85K Rows of Houses/Day Into CRM via API - Set up in 7 Days

A Swiss real estate agency asked us to find newly listed properties from private sellers before anyone else did. The target sites were the two largest portals in the country, Homegate and ImmoScout24, and the window for acting on a new listing turned out to be measured in hours.

Sources: homegate.ch, immoscout24.ch
Coverage: all property types for sale in Switzerland, parking spaces included; private sellers only
Frequency: every workday, May 2024 to January 2025
Volume: around 85,000 rows per day
Delivery: email report every workday at 7 a.m. plus upload into the client database via API
Setup and launch: 7 days from the go-ahead to the first delivery; ImmoScout24 added five weeks later

The challenge: the first three calls take the deal

The client described the dynamic on their market plainly. As soon as a private owner publishes a listing themselves, they receive around 20 incoming calls from agencies within the first day, each trying to position itself as the exclusive agent and collect the commission. If an agent is not among the first three callers, the lead is effectively gone.

That turned the project into a timing problem rather than a volume problem. A dataset delivered a day late has no commercial value at all, even if it is complete and clean. The pipeline had to run on a schedule the agency could build its morning around.

Telling private sellers from agents in disguise

The second problem was harder than the scraping itself. A meaningful share of listings that look private are posted by agents who hide their agency affiliation, mimicking private sellers to attract exactly the same calls. Feeding those into the report would waste the agency's most expensive resource, the first hour of the working day.

So beyond collecting the listings we built a filtering layer that analyses the key features of each listing to separate genuine private sellers from agency postings that present themselves as private. The client also supplied agents' phone numbers to exclude. When they flagged listings they thought we had missed, we traced every one: six had been published after the run and were already in the next file, eleven private sellers had been lost by a bug when contacts were saved, and two were agents caught by the client's own list. The bug was fixed, and the client's check of the next files came back clean.

How the CRM pipeline ran

• Scraping all available listings on both portals every workday.
• Filtering down to listings that appeared within the last 24 hours.
• Analysing key features to distinguish agency listings from private ones.
• Matching phone numbers across both portals and earlier reports.
• Generating a report containing only the private listings, with seller contact details.
• Uploading the report into the client database via API.

One CRM feed from Homegate and ImmoScout24

Homegate came first, with the first delivery on 14 May 2024. ImmoScout24 followed on 20 June with one extra rule from the client: phone numbers already reported from Homegate stayed out, so the sales team would not call the same owner twice. From then on both portals shared one contacts database, and the morning file had a sheet per portal plus a sheet of new private listings from both, without duplicate contacts. On the client's side, each record was checked once more on its way into the CRM, and any number already there was dropped.

What the agency got

Every workday at 7 a.m. the agency received a report with all new listings published on the two portals, including the contact details of private sellers. The daily dataset ran to roughly 85,000 rows, and the whole system was set up and launched in 7 days.

About three weeks in, the client told us what had changed in sales: with the private-seller filter in place, their salespeople were now getting through to private owners in time and closing deals. Before that, the scraping they had tried brought in too little, and not the listings that mattered. The feed ran until the end of January 2025, when the client paused it.

The report was the first thing the sales team worked through each morning, the only point at which the data was still worth anything.

See our Homegate scraper, our ImmoScout24 scraper and our lead scraping service; data can also come through our API or as a JSON feed. To set up a morning lead feed for your own market, contact our team.