How Berlin's Housing Databases Got Buried in Duplicate Images — and What It Cost the City

Berlin's Senate Department for Urban Development confirmed earlier this year that a systematic audit of the city's centralised housing database had uncovered tens of thousands of duplicate and misattributed property images — a problem that officials say traces back at least a decade, to the rushed early phases of the city's digitalisation drive. The audit, conducted under the umbrella of the Berliner Wohnraumatlas programme, found that district offices from Marzahn-Hellersdorf to Charlottenburg-Wilmersdorf had uploaded scanned records without standardised file-naming protocols, meaning the same building photograph could appear attached to three separate parcels in Lichtenberg while a new-build on Sonnenallee in Neukölln sat with no image at all.

The timing matters. Berlin is in the middle of a bruising political fight over rent caps and new social housing targets, and the Senate cannot afford to have its core property data questioned by landlord associations, tenant groups, or the courts. When a data foundation is unreliable, legal disputes over rent levels or building classifications become exponentially harder to resolve. The Mietendeckel debate never fully went away after the Federal Constitutional Court struck down the 2020 version; any successor policy will live or die on the quality of the underlying cadastral and housing data.

How the Duplication Problem Grew

The roots go back to 2014, when the Senate launched the first phase of its eGovernment Berlin initiative, pushing district offices to digitise paper planning files. Each of Berlin's twelve districts maintained its own scanning workflow. Tempelhof-Schöneberg used one contractor; Pankow used another. Neither handed over images in a compatible metadata format. By 2018, when the central FIS-Broker geodata platform — run by the Senate Department for the Environment, Urban Mobility, Consumer Protection and Climate Action — attempted to consolidate records, administrators found themselves inheriting a patchwork of JPEGs, TIFFs, and low-resolution PDFs with conflicting geotags.

A 2021 internal review flagged the duplication risk but was not acted on with urgency, according to documents later published under a Freedom of Information request to the Abgeordnetenhaus. The review estimated that roughly 12 percent of residential property entries in the unified database carried at least one incorrectly associated image. By the time the 2025-26 audit began under the current Senate's digital reform agenda, that figure had grown. The cost of manual remediation — cross-checking photographs against official cadastral data held by the Senatsverwaltung für Finanzen — is now projected to run into several million euros, though the Senate has not yet published a final contract figure.

What Comes Next for the Database

The Senate's current plan centres on deploying an automated image-deduplication pipeline, developed in partnership with the Zuse Institute Berlin on Takustraße in Dahlem, one of the city's leading applied mathematics research bodies. The system uses perceptual hashing to flag near-identical images and cross-references GPS metadata against the official parcel boundaries held in the ALKIS land register. A pilot covering around 40,000 properties in Friedrichshain-Kreuzberg and Reinickendorf is scheduled to run through the end of 2026, with a city-wide rollout targeted for 2027.

For ordinary Berliners, the practical stakes are concrete. Tenants challenging service charge calculations or landlords appealing against heritage-listing decisions both rely on accurate property records. Estate agents working along Karl-Marx-Allee and in the rapidly changing corridors around the BVG's new U5 extension stops have long complained that discrepancies between what a listing shows and what the official record contains slow transactions by weeks. A cleaner database will not solve Berlin's housing shortage, but it will at minimum stop bad data from making every bureaucratic step harder than it needs to be.

The Zuse Institute pilot results are expected to be presented to the Stadtentwicklungsausschuss, the Abgeordnetenhaus urban development committee, in November 2026. If the error rate drops below two percent — the Senate's stated benchmark — the full contract will be awarded in early 2027. If it does not, the Senate has indicated it will reopen the tender. Either way, the city is finally, methodically, dealing with a problem it spent ten years pretending was not there.