Humanly Bought CareerArc. The Staffing Business Is Next.

The recruiting industry has spent years selling employers software that helps them do more work. Humanly would rather sell them the candidates.

That’s the more consequential, if somewhat buried, story behind Humanly’s acquisition of CareerArc. The announcement describes recruitment marketing meeting AI-powered hiring, which sounds sufficiently familiar to disappear into the usual acquisition coverage. Another platform gets bigger; another funnel gets consolidated. Somewhere, someone updates a sales deck and messaging framework.

In our briefing, Kumar described a much more explicit ambition: building a technology-enabled staffing business, with AI doing more of the work and employers buying access to qualified, engaged candidates.

CareerArc gives Humanly another way to find those people. Social is one channel in that strategy, and an intermediate step toward a business that doesn’t have to buy a fresh introduction every time a customer needs to hire. The immediate target is job advertising spend. The longer-term ambition, Kumar said, is to compete for budgets going to companies like Manpower and Adecco. That should get considerably more attention than the integration roadmap.

Indeed, We Have a Problem

The backdrop is employers’ dependence on Indeed, and what happens when a source they’ve treated as infrastructure changes the terms of access.

Kumar described heavy Indeed dependence among Humanly’s target customers and declining applicant volume as organic distribution contracts.

He also acknowledged an important limitation: Humanly doesn’t yet have a complete picture of how those traffic changes translate into hires. Fewer applications and fewer hires are related problems, but they aren’t interchangeable metrics.

For precision, Indeed hasn’t eliminated every free listing. Its March 31 policy removed free visibility for certain single-source XML feeds when jobs could instead be supplied through an integrated ATS. Other organic routes remain.

For employers, however, the larger lesson doesn’t require a seminar on feed architecture. If one company controls access to most of your candidate flow, your recruiting strategy is partly dependent on its commercial decisions. Kumar isn’t suggesting customers can simply stop sponsoring Indeed jobs; they’ll still need that traffic. He wants Humanly to develop enough alternative supply that every hiring requirement doesn’t trigger another trip to the same checkout.

That means investing in employer brand, social distribution, direct engagement, and Humanly’s own job seeker product. Diversification becomes part of delivering the service, rather than another homework assignment for an already stretched talent acquisition team.

Social Gets a Job, Finally

CareerArc brings HireSocial and Job Maps, giving Humanly social recruiting capabilities and location-based job discovery. Kumar said the companies share roughly 10 customers, with the exact overlap potentially higher. He was measured about social’s contribution. He described traffic shares around 15% to 20% among those shared customers, rather than positioning social as a channel that would suddenly supply a third of demand. Those are traffic figures, not demonstrated sources of hire.

Social offers enormous reach: Pew’s 2025 survey found that 71% of U.S. adults, for example use Facebook and half use Instagram. It doesn’t follow that these users opened either app hoping to complete an employment application. As a standalone, immediate source of hire, social has an intent problem. Someone searching for a warehouse job has already made a decision that someone scrolling past a warehouse ad hasn’t. More impressions won’t necessarily close that gap. Humanly’s opportunity exists in the delta between generating impressions and driving applications.

Kumar described employer-branded pages and engagement experiences that give people a clear next action, alongside broader advertising for Humanly’s own job seeker product. That product is being built to coach candidates, engage them, and help them apply.

The two complementary strategies serve different stakeholders: one matches a candidate against a specific role at a particular employer, while the other can establish an ongoing relationship with Humanly, potentially making that candidate easy to engage for future opportunities.

Social doesn’t have to become the best direct source of hire to be helpful for hiring here; it simply has to produce enough relevant, available candidates for the engagement and qualification process to make sense economically. Compared to the pricing models of Indeed and LinkedIn, undercutting the competition, particularly for high volume workers, shouldn’t be particularly difficult.

RPO Is Now A Tech Play.

Kumar described an interim model that resembles recruitment process outsourcing: Humanly’s own team uses the tools, manages channels, and builds volume for customers. That significantly increases the scope and significance of the CareerArc acquisition. According to Kumar, the plan is to use CareerArc to support Humanly’s pipeline services, not just as a standalone product; in this model, social becomes one of several channels Humanly’s internal team leverages to deliver candidates.

There’s a decidedly unsexy (but commercially important) distinction between giving a customer access to a recruitment marketing tool and taking responsibility for actually using the tool. The first adds something to the customer’s workload. The second can remove a ton of work, provided the supplier actually delivers as promised.

Kumar described multiple ways customers could buy the combined offering, including an ATS with pipeline as a service or simply a qualified candidate list. Humanly would handle the front end of finding and qualifying people, with partners supporting preboarding and onboarding.

The ambition extends beyond selling another subscription to a large recruiting department. Humanly wants to supply more of the recruiting capacity that department would otherwise need to operate or outsource.

That puts it in a different budget conversation; a technology purchasing decision is obviously made when it’s being evaluated against other software. A service that produces usable candidates, on the other hand, can split allocations between recruitment advertising costs, recruiter bandwidth, and agency fees.

The spreadsheet gets more interesting when the alternative is paying someone else to do the whole job.

Talent Liquidity: Size Doesn’t Matter.

The phrase Kumar returned to was “talent liquidity”: having enough engaged, relevant people available that employer demand can be met without rebuilding the candidate pool every time.

His Las Vegas bartender example makes the idea concrete. Hiring activity for an employer such as MGM could help establish a deeper understanding of local candidates, their experience, and their availability. With the necessary permissions, future demand could be met by reengaging suitable people already in the network.

The real revenue opportunity would come from repeatedly serving demand within particular occupations and locations. A national database with millions of records is less useful to a single manager than a reachable group of qualified people who can actually work – and are actually available for – an open shift on the schedule.

Kumar said 62% of Humanly’s partners permit retargeting candidates to other companies. That’s a company-reported figure, and employer permission still has to be matched with appropriate candidate consent and data rights. An application to one company doesn’t automatically authorize a recruiting relationship with every other company in the network.

The operating challenge is equally important. Availability changes, as people accept jobs, move between markets, change availability and scheduling preferences, or simply stop responding. Maintaining liquidity means, in short, maintaining relationships, not preserving old screening results in a searchable archive.

Kumar described quarter-over-quarter growth in ready-to-work reach as a measure of success. The useful test will be whether that growth makes it faster and less expensive to deliver qualified candidates who actually progress.

Otherwise, the industry has invented another impressive database nobody wants to call.

Staffing Should Watch the Invoice

This is where AI-native services become a competitive value proposition.

A traditional staffing business needs people to acquire candidates, contact them, establish suitability, coordinate next steps, and maintain the relationship. Humanly is assembling a model in which software performs more of those repeatable tasks, supported by an operating team and partners.

Whether that produces better margins and outcomes remains to be seen. Qualification isn’t the same thing as a start, and an start isn’t the same thing as retention. Humanly will need to show that its candidates meet specific requirements and can convert into the kind of workers employers want to keep. Delivering a pipeline also doesn’t replicate everything a traditional staffing firm can provide to their clients. Things like payroll, employment responsibilities, working capital, and onsite operations remain substantial components of temporary staffing spend.

But Humanly doesn’t need to replicate every service before it starts taking spend. First, it can reduce how much customers pay to replenish candidate traffic. Then it can compete for sourcing and screening assignments. Over time, a dependable candidate supply can diminish the reasons an employer calls an agency at all.

There’s a risk for employers, too. Trading dependence on Indeed for dependence on a closed candidate network could reproduce the same problem under a different logo. Buyers should understand access, portability, consent, and what happens to candidate relationships when a contract ends.

Still, Kumar’s direction is clear. CareerArc is an input into a business selling recruiting outcomes, with social helping build the candidate supply that makes those outcomes possible.

The staffing industry’s next competitor may arrive with an ATS and an AI interviewer. The threat becomes real when the customer stops paying for access to the software and starts paying for the people it delivers.