Property matching: an AI that finds what the customer describes in your listings

Why filters are not enough

Classic portfolio search is a handful of checkboxes: transaction type, district, rooms, price range. They describe the skeleton of a listing, not what the customer is actually describing. For a customer who says “close to the sea, on a quiet street, within walking distance of a school for the kids”, the deciding part is exactly what the checkboxes leave out. That information usually lives in the listing description, and filter-based search never reads it.

How matching by meaning works

1. Four criteria first: The portfolio is not touched until transaction type, area, rooms and budget are collected. If one is missing, the assistant asks for it.
2. Hard gates: The transaction type and budget filter always apply. A customer looking to buy is never shown a rental, nor a listing far above their budget.
3. Meaning-based ranking: Listings that pass the gates are ranked by how close they are in meaning to the customer's intent; those below a threshold are dropped.
4. A short, honest list: Up to 5 listings are sent in order of fit. For a listing within the tolerance, the price difference is stated openly.

The order is deliberate: meaning-based search is flexible, but it never stretches the limits the customer stated explicitly. We collected all the rules that prevent wrong matches in how to prevent wrong listing matches.

Area: “nearby” has to be genuinely near

The assistant knows Türkiye's official geography: 81 provinces, 973 districts and more than 73,000 neighbourhoods. It knows which neighbourhoods genuinely border each other and places every candidate listing on a five-level proximity scale. If there is nothing in the requested place, adjacent neighbourhoods come first; listings in the distant level never enter the list. Neighbourhoods that share a name are pinned to the right city, and when it cannot be sure, the customer is asked. Details in region intelligence.

Budget: an anchor, not a wall

Customers usually state a round budget and keep a small margin in mind. That is why Realtiq applies a default 10% tolerance to the budget, which can be changed per office. The tolerance is narrow and intentional: a customer with 4 million is not shown a 6-million home, but the right home that missed by a little is rescued. We explain why in budget tolerance.

Matching quality is portfolio quality

Meaning-based search reads listing descriptions; the more detailed the description, the more accurate the match. A listing with an empty property-type field is assumed to be an apartment, so filling that field matters for land, commercial units or shops. The rules for preparing your spreadsheet correctly are in setting up your portfolio sheet.

Listings that never match are information too

Matching does not only find listings for customers; it also tells you something about your portfolio. The Dead Stock list in the CRM shows active listings that have not matched any customer for more than 30 days. Their price, description or the demand in their area is worth a second look.

Frequently asked

What does matching by meaning mean?

Comparing what the customer wrote with listing descriptions by meaning rather than word for word. A customer asking for "a quiet street" can be shown listings that describe that quality even if the description does not use those exact words. The transaction type and budget filter work together with this search.

Are listings above the budget shown?

Only within the tolerance. The default tolerance is 10% and can be changed per office; a customer who says 4 million can be shown listings up to 4.4 million. In that case the price difference is stated openly to the customer.

What if there is nothing in the requested neighbourhood?

The assistant says so openly and suggests listings in physically adjacent neighbourhoods, labelling them as neighbours. Distant listings are never offered as nearby alternatives; they do not enter the list at all.

How many listings are sent to the customer?

Up to 5 listings, in order of fit. Showing a few listings that genuinely fit, instead of a long list, makes it easier for the customer to decide.