Greenhouse leads on each of seven criteria, widest on candidate pipeline (5.2 to Lever's 1.8), AI screening control (3.8 to 0.5), job distribution (2.5 to 0.2) and applicant data governance (3.2 to 1.8) — six judges lean Greenhouse on each, and no judge leans Lever on any criterion. Integrations and handover leans Greenhouse, 4.0 to 2.5, though Lever's terms there promise API export of customer data at any time during the term. The genuine splits sit at the low end: sovereignty ties four judges (0.5 versus 0.0, both products listing jurisdiction, ownership, data residency and subprocessor exposure as unknown), and pricing transparency ties two (1.8 versus 1.2, neither vendor publishing prices). Lever's AI screening control split its judges: some credited named features, others refused credit absent disclosure of what the models evaluate or how to switch them off. Weighted totals: Greenhouse 2.7 to 3.9, Lever 1 to 1.7.
Choose Greenhouse if
You need depth across the candidate pipeline: the means read 5.2 to Lever's 1.8, with six judges leaning Greenhouse and zero leaning Lever.
Your team distributes roles widely across job boards: job distribution reads 2.5 to 0.2, and Lever's own verdict places its score there at 0-1.
You need control over AI screening: the means read 3.8 to Lever's 0.5, and the bench split on Lever's credit absent disclosure of what the models evaluate or how to switch them off.
Your applicant-data requirements go beyond contract-end deletion: Lever's verdict says its governance rests on contract-end deletion alone; the means read 3.2 to 1.8.
Your HR-IT integrator weighs integrations and handover: the means read 4.0 to 2.5, and that judge's weighted total reads 3.4 to 1.7.
Choose Lever if
You need contractual data-exit rights: Lever's terms promise API export of customer data at any time during the term, plus third-party API interoperability — the basis of its strongest scores, on integrations and handover.
Your recruiting ops team runs on dashboards: the verdict credits Lever's candidate pipeline score to reporting strength with out-of-the-box dashboards.
You want AI features named with stated purposes — matching, resume screening, unlimited interview transcripts and summaries, fraud signals — which some judges credited.
Your stack ties into email providers and LinkedIn through customer-granted access, or already sits in the Employ Inc. group alongside Lever, Inc.
Read this comparison as one judge. Each weighs the same scores by what they care about.
The Works Council Advocate
Sits on the Betriebsrat, which co-determines selection guidelines and questionnaires. Asks what any ranking evaluates, whether it can be switched off, who sees which applicant, and whether an internal candidate's file leaks into their personnel record. No answer, no agreement.
Greenhouse
This judge's pick
Lever
Criterion by criterion
Applicant management & hiring workflow
Greenhouse
Structured interview kits and scorecards ship in the entry plan alongside scheduling and reporting — the substance of structured hiring — and an audit log and enterprise-level data configuration exist, but only in the top plan. We found no public information on approval workflows for requisitions and offers, on a per-candidate history of every stage change and who made it in the lower plans, or on templated candidate communication in German.
Lever
The product is marketed as a combined ATS and CRM with dashboards, pipeline-health reporting and AI interview transcripts that streamline feedback and decision-making, but we found no public information on pipeline stages per job, hiring-manager access limited to their own vacancies, structured scorecards, or a record of who decided what. For co-determination purposes nothing here evidences a documented decision trail, and a per-candidate history export for an AGG dispute is not in evidence either.
Multiposting, career page & candidate experience
Greenhouse
The only candidate-facing evidence is the MyGreenhouse portal, which keeps candidates informed with real-time updates and surfaces opportunities — candidates can at least see their own status. We found no public information on job-board distribution, a branded career page, German boards such as the Bundesagentur für Arbeit, or a mobile application without an account.
Lever
We found no public information on job-board distribution, a career page, Google for Jobs markup, or the application form and its mobile usability — neither the Bundesagentur für Arbeit nor any German board is named on the captured pages. I cannot co-determine a questionnaire on candidate experience from silence, so I score at the floor for absence.
Retention, deletion & talent-pool consent
Greenhouse
The contract queues all customer data for deletion ninety days after termination, the privacy policy states personal information is deleted or anonymised when there is no ongoing business need, and a data-subject request portal exists. We found no public information on a configurable per-applicant deletion deadline after rejection, on recorded consent for talent-pool storage, or on a log of who viewed or exported an applicant file outside the top plan's audit log.
Lever
The only retention statement is contractual: after the term ends, customer data "will no longer be accessible and will be promptly deleted in accordance with Lever's data retention policy" — that protects the employer at contract exit, not the rejected applicant on a deadline. We found no public information on automatic deletion after a configurable period post-rejection, on talent-pool consent with expiry for the actively marketed CRM nurturing, on data-subject request handling, or on a log of who viewed an applicant file.
AI screening transparency & control
Greenhouse
AI runs through the product — AI recruiting, talent matching, fraud detection and identity verification, an AI Notetaker, Voice AI — with dedicated AI Terms incorporated into the agreement, published AI principles and a bias audit statement, and the master agreement stating the customer remains solely responsible for all hiring decisions. We found no public information on what any of these features evaluate, on a per-feature switch, on the model provider, or on an EU AI Act position, and the privacy policy reserves the right to use personal information to develop and train AI models — for a works council agreement that is not enough to sign.
Lever
AI screening that automatically prioritizes top candidates and AI resume screening are marketed, yet we found no public information on what the matching evaluates, where it runs, which model provider processes the CVs, or whether each feature can be disabled per account. A ranking whose inputs are undocumented and whose switch is unstated is precisely the co-determination case for recruitment AI under the EU AI Act; the features are individually named with stated purposes, which is why I sit just above the floor.
HR handover, integrations & API
Greenhouse
An API is offered with usage limits, rate limits and fees set per Order Form, hundreds of pre-built integrations are cited as a count, SSO ships in the entry plan, and the vendor sells its own onboarding product. We found no public information on a handover carrying a hire's data into an HR or payroll system common in DACH, on SCIM, on e-signature for contracts, or on any named integration partner.
Lever
An API export of all customer data is contractually promised, and the customer may grant Lever access to its own email providers and LinkedIn accounts — but that is the flow into Lever, not the handover of a hire into the HR system and payroll. We found no public information on maintained integrations with systems common in DACH, SSO, e-signature for offers and contracts, or webhooks with published limits, and nothing on whether documents or consent status travel into the personnel record.
European sovereignty
Greenhouse
Every applicant file would belong to a US contracting entity under New York law and exclusive New York jurisdiction, with primary offices in the United States and processing that the privacy policy says may occur in any country where providers and partners operate; subcontractors are permitted but we found no published list. Data Privacy Framework certification is a US framework, and we found no public information on EU hosting or on where the AI and parsing features run — for records belonging to third parties who never chose this vendor, that is the bottom of the scale.
Lever
The contracting entity is Lever, Inc. of San Francisco under California law, with binding arbitration in San Francisco, inside the Employ Inc. group; we found no public information on data residency, hosting, or a subprocessor list covering the AI providers and CV processing. Every applicant record belongs to a third party who never chose this vendor, so an American jurisdiction over their files with an unstated hosting location is not something a works council can agree to — I score at the floor.
Pricing transparency
Greenhouse
Three tiers are published with their feature lists, and the fee basis is stated publicly as "Customer's employee headcount, which shall also include all employees of Customer's Affiliates", with payment terms, auto-renewal and taxes-exclusive treatment documented. We found no public price figures for any tier, no add-on prices, no job-board posting costs and no API overage rates — fees are set on the applicable Order Form, so the annual invoice cannot be computed from public pages.
Lever
The captured pricing page shows feature bundles while we found no public prices and no statement of the pricing basis — seats, jobs or employees. The contract does publish payment mechanics — fees invoiced annually, payment thirty (30) days after electronic delivery of the invoice, fees non-cancellable and non-refundable, amounts "exclusive of" taxes, automatic one-year renewal — so the terms of payment are public while the invoice amount is not computable from these pages and a buyer still needs a sales conversation.
Sovereignty, side by side
Dimension
Greenhouse
Lever
Legal entity
Not determined
Not determined
Ownership
Not determined
Not determined
Data residency
Not determined
Not determined
Subprocessors
Not determined
Not determined
Facts, side by side
Only facts both products carry under the same definition — anything else would not be a fair row.