The State of Play

AI-powered candidate sourcing and outreach automation has solidified at established tier: 87% of organizations use AI in recruiting, with proof-of-concept efficacy now routine (3x faster sourcing, 40% time-to-hire reduction, $800/hire savings documented across independent deployments). The practice is operationally viable for volume hiring and cost reduction when human-in-the-loop controls are enforced. However, the 2026 inflection point reveals a massive credibility-versus-reality gap. Production deployments are real (Imast's 2026 report shows agentic sourcing agents surfacing 2,300 qualified profiles in 72 hours; Pin's 2026 user survey validates 95% quality improvement and 5x outreach response rates across 210 customers; Adecco's June 2026 deployment delivered 1.2M candidate interactions with 50% time-to-deliver reduction and 80%+ fill rates), yet adoption-impact remains fragmented: only 39% of HR teams see significant business value from AI recruiting, 88% report no material ROI gain despite adoption breadth, and only 31% can defend ROI claims with rigorous measurement methodology. The 2026 verification is stark: ManpowerGroup's June survey found 90% deployed AI in hiring yet fewer than 5% report transformational outcomes; SHRM data shows cost-per-hire and time-to-hire both increased over the three-year AI adoption period—contradicting the core ROI promise. Volume-based outreach automation collapsed in 2026: Explorium's analysis documents phase-3 degradation with <1% reply rates and 8-14% bounce rates; signal-first architectures now required to maintain efficacy. AI-only cold email (4.1% reply rate) underperforms human-written (10.4%) and signal-based personalization (15-25%), validating human-in-the-loop and precision-targeted approaches over fully autonomous messaging. Autonomous outreach automation positioned as the 2026 breakthrough shows 5-10% hire rates in practice versus 50% in hybrid human-oversight models. Candidate-side resistance is hardening: 50.5% of job seekers are rejected silently without feedback, 63.8% blame AI for the experience, and only 9.7% were clearly told AI was involved—creating a transparency crisis that drives 31.4% candidate abandonment. Specialized recruiting (niche technical roles) fails spectacularly: AI tools return 71% noise on nuanced queries due to keyword flattening. The limiting factor is no longer capability maturity but the compounding costs of poor implementation: organizations must now pay for rigorous fairness audits, transparency infrastructure, multi-channel diversification, governance frameworks, and cognitive load management for recruiters—offsetting claimed efficiency gains. Those pursuing volume-driven, autonomous-first strategies are hitting hard walls on hiring dysfunction, candidate experience, and brand damage. Venture capital validation persists—Contrario, a YC-backed agentic sourcing platform, achieved $6M annualized revenue in 6 months with 150+ placements, validating unit economics—but deployment success remains heavily concentrated among enterprises. Mid-market organizations report 72% of sourcing and outreach AI investments fail to generate ROI, with implementation execution and governance rigor as primary constraints.

Vendor consolidation around SeekOut (750+ enterprise customers), Findem, and hireEZ continues as the category matures. Adoption breadth is high: 87% of organizations report using AI in recruiting; 46% have sourcing automation deployed (plateau from 45% in March 2026); 47% of HR professionals specifically use AI for sourcing; $3.77B market in 2026 projected to reach $5.5B by 2031 at 7.85% CAGR. Positive deployments remain documented: OIIS (independent IT staffing) achieved 3x faster sourcing, 40% time-to-hire reduction, 3x outreach response improvement (12%→31%), €800/hire cost savings in 18-month implementation; MokaHR (3,000+ enterprise customers, 30%+ Fortune 500) reports 2-3x faster hiring cycles, 34% time-to-hire improvement, 36% cost reduction, 87% AI-human screening consistency; Imast's 2026 report documents agentic sourcing agents in production achieving 2,300 qualified profiles surfaced in 72 hours (vs. 3 weeks manual). Agentic AI (autonomous agents handling end-to-end source-to-hire) emerged as dominant 2026 market trend, with 46% of companies planning deployment; Paradox's Olivia chatbot handles 100+ simultaneous conversations, screens in <48 hours versus 5-7 days, demonstrating capability maturity at vendor level. Sourcing time savings of 30-50% and time-to-hire improvements of 40-70% are routinely claimed; first-year TCO documented as $60K-$250K for enterprise with ROI payback within two quarters.

However, critical deployment gaps are now visible across 2026 evidence. Only 62% of organizations have AI tools in production; among those, only 38% achieve meaningful scale (running automation on >50% of relevant requisitions); only 39% of HR teams report significant business value from AI recruiting, indicating an adoption-value gap where breadth does not correlate with impact. Most critically, only 31% use rigorous measurement methodology to assess ROI—69% rely on vendor dashboards or recruiter sentiment, indicating a credibility gap. Autonomous outreach automation shows 5-10% hire rates in practice (AI-only models) versus 50% in hybrid human-oversight approaches; volume-based cold email campaigns specifically collapsed in 2026 (Explorium reports <1% reply rates and 8-14% bounce rates), requiring shift to signal-first architectures (firmographic fit, intent signals, trigger events) to maintain 15-25% reply rates. Outreach infrastructure became critical lever in 2026: email authentication (SPF/DKIM/DMARC) now mandatory per Gmail/Microsoft/Yahoo policies; domain reputation and sending hygiene determine inbox placement more than message copy. Candidate experience is degrading: 50.5% of job seekers received rejections with zero human feedback in the past year; 63.8% of rejected candidates attribute poor experience to AI; only 9.7% were clearly informed AI was involved; 31.4% abandoned applications due to AI screening. Specialized sourcing fails systematically—boutique firms report 71% noise rates on niche technical queries as AI tools apply keyword flattening, discarding critical query nuance. Governance remains unresolved: 85.1% of AI screenings favor white-associated names; regulatory scrutiny intensified (EEOC/DOJ exposure, state-level audit mandates CA/NY/CO/IL); 45% of organizations lack formal AI governance frameworks despite claiming transparency is important. Internal organizational cost is rising: AI automation concentrates cognitive load on recruiters by removing low-effort tasks, creating decision fatigue and burnout, reducing recruiter productivity despite claimed efficiency gains. Paradoxically, cost-per-hire and time-to-hire metrics have both increased over the past three years (the AI adoption period), indicating implementation execution remains the critical constraint. Talent acquisition budgets are flat (only 30% expect growth), signaling market saturation and deployment focus shifting from capability to implementation quality, fairness auditing, infrastructure hardening, and organizational change management.

August 2026 evidence reinforces the value-realization bottleneck. LinkedIn platform data (180K+ connections) shows platform saturation: connection acceptance held steady (26.3%→28.1%) while reply conversion collapsed 32.2%→22%, a 32% relative decline year-over-year, confirming that saturation drives response fatigue. A survey of 1,500 US business managers (iprospectcheck, July 2026) shows only 13.8% of AI-using organizations report significant hiring improvement; 19.7% cite lack of human review as their primary concern. Only 31.6% of organizations use AI in hiring at all—among those, most see minimal impact. Practitioner data reveals the reality gap: a 12-month study of 46,216 LinkedIn InMails (Recruitrx) showed response rates of 22.1% for cold outreach versus 45.8% for prior-contact scenarios—the 2x advantage of targeting over volume; a separate analysis of 844K recruiting sequences (Noon AI) confirms that 64.9% of replies arrive after the first message, meaning single-touch automation captures barely one-third of willing candidates. Signal-based sourcing (hiring signals, funding announcements, leadership changes) achieves 3–7% reply rates versus <1% for generic cold outreach—a 10x lift—indicating that precision targeting, not automation scale, drives outreach efficacy. The hard constraint is not technology but execution: practitioner data (Supersourcing RPO study, July 2026) confirms that mature agentic deployments reliably own only 60–70% of recruiter hours (not the 95%+ vendors claim); time-to-hire compression to 7–10 days requires 60–70% of the funnel to be automated and 30–40% deliberately kept human. Executive search firm perspective (Jennings, August 2026) documents the outcome ceiling: despite sourcing cost falling by an order of magnitude, senior hire failure rates remain ~30%—unchanged since 2018—because the bottleneck shifted from finding candidates to diagnosing fit, scoping roles, and making judgment calls. Organizations pursuing purely volume-driven automation are hitting hard walls; those succeeding are those combining signal-based targeting precision, multi-touch sequences, human-in-the-loop screening, and rigorous governance.

• 2019: SeekOut and Entelo emerged as category leaders in AI-powered candidate sourcing. Market consolidation began (Entelo-ConveyIQ). Industry remained divided: early adopters validated sourcing automation, but experts warned about algorithmic bias and immature AI capabilities in hiring.
• 2020: Candidate sourcing moved to multi-vendor commercial deployment. Phenom reported 100+ deployments with 95% lead increase metrics. Findem launched with $7.3M Series A. Critical assessment intensified: academic research confirmed AI sourcing tools replicate demographic bias from training data; fundamental fairness questions remained unresolved despite automation efficacy.
• 2021: SeekOut reached 75+ enterprise customers and raised $65M Series B. Vendor maturity was clear, but enterprise adoption remained patchy: surveys showed 95% of recruiters wanted sourcing automation, yet only 9% used chatbots and 91% lacked job personalization. Implementation gaps and bias concerns persisted alongside documented successes in tech and defence sectors.
• 2022-H1: Findem raised $30M Series B; SeekOut expanded features (Talent Rediscovery, AI resume screening, 700M+ profiles). Adoption intent surged: 92% of HR leaders planned to increase AI use, 58% of recruiting budgets allocated to automation tools. Real-world deployment in production workflows confirmed via user reviews, but bias risks persisted—Brookings research documented algorithmic reproduction of demographic bias, and NYC regulatory mandates for bias audits emerged. Category solidified around 3-4 mature vendors with broad enterprise deployment, though implementation quality and fairness outcomes remained variable.
• 2022-H2: European adoption reached 28% with 19% planning deployment; vendor-managed services offerings scaled (Findem Sourcing Accelerator, SeekOut Spot). Real-world case studies multiplied (Project Legion veteran sourcing coalition). Critically, academic and legal assessment hardened: Cambridge University study (October 2022) challenged vendors' bias-mitigation claims; BLG legal analysis flagged regulatory momentum (Canadian AI Act likely to classify recruitment AI as high-impact). Governance gaps widened—only 28-31% of employers coordinated vendor bias reviews or assessed compliance. Adoption paradox crystallized: strong deployment growth alongside unresolved fairness risks and incomplete regulatory readiness.
• 2023-H1: Findem continued momentum with 3x ARR growth and major enterprise wins (PayPal, Nutanix); platform usage surged 275% for searches and outreach. New market entrants (HireGenie) claimed significant time savings. Production deployment confirmed at scale (government agencies, mid-market, staffing firms), but data quality challenges persisted. Expert analysis (interviewing.io) distinguished between automated outreach efficacy and unresolved candidate-quality prediction; fairness and effectiveness debates remained unresolved despite broad operational deployment.
• 2023-H2: Generative AI penetration accelerated with SeekOut releasing Assist for personalized outreach automation. Findem achieved Gartner Cool Vendor recognition, validating speed-and-quality pitch to enterprises. Production deployment scaled across vendor ecosystem (SeekOut, Findem, hireEZ), confirming category operational maturity. Adoption friction intensified: 49% of job seekers perceived AI recruiting as more biased than humans (ASA Workforce Monitor), and practitioner research documented widespread candidate experience failures—90% ghosting rates, complexity-driven opt-out, and persistent data quality issues in real-world deployments. Outreach automation worked at scale; the tension remained unresolved between operational efficiency and candidate fairness/experience degradation.
• 2024-Q1: Adoption reached mainstream status with AI talent acquisition rising from 26% to 53% year-on-year; staffing firms adopting AI were 31% more likely to see revenue gains. Findem launched agentic Intelligent Job Post automating full source-to-hire workflows. Analyst case studies documented substantial efficiency gains (95% candidate satisfaction handling 200K+ applicants, 423% interview increase with 85% lower drop-off). However, real-world data revealed critical deployment limitations: fully automated outreach achieved only 2.1% reply rates vs 11.3% with human oversight, with sequences degrading after 10-14 days. Sourcing automation proved operationally mature for volume but required human-in-the-loop to be effective.
• 2024-Q2: Vendor innovation continued with Findem's Copilot for Sourcing launch automating end-to-end candidate sourcing within ATS/careers pages. Ashby reported 46% lift in reply rates from AI-powered outreach, confirming measurable efficacy. Market adoption remained broad (60% of tech recruiters using SeekOut), but critical gaps persisted: Business Insider investigation documented widespread friction from AI hiring—incompatible formats, bot screening, spam-like outreach, candidate experience degradation. Category confirmed mainstream operational status with unresolved implementation quality and fairness challenges.
• 2024-Q3: Vendor platform evolution continued with SeekOut's agentic AI service (Spot) and Findem-HireBrain partnership expanding automation scope. Adoption metrics reinforced mainstream status: iHire survey showed 14.7% AI adoption in recruitment (up from 4.9% YoY), with threefold increase indicating sustained growth. However, critical friction signals intensified: Capterra survey documented that 38% of candidates would reject offers from AI-heavy processes and 60% prefer human involvement at any stage, signaling candidate experience risk. Legal challenges emerged with Mobley v. Workday (Workday's AI tools alleged to discriminate), with federal court allowing disparate-impact claims to proceed and EEOC amicus support, highlighting unresolved fairness and regulatory risks. Category remained operationally mature at mainstream adoption scale, but the tension between automation efficacy and candidate experience/fairness degradation remained unresolved.
• 2024-Q4: Vendor platform expansion accelerated with Findem launching Executive Search Platform and Remote releasing Recruit AI for global sourcing automation. Real-world deployment confirmed at scale: RingCentral's Findem deployment yielded 40% pipeline lift and 40% increase in under-represented candidate interest. However, regulatory scrutiny intensified: UK ICO audit found AI recruitment tools unlawfully filter candidates by protected characteristics without lawful basis, exposing compliance risks. Research highlighted automation bias risks in AI systems, while practitioner analysis suggested sourcing roles will evolve rather than disappear, with AI automating information gathering while humans manage relationship building and strategic talent attraction. Category solidified as mainstream operational capability with continued innovation, but the fairness and implementation quality tension remained central to adoption risk.
• 2025-Q1: Adoption reached 73% of companies implementing recruitment automation; SHRM data confirmed 32% automating candidate searches directly. Peer-reviewed research (n=469) found generative AI reduced bias in screening through algorithmic consistency and improved efficiency. New vendor entrants (hireEZ) claimed faster deployment with agentic platforms; market remained highly competitive. However, implementation friction persisted: surveys showed 42% of C-suite executives reported AI adoption was "tearing companies apart" due to organisational silos and poor ROI. Regional barriers remained—algorithm bias risks and data quality challenges limited applicability in markets with labour shortages. Category confirmed at mainstream production maturity with proven efficiency gains (30-40% cost reduction, 85% time-to-fill improvement), but successful deployment required human oversight and organisational change management.
• 2025-Q2: Vendor innovation accelerated with agentic AI product launches: SeekOut Spot (May) and Findem Talent CRM (May) both released autonomous sourcing and relationship management capabilities. Deloitte identified AI agents as defining 2025 trend; Korn Ferry reported 67% of HR leaders prioritizing AI as top trend. Academic research (May) acknowledged bias reduction potential while warning of systemic inequity risks from model training bias. UK ICO compliance audit (concluding June 30) documented unlawful filtering by protected characteristics in vendor tools, contradicting bias mitigation claims. Category sustained mainstream production maturity with proven efficacy (5X candidate volume claims, 800M+ profile access) but escalating regulatory scrutiny and unresolved fairness-efficacy tension required compliance testing, bias auditing, and ethical implementation frameworks for successful deployment.
• 2025-Q3: Market adoption accelerated with 57% of companies using AI in hiring (Resume.org survey, September 2025), and 1 in 3 companies expecting full AI automation of hiring by 2026. Vendor deployment metrics confirmed sustained efficacy: Findem's analysis (August 2025) cited 85% organizational satisfaction, Box's 3-5 hour time savings per role, and RingCentral's 40% pipeline lift and 40% rise in underrepresented candidate interest. Industry benchmarks showed AI tools cutting time-to-hire by up to 70% against SHRM's baseline 42-day standard hiring cycle. Fairness data emerged with conflicting signals: Warden AI's 2025 audit (August 2025) reported 85% of audited systems met accepted fairness thresholds, with AI systems delivering 39-45% fairer treatment for women and racial minorities compared to human-led processes. However, critical assessments persisted—Hueman RPO identified widespread implementation pitfalls including historical data bias, over-reliance on automation without human oversight, and candidate experience risks. Market remained competitive with multiple vendors (SeekOut, Findem, hireEZ) offering agentic AI platforms, but 76% of hiring managers continued to struggle finding qualified candidates and 70% of the workforce remained passive, signaling sourcing still drives operational friction despite automation. Category consolidated at mainstream operational maturity with proven efficiency and diversity outcomes, but implementation quality, fairness audit rigor, and organizational change management remained critical success factors.
• 2025-Q4: Adoption intent remained strong with 67% of recruiters planning increased AI investment (Lever, 1,200+ sample) and 83% prioritizing passive candidate engagement (LinkedIn research). Deployment efficacy confirmed: RPO.AI case study reported 65% time-to-hire reduction, 3x response rate improvement, 50% cost reduction; email automation delivered 356% higher response rates with multi-touch campaigns. However, fairness and effectiveness risks crystallized: University of Washington research found AI tools favor white-associated names 85% and male names 52%; Talroo analysis documented declining AI outreach effectiveness from candidate skepticism and scale fatigue; regulatory escalation accelerated with state-level bias audit mandates (CA, NY, CO, IL) and Workday facing federal court proceedings on discrimination claims. Data quality persisted as barrier (20-30% contact information outdated). Category remained mainstream operational maturity with documented limitations on fully automated outreach; successful deployment increasingly required rigorous compliance testing, continuous fairness auditing, and human oversight to close vendor-claimed ROI vs real-world implementation quality gap.
• 2026-Jan: Agentic AI consolidation around SeekOut (750+ customers), Findem, and hireEZ; adoption expanded to 80% of large companies and 73% implementing tools, but TA budgets flat signaling saturation. Channel disruption emerged: Gmail AI filters launched January 2026, reducing email efficacy (2-4% reply rates vs 18-25% on LinkedIn). Efficiency gains confirmed (30-50% time-to-hire, 20-40% cost reduction), but channel fatigue, data quality, and fairness risks became defining barriers to further expansion—successful deployment required multi-channel strategy and human oversight.
• 2026-Feb: Agentic AI deployment intent remained strong (82% HR leaders planning 12-month adoption) but execution barriers consolidated. Cost-of-deployment analysis documented TCO of $60K-$250K first-year, ROI within two quarters; arena practitioner assessment confirmed 40-60% sourcing time savings with human-in-the-loop, 50-70% time-to-fill improvement. Governance complexity emerged as critical blocker: compliance exposure (EEOC/DOJ), algorithmic bias, data privacy gaps, and operational fragility from weak governance requiring NIST RMF frameworks. Category remained mainstream operational maturity but adoption expansion slowing due to budget constraints and rising complexity in fairness/compliance deployment.
• 2026-Mar: Market validation confirmed but with critical signal degradation. Phenom's 2026 award winners (Elara Caring, Thermo Fisher, Dutch Bros, Aspen Group) documented measurable outcomes (2.3-day offer closure, 64% time-to-fill reduction, $257K annual savings), validating 40-60% sourcing time improvements with human curation. SAP SuccessFactors analysis (1.3M requisitions) documented "AI arms race" between candidates and employers: 39% of candidates now using AI to optimize applications, 51% willing to fabricate credentials, application volume doubled since 2021 yet signal quality declining. Candidate-side response intensified: 34% of job seekers used AI assistance (31% YoY increase), 70% customize resumes, 68% demand skills-based alternatives. Implementation failure signals accelerated: candidate ghosting peaked at 53% (3-year high), 67% of HR leaders report AI-generated applications slowing hiring, 85.1% of AI screening still favors white-associated names (Brookings/UW), email channel collapsed (2-4% reply rate post-Gmail filters). Market consolidation confirmed around SeekOut/Findem/hireEZ, but adoption saturation signaled: 80% of large companies implemented tools, TA budgets flat (30% expecting growth), suggesting competitive displacement complete but tier advancement blocked by signal-to-noise and fairness governance barriers. Adoption intent among talent professionals remained at 74% planning increased use (up from 45%), but execution reality — ghosting, bias, channel collapse — continued to constrain further tier advancement.
• 2026-Apr: Market confirmed mainstream status with 87% of organisations using AI-driven recruiting platforms and the global market at $707M growing at 7%+ CAGR. Korn Ferry survey found 50%+ of TA leaders planning fully autonomous AI agents for sourcing in 2026, while enterprise deployments (Hunkemöller: 78% time-to-hire reduction; executive search firms: 64-68% time-to-shortlist compression) validated agentic platform ROI. SHRM simultaneously flagged governance gaps blocking implementation despite technical readiness, and signal-based personalisation pressure intensified as DMARC enforcement and 20+ US state privacy laws reshaped email automation strategy.
• 2026-May: Adoption breadth reached 87% of organisations but the execution gap widened: iCIMS/Aptitude Research found 46% have sourcing automation deployed while 74% of candidates now use AI on their side — outpacing employer tooling and degrading signal quality. The ROI picture split sharply: Andiamo (boutique tech search) doubled client submissions and cut time-to-fill 76% with AI-powered semantic matching, while an NVIDIA 2026 report found 60% of organisations saw time-to-hire actually increase despite AI investment, with Forrester reporting a third still in pilot phases. Only 38% of organisations with tools achieve meaningful scale and 31% can defend ROI rigorously; autonomous outreach achieves 5–10% hire rates versus 50% in human-oversight hybrid models, and candidate-side transparency failures persisted (50.5% rejected without feedback, 9.7% disclosed AI involvement, 31.4% abandoned).
• 2026-Jun: Agentic sourcing deployments continued demonstrating production ROI — MokaHR's 3,000+ enterprise customers reported 63% time-to-hire reduction; PwC documented 70% sourcing time savings; Squirrels AI contacted 2,000 candidates/month for a named IT staffing agency; Adecco's global deployment across 10 countries reached 1.2M candidate interactions and 50,000 jobs filled with 50% time-to-deliver reduction. SeekOut democratized SMB access with Recruit Core ($149/mo, 1B+ profile index), and Yarco (UKG) processed 12,500 applicants with application-to-interview compressed from weeks to under 5 minutes. Bullhorn GRID data (1,000+ US employers) showed 54% with candidate search automation and 30% using agentic AI, with firms embedding AI fully 2x as likely to hit 75%+ fill rates. Critical structural signals: Stanford empirical research confirmed 26% of Black and 15% of Asian applicants rejected at adverse-impact rates across 3.4M applications at a single vendor; state AI employment laws (Colorado, Illinois, NYC, Texas) imposed bias audit requirements while DOJ challenged state-level mandates, creating patchwork compliance burden. A Phenom analyst review found only 27% of vendor-claimed AI features are actually live with customers. The defining constraint sharpened further: ManpowerGroup/Everest Group survey (90 C-suite leaders) found 90% deployed AI in recruiting yet fewer than 5% report transformational outcomes, while AI-only cold email achieved 4.1% reply rate versus 10.4% human-written and 15–25% signal-based — SHRM data confirms cost-per-hire and time-to-hire both increased over the AI adoption period, and volume outreach collapsed to sub-1% reply rates, making signal-first architectures the prerequisite for efficacy.
• 2026-Jul: Cold email decline continued (reply rates down to 3.43%, from 8.5% in 2019, amid Gmail/Yahoo/Microsoft DMARC enforcement), and new evidence showed AI SDR tools carry 50-70% first-year cancellation rates and a 9-12 point win-rate gap versus human-sourced opportunities. Findem Studio reached GA with 850M+ MCP-enabled profiles for agentic sourcing, and Fortune-company deployments (L'Oréal, Unilever, AWS) confirmed 30-50% time-to-hire reductions, though a ManpowerGroup/Everest Group C-suite survey again found fewer than 5% report transformational AI outcomes despite 90% adoption. Late-July evidence reinforced the adoption-value gap (87% of companies use AI in recruiting but value realisation lags), while a named case study (Garmony AI deployment) documented concrete before/after HR KPI shifts and Y Combinator's active funding of 58 recruiting startups confirmed sustained venture investment in the category.
• 2026-Aug: Large-scale benchmark data (13.2M LinkedIn requests, 844K recruiting sequences, 46K+ InMails) confirmed a widening acceptance-reply gap—connections easier to secure (26.3%→28.1% acceptance) but replies falling (22%, down from 32.2%)—with targeting precision (prior contact, signal-based outreach) outperforming volume by up to 10x. Adoption surveys showed a persistent value-realization gap (only 31.6% of managers use AI in hiring, 13.8% report significant improvement, 22.9% knowingly interviewed suspected deepfake candidates) and practitioner consensus hardened around selective automation ("automate the work, never automate the trust"), with agents reliably handling 60-70% of recruiter hours rather than the 95%+ often claimed, while senior-hire failure rates remain unchanged at ~30% since 2018 despite cheaper sourcing.