Mobile Proxy Real Estate Data

Real Estate Data Collection

Property is the clearest example of a vertical with no global source. Every market has its own incumbent portal, each with its own data model, its own conventions for describing a property, and its own idea of what counts as a price. The United Kingdom runs on Rightmove and Zoopla, the United States on Zillow and Redfin, France on SeLoger, Germany on Immobilienscout24, Spain, Italy and Portugal on Idealista, the Netherlands on Funda, Australia on realestate.com.au and Domain. There is nothing to scrape once and reuse.

These portals are built for domestic users and behave accordingly. Search interfaces, result sets, price history and sold-price archives are frequently abridged or simply withheld from visitors who do not look domestic, and several apply progressively harder bot mitigation to foreign address ranges — reasonably enough, since they have almost no legitimate domestic traffic from them. An exit inside the country removes both problems at once, and it is the difference between collecting the market and collecting a public teaser of it.

Monitoring Property Listings

The most valuable field in property data is one no portal publishes: how long a listing has been sitting. A portal shows you a property, not its history. Recording first-seen and last-seen dates against a stable listing identifier turns a series of snapshots into time on market, withdrawal rate, re-listing behaviour and price-reduction cadence — which between them describe whether a local market is tightening or softening far better than the headline asking price does.

listing_id        …          # portal's own identifier, kept stable
first_seen        2026-05-02  # you derive this; nobody publishes it
last_seen         2026-08-24
price_history     [(date, price), …]   # every observed change
status            listed | under_offer | withdrawn | sold
market            GB          # exit country used to collect
postcode_area     …           # the smallest geography the portal exposes

Two traps are worth naming. Portals re-issue identifiers when an agent re-lists a property, so treating the identifier as permanently stable will silently reset your time-on-market clock — match on address and characteristics as well. And a listing disappearing does not mean it sold; withdrawn and sold are different outcomes with opposite meanings, and portals are inconsistent about distinguishing them.

Regional Market Trends

Property is intensely local, which makes aggregate national figures the least useful thing you can compute. The interesting variation sits at the level of a postcode district or an arrondissement, and it frequently runs in the opposite direction to the national headline. Collecting at the smallest geography a portal exposes, and aggregating upward later, preserves that. Collecting at city level throws it away permanently.

This is also where multi-market coverage pays off, because the comparisons are the product: is supply tightening in one country while loosening in its neighbour, are price reductions becoming more common in a given city, is the rental share of listings growing. None of that is visible from a single national portal, and all of it needs the same collection running consistently in several countries at once.

Scaling Real Estate Scraping

The cadence discipline here is unusually forgiving, and most programmes get it wrong in the expensive direction. Property listings change on a scale of days, not minutes. A daily pass is almost always sufficient and a weekly pass is frequently fine, so an aggressive schedule adds no signal whatsoever while consuming exactly the access you need to protect for a multi-year series.

Because the value compounds with the length of the series, patient collection is strictly better than fast collection in this vertical. Treat the goal as running the same daily job for three years without interruption, rather than as covering every portal in the first month. A programme that survives is worth several that were comprehensive and stopped.

• One exit per national portal — Do not attempt a German portal from a Spanish address. The portal has no reason to serve it well and every reason to challenge it.
• Hold the session through a paginated search — Result sets are stateful; rotating mid-pagination produces duplicates and gaps in the same run.
• Collect at the smallest geography offered — You can always aggregate upward. You can never disaggregate a city average afterwards.
• Store the raw listing page — Portals restructure their markup regularly, and the raw capture is what lets you re-extract a field you did not originally parse.
• Alert on collection health, not just on data — A silent parser failure on one portal looks exactly like a quiet market, and both produce a flat line.

For the general collection discipline see web scraping best practices, and for the sampling design behind multi-market comparison, market research.

Reach Each National Property Portal

Live PXM2 locations — pick the countries whose property markets you track and collect from inside each:

Frequently Asked Questions

Related Mobile Proxy Guides

Property is a national-portal problem, which makes the multi-market sampling and pricing guides the closest neighbours.

Business use cases

Core mobile proxy guides