What we have built for agents and developers
For BS Immobilienkontor, an estate agency with four branches in north-west Germany, we built the WordPress site including the OpenImmo integration. The listings — around two hundred active offers — are not typed into WordPress; they come in from the agency software, with images, key data and the mandatory energy-certificate details. Whatever is marked as sold in the office disappears from the site on the next import run. The contact routes are separated by intent, because a buyer, an owner and an investor have nothing in common except the page they land on.
For AMR Wohnbau, a residential developer, project marketing runs on a WordPress site with fixed building blocks for project presentation, image gallery, key data and the enquiry route. The company publishes new projects itself as soon as they are ready for marketing — there is no development step between the decision and the publication.
Both projects are documented in our work section; the real-estate sector page describes the thinking behind them: the website is an output channel, not a second data store.
OpenImmo: from the back office onto the website
OpenImmo is the XML exchange format of the German real-estate industry. Practically every agency software exports it: an XML file with the property data plus images, usually delivered as a ZIP via FTP or SFTP into an agreed directory. Setting up an integration means agreeing the delivery path, telling full and partial deliveries apart, evaluating the action type per listing, catching empty deliveries — and logging every run, so it stays traceable what was imported and when.
We define the field mapping against a real export, not against the manual: property types, marketing types, area fields and the energy-certificate values sit in different places depending on the software. That is why every project starts with a test export from your agency software. You then get a written assessment of what maps directly, what needs mapping work and where values are missing.
If a mandatory value is missing from the export, it shows up in the log instead of silently producing an incomplete listing. The import does not decide whether a property belongs online — the data in the back office decides that.
Leads: separate routes instead of one collective form
The first step towards better enquiries costs no software, only structure: one enquiry route per intent. A buyer names the property and their timeframe, an owner describes their property for a first valuation, an investor asks differently from both. Enquiries arrive in a form you can actually work with — no round of follow-up questions.
The second step is storage: enquiries land not only in a mailbox but also in a searchable overview with status and notes. An email that ends up in spam or gets deleted by accident is then no longer a lost lead. This is exactly how we run our own website.
If your needs grow beyond that — several branches, matching prospects to properties, analysis over years — we build lead and customer management as custom software. Our ongoing CRM project in the vehicle trade shows what that looks like: dedicated objects for customer, case and stock, duplicate detection across several sources, roles and permissions. We do not sell an off-the-shelf agent CRM — if an existing product fits better, we say so.
More than one website: landing pages on their own domains
An agency website answers many questions — a landing page answers one, but completely: the valuation in a specific town, the specialist field, the individual development project. We run such pages as part of the website or on their own domain; additional domains serve the same content and point to the main address via canonical links, so visibility accumulates in one place instead of spreading thin.
Where structured property data exists, landing pages can also be generated from the data — one page per concrete search question. Structure, values and mandatory details come from the database; a language model only handles the enrichment, such as turning attribute lists into readable paragraphs. Without a data basis no sentence is produced, and whether a page appears is decided by the state of the data — not by the model. Our page on AI-assisted landing pages explains the details.
We handle operations and maintenance on our own servers in Germany: updates, backups, monitoring and TLS, with maintenance agreements in four tiers.