Below is a concise, evidence‑based teardown of Agency Leads (agency-leads.com)—what they appear to do to produce those “too‑good” leads (company name, hiring manager contacts, current and past roles, and confirmation that the company uses recruiting agencies)—followed by a concrete, compliant blueprint you can use to replicate the strategy.
Executive summary (what they do in plain English)
- They aggregate job postings at scale
Agency Leads runs two products:- Open Web Leads (OWL): a job board scraper that “automatically search[es] the top job posting websites” and pulls job details and employer contact info into their system.
- Agency Leads (core product): a curated, human‑verified feed limited to companies confirmed to use staffing agencies for hiring, with job details and hiring‑manager contact info.
- They filter for “agency-using” employers
The “verified” badge means they “confirm the accuracy of the job listing and, most importantly, verify that the hiring company uses staffing agencies and recruiters.” - They enrich each job with decision‑makers and a “likely gatekeeper”
They provide a directory of relevant contacts; the check mark contact is chosen by an algorithm that (per their own FAQ) uses LinkedIn profile‑view signals between the associated staffing firm and the end client to infer the most likely hiring gatekeeper. - They operate a real‑time, QA‑assisted pipeline
Their “AI system automatically sources quality leads daily,” and a “Lead Genius team” reviews and verifies the company info, job details, and hiring‑manager contact information before delivery. - They package this with saved searches, alerts, exports, and ATS/CRM integrations
Saved searches, export, “ticker”/alerts, and a “contact search” feature are documented in their support/FAQ; integrations include Bullhorn, HubSpot and RecruitCRM.
Scale note: they publicly claim 225k+ verified leads across the US/UK/CA/AU and daily updates.
How this likely produces the results you’re seeing
| Your observation | Likely mechanism behind the scenes |
|---|---|
| “They give company name & current/past roles.” | Continuous ingestion of job ads from job boards + ATS job boards, with a data store that tracks first‑seen/last‑seen dates per role; “job posting date(s)” are surfaced in the UI. OWL is explicitly a job board scraper; the core product is the curated slice. |
| “All hiring manager contacts.” | Contact enrichment against external B2B datasets, then ranking. The UI shows a contacts directory, with one contact “check‑marked.” |
| “They confirm the company is working with at least one recruiting agency.” | They only include “verified buyers”—i.e., employers confirmed to use staffing agencies. Likely signals include postings by agencies for named clients and other corroborating patterns; their FAQ mentions an associated staffing firm per posting available via chat. |
Replication blueprint (compliant, efficient, and practical)
Goal: Build a lean, legally safe pipeline that reproduces the output: (1) a live stream of “agency‑using” employers, (2) job/role timelines, (3) ranked hiring‑manager contacts with verified email/phone, and (4) export + alerts.
1) Data ingestion (jobs at scale)
- Use official, allowed sources first (fastest path to coverage + compliance):
- Adzuna Job Search API (broad aggregator): fetches current listings across major boards.
- ATS job-board APIs (direct company career pages):
- Greenhouse Job Board API (public job JSON).
- Lever Postings API (public job JSON).
- SmartRecruiters Jobs API.
- Collector design: Run incremental crawls (country/role keywords) every few hours; store raw JSON per source with a source_id + source_hash for de‑duplication.
Why this maps to Agency Leads: OWL is their “job board scraper” layer; you’re recreating it with official APIs + a single aggregator API (Adzuna).
⚠️ Compliance guardrail: Avoid scraping sites that prohibit it (e.g., LinkedIn). Their FAQ references LinkedIn‑derived signals, but LinkedIn’s user agreement prohibits scraping/automation; use licensed data instead (see §3).
2) Normalize, de‑duplicate, and track history
- Normalization: Map titles → taxonomy (role family, seniority), locations → ISO geos, compensation fields.
- De‑duplication: For each (company, title, location), select the “best” record by recency + completeness; retain first_seen and last_seen timestamps to build role timelines (the “what they’ve hired in the past and when”).
- Company resolution: Enrich each posting to a canonical company (domain + legal name) using a company enrichment API (PDL or Clearbit).
3) Detect “agency‑using” employers (your unique filter)
Build a lightweight classifier that assigns an Agency‑Involvement Score (AIS):
- Primary (hard) signals:
- Posting originates from a staffing firm (employer domain ∈ curated staffing list). Start with public rosters (e.g., Staffing Industry Analysts’ “Largest Staffing Firms” list) and expand.
- A third‑party recruiter posts an identical role for a named client (match on title/location/description n‑gram similarity).
- Secondary (soft) signals:
- Job copy with phrases like “our client,” “on behalf of,” etc.
- Application link goes to agency ATS (vs the employer’s domain).
- Decision rule: AIS ≥ threshold → verified buyer of staffing services (what Agency Leads markets as “verified”).
Their FAQ also notes an “associated staffing/recruiting firm” per posting (revealed on request), indicating they track the agency ⇄ end client relationship. Mirror that linkage in your data model.
4) Enrich to “all hiring manager contacts” (with verification)
- People/role discovery:
- People Data Labs – Person Enrichment (title, seniority, org).
- Apollo People Enrichment (contact details + employment history).
- Clearbit Enrichment (company/role context).
- Email/phone validation: run ZeroBounce (or Kickbox) to verify deliverability and reduce bounces.
Rank the “likely gatekeeper”:
You can’t (and shouldn’t) replicate their LinkedIn profile‑view method. Instead, compute a Hiring‑Influence Score (HIS) per contact:
- Role proximity: match the job family to the org (e.g., SWE → Dir/VP Eng; FP&A → CFO/Controller).
- Seniority & team ownership: IC < Manager < Director < VP < C‑Level.
- Recency: current title tenure.
- Talent function weight: include TA/Recruiting leadership for BD entry points.
- Signal adders: past posting history for similar roles at the company; public mentions as hiring manager.
Top HIS → check‑mark equivalent in your UI.
Agency Leads’ check‑mark is “decided by an algorithm … based on LinkedIn profile views from the associated firm to the end client.” Use the safer proxy above to avoid ToS issues.
5) Human QA loop (the “Lead Genius” step)
- SOP: For leads above threshold, a researcher quickly verifies: company ↔ role ↔ contact triad (title still current, email verified, role actually open).
- Feedback: Analysts flag false positives to improve the AIS/HIS models.
This mirrors their claim that a human team validates before delivery.
6) Delivery: saved searches, alerts, exports, integrations
- Saved searches + email alerts: store user queries and push daily digests of new/changed matches (this is akin to their “ticker” & alerts).
- Exports & CRM/ATS: CSV/XLSX export; native connectors to HubSpot/Bullhorn/RecruitCRM first (they list these).
Minimal technical stack (quick to stand up)
- Ingestion: Serverless jobs hitting Adzuna + ATS APIs; store raw JSON → PostgreSQL / BigQuery.
- Processing: A daily dbt pipeline to normalize/deduplicate; embeddings (optional) for fuzzy job matching.
- Enrichment: Queue to PDL/Apollo/Clearbit; email verification via ZeroBounce batch.
- Scoring: Simple logistic or gradient‑boosted model for AIS; rule‑based → learned HIS over time.
- App: A small React or internal tool (filters, saved searches, contact pane), CSV export, webhooks to CRM.
- QA console: lightweight interface for reviewers to approve/flag leads.
Data model (essentials)
- companies(company_id, name, domain, country, …)
- agencies(agency_id, name, domain, …)
- jobs_raw(source, source_id, payload, fetched_at)
- jobs(job_id, company_id, title_norm, family, location_norm, first_seen, last_seen, status)
- job_agency_links(job_id, agency_id, evidence_type, evidence_score)
- contacts(contact_id, company_id, name, title, seniority, email, phone, last_verified_at, source)
- job_contacts(job_id, contact_id, his_score, is_primary BOOLEAN)
- signals(job_id, signal_type, value, weight) — used to compute AIS/HIS
- alerts(user_id, saved_query_id, delivered_at, payload)
What to buy vs. what to build
- Buy/licence: Adzuna + ATS APIs access (no scraping headaches), contact data (PDL/Apollo/Clearbit), and email verification (ZeroBounce).
- Build: Normalization, dedupe, AIS/HIS scoring, UI with saved searches/alerts, and the QA loop.
Compliance & risk (how to stay out of trouble)
- Avoid scraping platforms that prohibit it (notably LinkedIn). Courts have upheld LinkedIn’s User Agreement breach claims against scraping; use licensed data and official APIs instead.
- Respect job‑board ToS and robots.txt and prefer aggregator/ATS APIs. (Cloudflare’s rising default anti‑scraping posture is another reason to stay API‑first.)
- Privacy laws: ensure GDPR/CCPA basis for processing B2B contact data (legitimate interest + opt‑out, suppression lists, etc.).
Practical scoring recipes you can copy
Agency‑Involvement Score (AIS)AIS = 4*(AgencyDomainMatch) + 3*(DuplicateFoundAtAgency) + 2*(“our client” phrase) + 1*(Non‑company apply URL) – 2*(Only company‑domain postings)
Threshold → mark employer as “active with agencies” (your verified buyer).
Hiring‑Influence Score (HIS)HIS = 4*(RoleFamilyMatch) + 3*(SeniorityWeight) + 2*(ManagerOfHiringTeam) + 1*(RecentTenure < 24m) + 1*(TA Leadership for BD) + 1*(Past similar hires)
Top HIS → your primary contact (the “check mark”).
Where each key claim above came from (selected sources)
- OWL is a job board scraper; scrapes “top job posting websites”; pulls job + employer contact info.
- Core product = verified companies that use staffing agencies; AI sources daily; human team verifies job, company and contact info; 225k+ verified leads across US/UK/CA/AU.
- “Check‑mark” gatekeeper chosen by algorithm using LinkedIn profile‑view signals between the associated firm and end client; associated staffing firm per posting available via chat; ticker/alerts & exports documented in support/FAQ.
- Integrations list (Bullhorn, HubSpot, RecruitCRM, etc.).
- API‑first replication path: Adzuna; Greenhouse Job Board; Lever Postings; SmartRecruiters.
- Contact enrichment & email verification: PDL, Apollo, Clearbit; ZeroBounce.
- Staffing‑firm seed list: SIA Largest Staffing Firms (starting point to curate agency domains).
- LinkedIn anti‑scraping precedent / ToS risk (hiQ litigation; enforceability of user agreement).
If you want to out‑execute them
- Add sector‑specific intent (funding rounds, new plant openings) to prioritize outreach.
- Score “replaceable roles” (where agencies historically help) to focus BD.
- Close the loop: capture outcomes (won/lost, meetings set) and feed them back into AIS/HIS to auto‑tune thresholds.
Bottom line
Agency Leads appears to win by combining broad job aggregation (OWL) with an agency‑use filter, contact enrichment, and a human QA layer, then shipping it with alerts, exports, and integrations. You can replicate the essence without violating ToS by leaning on authorized job/ATS APIs, licensed enrichment, email verification, and straightforward scoring + reviewer workflows. The blueprint above gives you the exact building blocks—and avoids the legal/operational pitfalls.