As AI Advances, Candid Cuts Staff, Bets Its Future on Data

Nonprofit data provider Candid is undergoing a major reorganization and has laid off 42 people, about one-fifth of its work force, to try to hitch its future to the fast-paced growth of artificial intelligence.

The nonprofit will focus on data collection about the nonprofit sector, beyond data provided by the federal government, in an attempt to ensure facts and figures produced by AI engines are accurate and timely.

Created from the 2019 merger of the Foundation Center and GuideStar, Candid has experienced a decline in revenue over the past several years that its CEO, Ann Mei Chang, ascribes to dwindling interest among foundations in philanthropy “infrastructure.” She said the nonprofit has also been buffeted by a drop in the number of users for its products, such as Foundation Maps, a data visualization tool, and its search functions that provide snapshots of foundation giving.

Candid’s projected 2025 revenue is at its lowest mark since 2018, at $38.5 million, down from a high of $55 million five years ago.

After nonprofits were required to file their returns to the IRS digitally, organizations like Candid that provided search products based on the filings became less vital, as competition emerged from other data projects that began to provide free filing searches, said Nathan Dietz, a professor who researches nonprofits at the University of Maryland. 

“Now that all filing nonprofits have to file their 990 returns electronically, a lot of organizations have needed to make changes to their business models,” he said.

For Chang, it is the rapid growth of AI and the increasingly fragile state of government-provided data that have been the drivers behind the shift in strategy. As more people turn to AI for information, they will only get an accurate picture of the state of the nonprofit sector if the data provided to AI platforms is professionally vetted, Chang said. 

“AI is only as good as the data it’s trained on,” she said. “If we’re going to be able to leverage AI effectively for the sector, we need it to be trained on good data.”

A Data Provider for AI

Chang envisions hiring more staff to focus on curating new data sets about the sector, such as figures on foundation grants supplied directly by foundations so they are truly up-to-date and scraping the internet rather than relying on outdated government data. 

Obtaining reliable data has been a challenge for Candid and other organizations and researchers. Over the past several years, some grants data has been mysteriously missing from Internal Revenue Service reports, which gives researchers and nonprofit executives an incomplete picture of foundation activity.

Chang and other nonprofit leaders have pressed the agency over the data gaps and expressed concern over whether a short-staffed IRS will continue to disseminate data on a regular basis. Other federal data collection efforts, such as the U.S. Department of Agriculture’s measurement on national “food insecurity,” were terminated during the first year of President Trump’s second term.

Even when there are no glitches in IRS data distribution, the nonprofit figures the IRS produces for the public are usually up to two years out-of-date. That is of little use to nonprofits, which are under a great deal of pressure to raise money due to federal funding cuts and increased demands for service. With an expanded data focus and current information gathered directly from grant makers, she said, Candid hopes to fill the gap.

“People need more current data in order to make good, relevant choices about where they’re going to invest, and how to coordinate with other partners,“ she said.

Five years from now, Chang predicted, Candid will provide data to all of the major AI engines as well as any next-generation AI products that emerge.

Newsletter to Move to Chronicle of Philanthropy

As part of the restructuring, Candid has discontinued its Philanthropy News Digest newsletter. Candid provided former subscribers’ information to the Chronicle of Philanthropy unless they opted out. Those who did not opt out will get the Chronicle’s Philanthropy Today newsletter.

Candid has also terminated its job site, which will be folded into the nonprofit employment site Idealist. Both arrangements were noncash barter agreements. Both the Chronicle and Idealist will promote Candid offerings to their users at no charge.

Under the deal, Chronicle content will be available to an expanded reader base. “These are people who are engaged in nonprofit work,” said Stacy Palmer, CEO of the Chronicle of Philanthropy in an interview. “They’re the kinds of people we want to reach.”

A Bet That Accurate Data Will Have Value

In an era when federal data is becoming less consistent and less reliable, other sources of data are increasingly important, said Cameron Hickey, president of the National Conference on Citizenship. Like Candid, the conference is seeking to produce its own data sets, anticipating further interruptions at the federal level. 

“We need to have a resilient public data infrastructure,” he said, “one that is not at risk of the whims of one political administration or another.”

In the age of AI, Hickey said, it is possible for someone to “overload the system” with bad information. If a data provider like Candid could provide access to the best quality information over AI platforms, it could benefit nonprofits and the broader public — and potentially, Candid’s bottom line. 

“It’s a really critical thing to do for the community that depends on that data and a smart business opportunity if they’re able to make money off it from those gigantic, multibillion dollar companies” that manage AI systems, he said.

Exploring New Models for an AI Future

Candid, like every other group, is trying to figure out how best to navigate a future with more people relying on AI. 

In December, Candid inked a deal Chang calls an “experiment” to integrate its data with Anthropic’s Claude AI assistant. Users will get verified Candid data on the nonprofit sector and will be directed to Candid, where they can access additional data with a paid subscription.

Agreements like the one with Anthropic protect Candid from having AI platforms simply “snarf up” its data, Chang said. But the deal doesn’t yet guarantee a revenue stream for Candid, which has provided its data for free.

Chang believes AI platforms and end users will both pay a premium to access data that Candid has procured on its own and for federal data that they have analyzed and interpreted. “Nobody’s quite figured out how all this is supposed to work,” Chang said. “Yet in the ideal world, the AI engine should be paying for copyrighted material, which includes our data.” 

Corrections: In a previous version of this article, Ann Mei Chang’s name was misspelled in a caption. Also, the piece said the merger that created candid was between GuideStar and Foundation Source instead of the Foundation Center. It also misstated how Philanthropy News Digest subscribers will be transferred to the Chronicle of Philanthropy. Candid provided former subscribers’ information to the Chronicle of Philanthropy unless they opted out. Those who did not opt out will get the Chronicle’s Philanthropy Today newsletter.