San Francisco's City Website Has Thousands of Duplicate Images — and It's Costing More Than Anyone Admits

San Francisco's official city website, SF.gov, is carrying somewhere between 12,000 and 18,000 duplicate image files across its content management system — redundant photos, icons, and graphics uploaded multiple times by different departmental staff, according to a Digital Services internal review completed in May 2026. The bloat is not cosmetic. It translates directly into cloud storage costs, slower page-load times for constituents, and hundreds of hours of staff time spent managing a library that nobody has formally cleaned since the city migrated to its current Drupal-based platform in 2019.

The timing matters. San Francisco's Department of Technology is midway through a $4.2 million modernization contract aimed at making SF.gov accessible and fast on mobile devices — a priority after the city's 2025 Digital Equity Report found that 34 percent of Tenderloin and Bayview-Hunters Point residents rely primarily on smartphones to access government services. Duplicate images inflate page weight, and page weight punishes mobile users on constrained data plans. Fixing this is not a tech curiosity; it sits directly in the path of the city's equity goals.

What the Numbers Actually Show

The May audit, conducted by the city's Digital Services team based at 1 Dr. Carlton B. Goodlett Place in Civic Center, flagged that duplicate images account for roughly 22 percent of all assets stored in the SF.gov media library. Cloud storage for the platform runs through a vendor contract that renews annually; the city pays a per-gigabyte rate at scale, and the audit estimated that eliminating confirmed duplicates would reduce the media library's total footprint by approximately 40 gigabytes. At current enterprise cloud pricing tiers — typically ranging from $0.02 to $0.08 per gigabyte per month depending on the tier — that is a modest but real line item, compounding every month the problem goes unaddressed.

Labor costs are less modest. The audit found that content editors across 47 city departments spend an average of 2.3 hours per month hunting through duplicate or near-duplicate images to find the correct, rights-cleared version before publishing a new page. Multiplied across departments and annualized, that figure approaches 1,300 staff hours per year — hours that digital managers say could be redirected to accessibility reviews and Spanish- and Cantonese-language content, both of which SF.gov still produces at lower volumes than city policy requires.

The problem is structural, not accidental. When SF.gov launched its current architecture in 2019, the Department of Technology gave individual departments autonomous upload permissions without a centralized tagging or deduplication protocol. The Recreation and Parks Department alone — which manages more than 220 parks from the Panhandle to Crocker Amazon — has uploaded versions of the same Golden Gate Park aerial photograph at least nine times in different file formats, according to the audit summary provided to The Daily San Francisco.

What Comes Next for SF.gov's Digital Housekeeping

The Department of Technology has issued a request for proposals for an automated deduplication tool that would scan the media library using perceptual hashing — a technique that identifies visually identical or near-identical images even when file names differ. Three vendors responded before the June 30 deadline. A selection is expected by September 2026, with implementation targeting the first quarter of 2027.

In the interim, Digital Services has assigned two staff from its Mission Street office to manually review and retire the highest-traffic duplicate assets — roughly the top 500 images by page-view count. That work began July 1 and is expected to take six weeks.

For San Francisco residents, the practical upshot is incremental but real. Pages on SF.gov that currently load in 4.1 seconds on a mid-tier mobile connection — a benchmark recorded in the city's own performance dashboard — could drop toward the 2.5-second target the Department of Technology set in its 2025-2026 strategic plan. That gap is the difference between a resident completing a permit application on a Muni bus and giving up before the form renders. On a city website that logged 11.3 million unique sessions in fiscal year 2024-2025, shaving that load time is not a minor adjustment.