A Berlin Property Portal That Lifted Qualified Leads 63 percent

Losing buyers to speed

When a Berlin property platform first came to us, the symptom was easy to name but painful to live with. Buyers were leaving before the page even finished drawing. On a phone, the old site needed several seconds to show a single listing, and in Berlin that is plenty of time for someone to bounce back to Google and tap a faster competitor instead. The team could watch it happen in their analytics: high intent visitors arriving from search, then vanishing inside the first few seconds. Every one of those was a buyer the agents never got to talk to.

The cost was not only lost sessions. Slow pages and shaky Core Web Vitals were quietly dragging the whole catalogue down the rankings, in one of the most contested property markets anywhere. The marketing team was paying more for traffic that converted less, and leadership could feel the business leaking value month after month.

The problem under the hood

Once we looked past the loading spinner, the real issues were structural. Search ran straight against the main database, so the more listings the catalogue held, the slower a filter became, exactly the wrong way round for a portal that wanted to grow. Pages were assembled in the browser, which left search engines staring at thin HTML and left phones doing work they should never have been asked to do. URLs were messy, structured data was missing, and there was no clean way for marketing to see which journeys actually produced a lead. The site was not just slow. It was built in a shape that fought every goal the business had.

What we rebuilt

We rebuilt the portal from the ground up on Next.js, with a clear priority order: make it fast on a mid range phone first, make it easy for search engines to read second, and make every lead measurable third. Rendering moved to the server, listing pages were generated ahead of time and refreshed as inventory changed, and a dedicated search index took the pressure off the main database so a filter stays quick whether the catalogue holds two hundred thousand homes or two million.

• A full rebuild on Next.js with server rendering and listings regenerated as inventory changes
• A dedicated search index tuned for under one second results across millions of listings
• Image optimization, edge caching, and a strict mobile first performance budget
• Clean URLs, structured data, and programmatic listing pages for search visibility
• A lead capture flow instrumented end to end so marketing can see what actually converts

What we delivered

What changed

The difference showed up first in the numbers the team had been worried about, then in the ones that pay the bills. Mobile pages now paint in about nine tenths of a second, comfortably inside the green band for Core Web Vitals, and search returns results in under a second even across two million indexed listings. As performance and structure improved together, rankings recovered and organic traffic climbed 28 percent. Most importantly, the agents started getting better. Within two quarters of launch, qualified leads were up 63 percent, because the people arriving from search were now staying long enough to enquire.

One moment stuck with the team. A senior agent had built a habit of apologising on calls while a listing crawled open on the prospect's phone, filling the silence so the buyer would not hang up. A few weeks after launch she mentioned, almost in passing, that she had stopped doing it, because the page was already there before she finished saying hello. Small as it sounds, that was the day the rebuild felt real to the people using it every day.

The new portal is a different business to run. Pages open instantly on a phone, we are climbing back up the rankings, and the agents keep telling me the leads coming through are far better qualified than before.— Marketing Lead, a Berlin property platform

The stack

• Next.js
• TypeScript
• React
• PostgreSQL
• Elasticsearch
• Vercel edge CDN
• Tailwind CSS