CPCs, CPAs, and other ad costs up but hiring and postings down

Andrew Flowers, chief economist of AppCast, Inc., was the guest for the April 2026 Job Board Leaders’ Monthly Roundtable, an online gathering of dozens of leaders of job boards globally.

Andrew walked us through data covering hundreds of millions of job ads, millions of employers, thousands of job boards, and a couple of dozen countries. The data varied by segment but, on balance, showed that hiring was down last year yet, at the same time, costs to advertise were up by about 22 percent.

Listen in as the group collaborative and collegially share what they’re seeing, ask questions, and get those answered.

AI-generated notes:

Introduction of Andrew Flowers and AppCast: Steven Rothberg introduced Andrew Flowers, the Chief Economist at AppCast, who was invited to discuss AppCast’s research, specifically a counterintuitive finding (00:00:00). Andrew Flowers, who has been with AppCast for about four and a half years and leads their data team, previously worked as an economist for Indeed (00:02:29). AppCast, founded in 2014, is described as the leading recruitment marketing platform globally and the largest buyer of job advertisements on major job boards like Indeed, ZipRecruiter, and LinkedIn (00:03:37).

AppCast’s Business Model and Global Footprint: AppCast utilizes technology to programmatically distribute open jobs across many job boards for their thousand-plus clients, which include direct employers and agencies. AppCast expanded its products beyond programmatic job board channels to include search, display ads, and career sites through the acquisition of Bayart Advertising (00:03:37). AppCast’s parent company is the Stepstone Group, and their market presence is primarily focused on North America, United Kingdom, and European Union with lesser presence in Asia, Africa, and other markets (00:05:00).

Context of the US Labor Market Softening in 2025: Andrew Flowers provided context on the US labor market, noting that hiring declined and the labor market softened in 2025, becoming more favorable for employers (00:07:08). The ratio of US job openings to unemployed people has cooled substantially, moving from a 2:1 ratio to 0.909, meaning there are only 99 job openings for every 1,000 unemployed people (00:08:12). This softening is reflected in Indeed job postings, which declined by 16% over two years leading up to January 2024 (00:09:05).

Measuring Job Postings and Labor Market Decline: Steven Rothberg inquired about how AppCast counts active job postings; Andrew Flowers explained that they deduplicate job postings daily, excluding expansions or title changes, to get a unique job posting count posted on the Indeed platform (00:10:02). Overall, whether looking at job openings from the Bureau of Labor Statistics, job postings on job boards, or actual hiring, all measures indicated that the labor market declined in 2025 compared to 2024, with fewer job openings and a slower growth rate of employment (00:12:23).

Low Hire, Low Fire Economy: The US economy is characterized by a low hire and low fire environment, with the hire rate declining substantially to its lowest in over a decade. Andrew Flowers clarified that this is not due to a deep recession with a big spike in layoffs, as the layoff rate remains historically low; instead, it is driven by slow hiring, making it difficult for job seekers to find employment (00:13:25).

Recruitment Marketing Metrics and Counterintuitive Findings: Despite the slow hiring and softening labor market, Andrew Flowers revealed the counterintuitive finding that recruitment marketing metrics, including costs, rose in 2025. The cost per click (CPC) rose by 22% in 2025, which Andrew Flowers attributed to a composition effect where advertising spend concentrated into hard-to-fill occupations like healthcare, which command higher bids (00:17:23).

The Scarcity Game and Hard-to-Fill Roles: The explanation for the rising costs is that employers shifted their programmatic spend to roles that are difficult to fill, while using organic channels for easy-to-fill positions, especially white-collar roles (00:18:24). Healthcare application costs, for instance, are at $35 per application, while costs for other occupational groups declined. Andrew Flowers concluded that the job board game is less about scale and volume and more about the scarcity of specific, hard-to-fill roles (00:19:35).

Outliers in High Cost-per-Application (CPA) Roles: The occupational groups showing the highest CPAs were identified as healthcare, education (including teachers and administrators), and science and engineering (including AI-related roles) (00:21:38). These roles tend to be non-automatable, in-person service delivery jobs, while the science and engineering category is related to the AI boom, requiring hard-to-hire talent like data scientists (00:22:32).

Year-over-Year Changes in CPCs and Conversion Rates: Regarding year-over-year changes in costs, Andrew Flowers explained that for some white-collar roles, the decline in CPA was often driven by an “avalanche of new people applying” and expanded conversion rates, even if bidding remained stable or slightly increased (00:23:36). Conversely, in other roles, the decline in CPA was actually caused by weakened demand and a decline in basic bidding (00:24:38).

Job Seeker Conversion and White-Collar Fear: Data on job seeker conversion rates from 2024 to 2025 showed that white-collar fields like legal, technology, finance, and consulting had the biggest jump in applicants. Andrew Flowers suggested this is due to job seekers in these roles being fearful of job security and using mass AI tools to apply (00:25:25). Roles with declining conversion rates included healthcare, hospitality, transportation, and education (00:26:19).

Global Labor Market Trends and AppCast Data: Steven Rothberg asked if the US trends were consistent globally; Andrew Flowers presented data from over 15 countries, primarily in Europe and North America, showing varied cost-per-application (CPA) changes across regions (00:28:30). Germany, Austria, and Poland showed big declines in CPA, while the United Kingdom held steady, indicating that the labor market story is not uniform across all countries (00:29:38) (00:31:35). AppCast’s data in many of these international markets, particularly where pay-for-performance is less common, is administered through the Stepstone Group (00:30:36).

Early Career Hiring and Macroeconomic Factors: Gerry Crispin raised a question about the extreme slowdown in hiring for early career candidates, hypothesizing that AI and automation might be a key issue (00:33:42). Andrew Flowers provided four points, arguing that the slowdown is mostly driven by macroeconomic factors like high interest rates, which disproportionately affect young people (00:40:05). Steven Rothberg agreed that the lack of turnover (the “great hug”) is a more significant factor than AI in preventing early career candidates from getting hired (00:37:37).

AI Impact and the Shift to Trades: Andrew Flowers acknowledged that AI has some impact, noting that young people in highly exposed occupations (finance, software development) are struggling to secure jobs following the introduction of large language models (00:42:15). However, Andrew Flowers emphasized that it depends on the job, as the headcount for young people in jobs unexposed to AI, such as physical therapy, teaching, and the skilled trades (HVAC, plumbing), is booming (00:43:20).

Spend Moving Away from Job Boards: Ben Groves asked if spend is moving from job boards to other sources like social or display. Jason Gorham provided anecdotal evidence of job boards shifting their models to display pay-per-click (00:47:34). Andrew Flowers stated that competitors to job boards are search/social channels (which are complementary for niche roles or Gen Z workers) and large language models (LLMs) (00:50:01).

Long-Term Competition from Large Language Models: Andrew Flowers noted that the competition from search and social channels is not existential due to the high-intent audience on job boards. The existential competitor is LLMs (like Claude or ChatGpt) moving into the job search channel, which would require an exchange layer to distribute jobs across models, though Andrew Flowers believes this threat is not immediate for 2026 but could be significant by 2030 (00:51:01).

Conclusion and Next Meeting: Andrew Flowers shared their email in the chat for further specialized data inquiries, such as regulatory and compliance data (00:56:03). Steven Rothberg thanked Andrew Flowers for their informative presentation and announced that the next meeting’s guest would be Tim Dineen on May 14th at 11 a.m. US Eastern and those who want an invite should email him at Steven@CollegeRecruiter.com.