Big Data to Fight Poverty, Part 2: What We Learned From Our Data (so far)

Recommendation: Read Part 1 in the three-part series for important context before taking in this one.

In this post I want to share a glimpse of what is possible when we put big data to work to fight poverty. First, let’s get clear about where all this data came from.

The One Degree from Alluma platform has been used by over 500,000 people searching for social services or resources, as well as by the professionals who support them (case managers, social workers, etc). To access the platform, our community uses One Degree’s iOS or Android app, our website at, or our text-to-search SMS tool. On the platform, individuals are able to find, manage, save, and track social services and community resources. We call our platform users “members” of our community. Our sharing features enable members to add ratings and reviews, and share resources with friends and family by email, Facebook, or text message. We also encourage the community to contribute to the platform — anyone can report errors or add and update resources they know about. Professionals are also able to use the platform to make referrals to their patients or clients.

One Degree is optimized to help a person find exactly what they are looking for — such as a food pantry in a specific neighborhood or a homeless shelter that serves LGBTQ individuals — as quickly as possible. (If you want to dig into this more, we have an informational website about our taxonomy and how we organize resource data.)

All the data included in this analysis was collected in one of three ways:

  1. By directly surveying community members;
  2. By capturing search behavior and demographic information of site visitors not logged in;
  3. By capturing search behavior and demographic information of logged-in members.

We do not require individuals to have an account on One Degree just to search for help. People who do sign up for an account are able to access richer features, such as saving resources for later, tracking their outcomes, and getting reminders — all for free. Because logged-in members provide the richest data, we focused most of our analysis on a representative sample of the more than 19,000 members of the community who have One Degree accounts. We have anonymized and aggregated results for the purpose of this analysis. (It’s also worth pointing out that we store all personally identifiable information on a HIPAA-compliant database.)

Here are examples of the kinds of data we captured on our platform:

  • Devices used and location (usually zip code)
  • Categories searched, and over how many sessions
  • Keywords typed into the search box
  • Frequency and time of year of searches

We log basic information in the background during platform searches including location (when available), timestamps, and device used. As many in the tech industry know, this kind of data isn’t perfect. For example, some security-conscious users adjust browser settings to block collection of location data, or they may be using a proxy. By and large, however, the data in the aggregate is reliable for identifying trends and insights.

What we found

Most low-income people have a broad spectrum of needs at any given point in time.

We found that most people search for a diversity of resources within a single search session or across multiple sessions.

  • Of those that searched in the Employment category specifically, and then searched for additional resources, 77% were looking for resources outside Employment, in categories like Food, Health, or Housing.
  • In an average month, people search for 4.4 different resource needs
  • 66% of people search for more than one resource need and 28% search for 5 or more resource needs:

Most people using One Degree have a variety of needs across multiple categories.

Real profile example: A 25-year-old Spanish speaking woman in the Mission District of San Francisco downloaded One Degree’s Android app and then within the week she:

  1. Searched for homeless support, apartment waitlists, and tax preparation help
  2. Rated and reviewed a tax preparation resource
  3. Searched for books & supplies

We did not find a significant difference between searches by professionals, such as social workers or community navigators, and searches by low-income individuals or families.

This really drove home how ill-suited the current safety net is to meet the scale of the problem facing low-income individuals and families. The safety net isn’t really organized to help people who experience hardships in more than one category at the same time.

There are thousands of nonprofit organizations and government agencies in our communities that, by-and-large, focus on meeting only one specific category of need. For example, a San Francisco mother with a child might be best served by WIC SF for food, emergency shelter from A Safe Place, and prenatal care from the Women’s Community Clinic. But to find each of those resources, she has to navigate the landscape of hundreds of organizations and providers to figure out which can or cannot meet her needs. This means she’d have to research multiple agency websites, online PDFs, review paper pamphlets, or even visit or call the offices.

Families can spend as many as 20 hours per week just trying to make ends meet with services provided by multiple organizations — it basically adds a part-time job to their already hectic lives.

This is the challenge low-income families face that One Degree is trying to help solve. Our goal is to be the one place available anytime and anywhere to meet a diversity of specific needs.

We can learn about real-time needs at the regional and neighborhood-level

Because One Degree has served the San Francisco Bay Area since our launch, we have a longer search history from individuals and families searching for resources there. Looking at this dataset, we are able to zoom in to the region, and in some cases even to the neighborhood level.

Housing is the Most-Searched Category in the Bay Area

Unsurprisingly, Housing is the predominant resource area searched by low-income people in Bay Area communities, representing more than 27% of all searches, that’s nearly twice as many as the next highest resource area (Family & Household, which includes things like baby supplies or cash for utilities):

Bay Area housing searches on One Degree

One example of a person using One Degree for help with housing is Seisha, mentioned in Part 1 of this series. Seisha is a young mom of two young children living in East Palo Alto in Silicon Valley. When Seisha faced a major rent increase that threatened her family’s housing situation, Seisha looked on One Degree for helpful resources.

“I was able to go to the site, look up ‘rent increase rights’ and be led to the right resources to find out what my rights as a tenant are. … It’s nice to go to the [One Degree] site and have options for good nutrition, health, and hospital resources. It’s a one stop shop that makes it easy for families and parents.” — Seisha

With two active girls and attending school herself, Seisha doesn’t have a lot of time to navigate the web, visit offices, or make calls to find what she needs for her family. With the sky-high cost of living in the Bay Area, Seisha found resources on One Degree that helped her stay in her home and make ends meet without a major time investment.

We also zoomed in to look more closely at searches in “Housing” to see what we could learn:

We saw an above average correlation between those looking at resource category “Housing search” and those looking at resource category “Affordable home ownership,” which was also in the top ten most popular housing-related searches. To us, this suggests that home ownership is still seen by many as a way to find stability in their housing situation, despite the fact that with extremely inflated home prices, it is largely an unattainable goal for most low-income — and even middle-income — families in the Bay Area.

At the Neighborhood Level

We were also able to identify trends at the neighborhood level that map to some of the known realities facing low-income individuals and families in those areas.

In San Francisco’s Bayview district, an outer-ring neighborhood with known transit challenges, “Transit passes & discounts” surpassed “Housing search” as the top resource need.

In San Francisco’s SOMA neighborhood, which has the highest concentration of San Francisco’s homeless individuals, “Section 8” is the top search, followed by food services.

A person searching for “Affordable housing” in SOMA is much more likely to also be searching for “Homeless support services” compared to other neighborhoods.

San Francisco’s Mission District is home to a population living in large tent encampments, and we found that “Shelters” is by far the top search.

The Mission also has an especially large Latinx population. More people here are searching in the “Immigration law” and “ESL classes” categories than in other San Francisco neighborhoods.

We see needs change with the seasons

It’s not surprising that we see needs change with the seasons as indicated by changes in search behavior. While not new, the seasonal changes provide validation that One Degree’s data is accurate and reliable.

Throughout an entire year’s searches, “School supplies” doesn’t make the top 20 most searched categories, but in August it surges to the eighth most-searched category just behind “food” and “food pantries”.

Much of the year, the “Meals” category isn’t in the top 15 most-searched categories. However in November, the “Meals” category is the №1 searched category.

Likewise, “Toys” doesn’t even make the top 20 categories overall in a year’s worth of search data, but in December “Toys” rises all the way to the third most-searched category.

Nomenclature matters

When it comes to communicating with low-income families, are we understanding one another? By noticing what terms people are actually using to look for help (or not using), we can apply that knowledge to make sure we’re communicating clearly.

One finding we found interesting was about CalFresh. CalFresh is the largest food assistance program in the nation, serving over 4 million low-income Californians annually.

It was surprising to discover that “CalFresh” was searched only a trivial number of times on One Degree by people typing in their own keywords, compared to the term “food stamps,” which was the fourth most searched food-related term.

This indicates that “food stamps”, the colloquial name for the California SNAP program, remains resonant and popular, whereas “CalFresh” is largely a term used by professionals. We intentionally use the term “Food stamps” instead of “CalFresh” in the One Degree platform’s search categories for this reason.

We also found that “Section 8” as a more widely searched resource need than “Housing vouchers” or “Deposit & rent assistance” despite the very limited availability of Section 8 vouchers. This tells us that “Section 8” is still a widely recognized term that seems to have become synonymous with ‘housing assistance’ in general.

The One Degree team is continuously applying this kind of information to new product features and updates to our taxonomy. As more and more organizations and families use One Degree, the dataset becomes richer. It’s true there are proprietary databases or standards out there that purport to organize social services information that serves organizations’ needs. Here at One Degree, our standard is centered on reducing barriers to low-income communities who are looking for the services they need as quickly as possible.

We don’t use practitioner jargon; we want to meet our community members where they are.

How might the social services sector more broadly apply this kind of data to its work serving low-income families? I tackle that question in Part 3 in this series: The Big Opportunity.

This data analysis was supported by a grant from The James Irvine Foundation and from the California Endowment.

Andrea Wood is the former Head of Development at One Degree. Previously, Andrea worked as Director of Advocacy and Fundraising at Mozilla and as Senior Director of Client Services at