How San Francisco's City Websites Ended Up Drowning in Duplicate Images — and What It Took to Notice

San Francisco's official digital infrastructure is carrying thousands of duplicate images — the same photographs, icons, and graphics uploaded multiple times across different city department portals — a problem that accumulated quietly over nearly a decade of piecemeal web expansion and is now driving a formal remediation effort by the city's Department of Technology.

The issue matters now because SF.gov is in the middle of a $4.2 million modernization push that began in fiscal year 2025-26, aimed at consolidating the fragmented content management systems that different agencies built independently after the city first launched its unified web presence in 2017. Auditors working with the Department of Technology found that duplicated media assets were inflating cloud storage costs and degrading page-load performance — a particular problem for residents trying to access permit applications and benefit enrollment tools on mobile networks in neighborhoods like the Tenderloin and Bayview-Hunters Point, where broadband penetration remains uneven.

How the Duplication Problem Grew

The root cause traces back to a structural decision made when the city began migrating departments onto a shared Drupal-based CMS platform starting around 2018. Each department — from the Municipal Transportation Agency to the Department of Public Health — was given its own content team with upload privileges but no shared asset library and no deduplication policy. When a communications staffer at, say, the San Francisco Public Utilities Commission needed a photograph of City Hall for a press release page, they uploaded their own copy rather than referencing a central repository. So did the next staffer. And the next.

The San Francisco Public Library system's digital branch and the Office of Economic and Workforce Development both ran into the same pattern. By the time the Department of Technology commissioned an internal content audit in late 2024, preliminary findings pointed to image redundancy rates high enough to justify dedicated engineering resources. The audit drew on automated scanning tools that flagged files with identical hash values stored under different filenames across departmental subdirectories on the city's Amazon Web Services and Microsoft Azure partitions.

This is not unique to San Francisco. New York City's NYC.gov underwent a similar consolidation between 2019 and 2022 after its own audit found tens of thousands of orphaned and duplicated media files. But San Francisco's problem was compounded by the tech sector's influence on city hiring. Between 2020 and 2023, the city brought in a wave of short-term contract developers — many cycling through as the broader Bay Area tech economy yo-yoed through pandemic hiring freezes and subsequent layoffs — who built one-off solutions without coordinating with the central IT architecture team based at 1 South Van Ness Avenue.

The Cleanup and What Comes Next

The Department of Technology is now piloting a centralized Digital Asset Management system, with a phased rollout that began in March 2026 for three pilot departments. The goal is a single searchable media library where any city staffer can retrieve an approved image rather than upload a fresh copy. Deduplication scripts running on existing cloud infrastructure are also being used to collapse redundant files retroactively, a process that the city estimates will trim cloud storage consumption meaningfully before the end of the current fiscal year on June 30, 2027.

For residents and businesses that interact with city services through SF.gov — filing for a permit at the Planning Department's online portal, checking Muni schedules, or navigating the Human Services Agency's benefits pages — the practical effect should be faster load times and fewer broken image links, which have been a recurring complaint logged through the city's 311 service request system. In fiscal year 2024-25, the 311 system received more than 800 web-related service requests, a subset of which cited non-loading or incorrectly displayed images on official city pages.

City technology staff say the longer-term fix requires governance as much as engineering: a formal media upload policy, mandatory training for departmental content editors, and quarterly audits built into standard operating procedure. Without those, engineers warn, a new tranche of duplicate files will accumulate within two years of any technical cleanup — leaving the city back where it started, paying cloud storage bills for images nobody needed to upload twice.