brief (budget, suburb, land size, must-haves) against homes not currently listed
Reading back through past conversations to find owners who once said 'maybe in a year or two'
Drafting tailored approach letters and follow-up notes that do not read like a mailmerge
Keeping a tidy record of who has been contacted, when, and what they said
None of this is complicated. It is just volume, and it competes with everything else an agent does. When the research slips, off-market opportunities quietly walk to a competitor.
The information also goes stale fast. An owner who was not ready in March may be ready in September after a job change or a growing family. Keeping that picture current across a few hundred prospects, by hand, is the part that quietly falls over first.
Where Claude fits in the workflow
The useful mental model is that Claude reads and drafts against material you supply. You hand it your buyer briefs, your notes, your street-by-street prospect list, and it does the reading, sorting and first-draft writing that eats your evenings.
A typical loop looks like this. An agent has a buyer chasing a four-bedroom home on a big block in a tight pocket of a city fringe. They paste the brief plus a spreadsheet of owners in the target streets, with tenure and last-sale notes, and ask Claude to shortlist the ten owners most likely to consider selling and explain the reasoning for each. Claude returns a ranked list with a short rationale, then drafts a warm, specific approach note for the top three. The agent edits, adds the local detail only they know, and sends.
The same pattern covers buyer-match summaries, recap notes after an appraisal, and turning a rambling voice memo into clean CRM notes. It is the admin around the relationship, not the relationship itself.
It also works in reverse. When an owner quietly signals they might sell, an agent can ask Claude to scan their active buyer list and surface the three or four buyers whose brief fits best, with a one-line reason for each. That turns a vague note into a shortlist you can act on the same afternoon, which is often the difference between holding the listing and losing it.
The AUD maths for a single agent
Say an agent spends around eight hours a week on off-market research and the writing that goes with it. Trimming that to roughly three hours with Claude doing the first pass gives back five hours a week. At a loaded cost of about $65 an hour, that is close to $15,000 of an agent's time returned over a year, redirected to listing appointments and buyer calls.
The bigger number is on the other side. A single extra off-market sale on a $1.2M home at a 2% commission is about $24,000 in gross commission. The research tooling does not have to be perfect. It has to help you convert one or two matches a year you would otherwise have missed, and the case pays for itself many times over.
Guardrails you cannot skip
Off-market work sits close to people's private circumstances, so a few rules are non-negotiable. Claude should draft and summarise from information you already hold lawfully, never invent owner contact details or personal facts, and never be used to scrape or assemble a profile of someone from across the web.
Under the Australian Privacy Act, how you collect and store owner information matters, and an AI drafting tool does not change your obligations. Treat every Claude output as a first draft: verify names, ownership and figures against your title and CoreLogic sources, keep a human sign-off before anything goes to a prospect, and make sure your approach notes are honest about how you came to contact them. Used this way, Claude speeds up the parts that should be fast and leaves the judgement with you.
If you run a real estate business and want a research workflow built around your own data and Australian privacy obligations, we can map it out with you. Book a brainstorm and we will walk through where it fits.
