Automatic buyer↔property matching: your agency’s ‘For You’ feed
Your buyer database probably has 3 to 5 ideal clients for every new property that comes in. Without AI, nobody has time to cross-reference them.
Every agency has a CRM full of registered buyers — people who came in once, said what they were looking for, got two or three suggestions, and got lost in the noise. When a new property comes into the portfolio, nobody has time to comb the base by hand looking for interested buyers.
An automatic matcher does this work every morning, before the advisor starts the day.
What the system does
enters portfolio]) --> B[System reads
full sheet] B --> C[Scans active
buyer base] C --> D[Match score:
budget, area,
type, timing] D --> E{Top 5
matches} E --> F[Notifies
responsible advisor] F --> G[Advisor decides
whom to contact] G --> H([Personalised
outreach])
The system is precise, not exhaustive: prefers 5 very good matches to flooding the advisor with 30 possible ones. The rule we use is conservative — high minimum score, and never more than 10 matches per property.
What AI adds beyond a CRM filter
A good CRM already filters by budget and layout. Where AI adds:
- Interpreting "soft" criteria. A buyer who registered "a place with character" or "something I can personalise" doesn’t filter by checkbox — AI reads the sheet and understands intent.
- Recent intent signals. Who visited the site in the last 15 days, who interacted with emails, who asked for a second viewing on another property — all weighted.
- Life context. A person who registered interest 8 months ago "for next year" is now at the right timing. The CRM doesn’t do that math.
What changes for the advisor
- Warm outreach, not cold. "I saw you were looking in Alvalade — this one came in yesterday, want to see it before it hits the portal?" gets 40-60% reply rate, versus 5-10% on generic outreach.
- Clear priority. The advisor opens the CRM in the morning and sees whom to contact today — no need to do the curation work.
- Fewer properties lost. New properties stop "waiting to find their owner" — the owner is found before the property goes public.
The numbers usually seen
On an active base of 2-4k registered buyers:
- Average time to first viewing on a new property: from 12 days to 4 days.
- Properties sold "off-market" (before public listing): 5% → 25-35%.
- Preserved commissions — off-market sales don’t pay portals and don’t face competition.
Where AI can fail (and how to protect against it)
- Stale base. If half the contacts are from 2019, matching suggests people who already bought elsewhere. Cleaning the base first is mandatory.
- Overfitting to old preferences. Someone who said "only 2-bed" 3 years ago is now married with kids. The advisor always has the option to loosen criteria manually.
- Silent dropping of valid buyers. Never discard completely — "not matched" stays in a second tier, visible if the advisor wants.
The pattern is clear: AI prepares, the advisor decides. And the CRM stops being a graveyard.