Assessment of the San Francisco Shelter System

San Francisco residents consistently report that homelessness is the most important issue facing our city. In 2024, over 8,300 people were experiencing homelessness on a single night in January, and more than 20,000 sought homelessness services over the course of the year.

The San Francisco Department of Homelessness and Supportive Housing (HSH) manages the City’s Homelessness Response System; emergency shelter is a critical component of that system. Although the goal of the Homelessness Response System is to move people into permanent housing, temporary shelter provides a place for people to go when they are in crisis and connects people to services that support them in moving into permanent housing.

Understanding how well the shelter system is working is an important part of assessing how well the City is meeting its overall goals for addressing homelessness, especially in decreasing unsheltered homelessness.

The San Francisco Controller’s Office conducted this assessment of the City’s shelter system to provide a neutral perspective on its service delivery and performance. We use a mixed methods approach, including interviews, focus groups, benchmarking, data analysis, budget analysis, and equity analysis.

This report is organized in three primary sections:

1. A broad overview of the shelter system. The goal of this section is to introduce readers to a complex system and equip them with facts to make informed judgements about the system’s performance.
2. An analysis of shelter client experiences and outcomes.
3. Findings around shelter policies and operations, which outline the current strengths and areas of concern in the shelter system.

System Overview

Introduction to Shelter | The fundamental purpose of shelter is to provide a safe, clean, and dignified place for people who would otherwise be unsheltered to meet their most basic needs while they search for stable housing. Shelter is a short-term, emergency resource. People are still considered to be experiencing homelessness while staying in shelter.

San Francisco’s Shelter System Operations | Most shelters in San Francisco are funded by HSH and operated by nonprofit service providers. HSH staff monitor, evaluate, and provide technical assistance to shelters, while nonprofits manage day-to-day shelter operations.
Each shelter client is entitled to baseline living conditions, social services, and amenities. In addition, clients have access to a range of services on-site, depending on the shelter. These commonly include case management, physical and behavioral healthcare, and benefits enrollment assistance.

Size of the Shelter System | In September 2024, HSH funded 33 shelters with 3,228 beds/units operated by 20 different nonprofits. The adult system is substantially larger than the other systems, representing over 87% of all shelter beds/units in San Francisco, and has expanded since 2022.
Shelter sites are clustered in the eastern part of the city, primarily in the Tenderloin, with sites in SOMA, the Mission, and Bayview-Hunters Point.

Comparisons to Peer Jurisdictions | This report examined 12 peer jurisdictions that were chosen for their similarity to San Francisco.

• San Francisco operates more homelessness resources (shelter and housing) per capita than most peers.
• San Francisco allocates a larger proportion of total bed inventory toward permanent housing than most peers and the national average. It allocates a smaller proportion toward shelter. Most peer shelter systems don’t have enough beds to accommodate everyone who is unsheltered.
• San Francisco reports serving the highest share of shelter clients with severe mental illness or chronic substance abuse among peers, though it is unclear if this is a true difference in client population or reflective of the City’s emphasis on diagnosis and treatment.

Number of Shelter Clients Served | The number of people served through the shelter system has consistently increased since 2021, from 6,859 to 9,913 per year. This increase in clients corresponds to both an increase in shelter capacity and an increase in the number of people experiencing homelessness between 2022 and 2024.

Shelter System Demographics and Equity Analysis | In Fiscal Years 2023 and 2024 (FY23, FY24), most shelter clients were Black (28%) or White (27%). The majority of clients in the family shelter system were Latine/Hispanic (61%), while the majority of clients in all other systems were not (22%).

More men than women are homeless. Those with marginalized gender identities—transgender, non-binary, questioning, or other gender identities—are much less likely to be in shelter than cis-gendered people experiencing homelessness.

A large proportion of clients identify as having a disabling condition (44%) or substance use disorder (30%), despite a significant amount of missing data for both conditions.

Shelter System Budget | Approximately 25% of HSH’s budget went toward shelter in FY24, or roughly $176 million. The shelter budget relies significantly on one-time funding sources, including state grants.

Actual Spending by Population and Shelter Types | In FY23 we estimate that, on average, Adult and Transitional Age Youth (TAY) programs cost $126.25 per filled bed/unit per night while family shelter costs $221.81. Non-congregate shelter was costlier than congregate, and crisis intervention programs were more expensive than emergency shelter or navigation centers. Exact spending is challenging to assess due to the structure of contracts and payment data.

Findings: Client Experiences and Outcomes in Shelter

Client Safety | In focus groups, shelter clients generally viewed safety as adequate, though theft of personal belongings was an issue. Female focus group participants reported more safety concerns than their male counterparts in adult congregate shelters. HSH requires shelters to have safety or security personnel on-site during operating hours, and has a number of policies to maintain safety, including a weapon check policy. Still, safety incidents occur on a regular basis. Between October 2022 and May 2023, shelter providers reported an average of 37 threats or acts of violence per month across the entire shelter system.

Living Conditions | Most clients noted the facilities were as clean as they could be. Many shelter providers attempt to create welcoming environments for clients, but shelters typically lack the same privacy and comfort as housing. Some clients wanted additional social programming at their sites. Clients also noted dissatisfaction with the quality of shelter meals, specifically frozen meals. This represents a trade-off, as HSH has moved toward flexible mealtimes and procured more frozen meals that can be reheated on demand for clients whose schedules may not overlap with traditional mealtimes.

Overdoses and Deaths | Overdoses and overdose reversals occur frequently in shelter, particularly in shelters serving adult populations. The shelter system takes numerous precautions to prevent fatal overdoses.

There is a high level of oversight and safety precautions in shelter, especially in congregate settings where there are many people around, which creates a safer environment for drug users. This allows staff to quickly catch drug overdoses when they happen and step in to reverse them.

Death as a reason for exit from shelter has decreased significantly in the past four years. This trend may partially reflect impacts from the COVID-19 pandemic and aftermath.

Patterns and Length of Shelter Stays | While shelter is an emergency intervention rather than a housing solution, almost half of stays are for longer than one month. Median length of stay was highest during the COVID-19 pandemic when the City removed any limits and managed a large number of Shelter-in-Place Hotels. Length of stay has decreased over the last two years as the overall system has worked to return to normal functioning. The average adult shelter bed served 3.5 people per year. Over the last four years, over half of people who appear in the shelter system have a single stay.

Shelter Client Housing Outcomes | Over half of clients don’t have a record of where they went upon leaving shelter, which limits the conclusions we can draw about shelter outcomes. This varies widely across shelters, with some sites reporting known exit destinations for over 90% of clients and others for just 5%.

Clients experience mixed outcomes even in available exit data. Across all clients in the reporting period, only 13% exited to permanent housing. This was higher for families and TAY than for adults. Generally, clients in non-congregate shelter were more likely to exit to permanent housing than clients in congregate shelter.

One of the reasons for the low exits to permanent housing may be the limited availability of permanent housing options both within and outside of the homelessness response system. Providers regularly reported that they struggle to figure out how to help clients if they don’t qualify for Permanent Supportive Housing (PSH). The majority of people who are assessed via Coordinated Entry do not qualify for PSH.

Subsidized housing programs outside of the homelessness response system often have long waiting lists, and clients may not be eligible for certain local or federal housing programs due to lack of income, immigration status, or criminal background. Private rental units are likely out of reach without a rental subsidy.

In the future, HSH should evaluate the efficacy of case management services to better understand whether the investment in them is impacting client outcomes. In addition to case management, both providers and clients expressed a desire for more wrap-around services to meet clients’ needs.

Equity in Experiences and Outcomes | Client focus groups revealed some differences in perceptions of treatment by race, but limited data makes it difficult to draw strong conclusions. Latine or Hispanic and monolingual Spanish speakers were somewhat more likely to report that they felt shelter staff were not empathetic or supportive. Congregate shelters were more likely to serve Latine or Hispanic clients, while non-congregate shelters were more likely to serve Black and White clients. This difference could be attributable to other characteristics, but HSH should explore additional data to assess whether the shelter system is serving Latine or Hispanic clients equitably.

Few differences in exit outcomes exist by race or ethnicity, and differences may be attributable to other client characteristics. White and Native American clients are slightly more likely to exit into unsheltered homelessness than other racial/ethnic groups.

There is some evidence of inequity by gender, and HSH should continue to explore ways to support female clients.

Findings: Shelter Policies and Operations

Partnership Between HSH and Providers | Day-to-day collaboration between HSH and providers is going well. Shelter providers interviewed universally reported positive experiences with their HSH Program Managers, who they viewed as highly responsive, problem-solving partners.

Maintaining fair and comprehensive shelter policies that satisfy all stakeholders is challenging, and requires HSH to balance competing goals and priorities. For example, maintaining low-barrier shelters sometimes conflicts with the goal of providing safe, clean, stabilizing environments for all clients. Both shelter clients and providers expressed frustration with the application of current shelter rules, and had specific feedback about policies they’d like to change.

HSH should continue work to make monitoring more comprehensive and outcomes-focused. All shelter contracts include performance measures, but many track inputs and outputs, not outcomes the City cares about improving. Many shelter contracts only include a single outcome measure, which tracks client satisfaction with services on site. HSH should continue work to develop simple, consistent, and meaningful performance measures as part of its Performance Measurement Plan.

Resource Challenges for Providers | Both shelter providers and HSH staff noted a significant gap between people’s expectations of shelter and the resources allocated to meet those expectations.

During interviews, providers continually said they lacked adequate resources and authority to manage street conditions around their site. The most common concern was a lack of staffing. Providers said when they could find staff, the City’s Shelter Grievance Policy limited their ability to enforce consequences for most behaviors outside the building.

Providers also frequently noted challenges supporting high-need clients. Some clients require skilled nursing, social work, and/or therapy at levels that shelter is not resourced to provide. HSH has explored senior-specific shelter models which could provide more active and concentrated supportive services for populations most likely to have physical disabilities, however there is no confirmed plan or timeline to offer these models at this stage.

Shelter providers and HSH staff noted that HSH’s Harm Reduction policy is working well for getting drug users into shelter and engaging with services. However, they felt that Harm Reduction needed to be paired with treatment options for clients who want them. HSH is currently running a 20-room pilot program with the San Francisco Department of Public Health (DPH) to provide immediate shelter, and access to prescription addiction medication and residential treatment.

Data Availability and Quality | There are significant challenges in working with available data that make it difficult to assess the impact of services, or answer key questions about how shelter is functioning. Half of exit destinations in FY23 and FY24 were recorded either as missing or “other”, and some demographic information, particularly reported disabilities or substance use disorder, have large numbers of missing reports.