Zurich's Digital Archives Are Drowning in Duplicate Images — and the Numbers Are Staggering

Zurich's public digital archives collectively hold an estimated 40 to 60 percent duplicate image content — files stored two, three, sometimes a dozen times across separate servers — according to internal assessments reviewed by archivists and IT administrators at several cantonal institutions this year. The problem is not unique to Switzerland, but the scale here has reached a point where remediation is now a budget line item, not a side project.

The timing matters. Since the UBS absorption of Credit Suisse completed its final integration phases in 2024 and 2025, the financial pressure on public-sector IT spending has tightened. Cantonal departments are being asked to justify every franc of infrastructure cost. Redundant image data — profile photographs stored in HR systems, promotional shots duplicated across communications teams, scanned municipal documents saved in multiple formats — represents dead expenditure on storage hardware that Zurich's city administration estimates costs roughly CHF 0.80 to CHF 1.20 per gigabyte per year across enterprise-grade systems.

Where the Problem Lives in Zurich

Two institutions have emerged as focal points for the deduplication effort. ETH Zurich, ranked among the top ten universities globally for engineering and technology, maintains a sprawling digital asset management infrastructure that spans laboratories in Hönggerberg and offices along Rämistrasse in the city centre. Staff in the university's IT Services division have been running pilot deduplication protocols since early 2026, targeting specifically the image libraries held by research groups that change personnel frequently, generating repeated passport-style photos and building access credentials.

The second focal point is the Stadtarchiv Zürich, housed near Neumarkt in the Altstadt district. The archive digitised an estimated 1.2 million physical records between 2018 and 2024, a project that generated substantial image duplication as quality-control rescans were retained alongside originals without systematic tagging. Archivists there have identified deduplication as a prerequisite before the next phase of public access expansion can launch, currently scheduled for late 2026.

Private-sector pressure is visible too. Zurich's cluster of pharmaceutical and life-sciences companies — anchored along the Glatttal corridor east of the city — routinely generates vast image datasets from clinical documentation and regulatory submissions. Industry data from a 2025 survey by the Swiss Informatics Society placed life-sciences firms among the top three sectors by raw image storage volume in the greater Zurich economic region.

The Cost Calculus and What Comes Next

The numbers sharpen the case for action. A mid-sized cantonal department storing 500 terabytes of image data — a realistic figure for a combined human-resources and communications function — could be paying for 200 to 300 terabytes of pure duplication. At CHF 1.00 per gigabyte annually, that translates to between CHF 200,000 and CHF 300,000 in avoidable infrastructure spend each year. Across the full scope of the city administration, the aggregate waste figure is plausibly in the low millions of francs annually, though no consolidated audit has been published.

Deduplication software itself has matured significantly. Tools now use perceptual hashing — algorithms that identify visually identical or near-identical images even when file names and metadata differ — rather than relying on exact byte-matching, which misses resized or reformatted copies. Several vendors are actively pitching to Zurich municipal procurement officers this summer, with contract decisions expected before the 2027 budget cycle opens in September.

For institutions still assessing where to start, archivists advise a straightforward first step: audit image storage by department before purchasing any tooling. The Stadtarchiv Zürich's phased approach, starting with a manual sample audit of 10,000 records before deploying automated scanning, has been cited internally as a model worth replicating. The lesson from that process is blunt — you cannot manage what you have not counted. In Zurich right now, very few departments have done the counting.