Greenhouse

What it is

Greenhouse is the default ATS for in-house B2B SaaS recruiting, and the product that turned structured hiring — scorecards, interview kits, rubrics, source-of-hire reporting — into a category expectation rather than a differentiator. TPG has held a majority stake since January 2021, a $500M investment at roughly an $825M valuation, and Greenhouse has run as a standalone platform since.

Two changes in 2026 retire the older read that Greenhouse ships AI late. On 7 May it announced a first-party MCP server, which reached open beta on 6 July across every subscription tier. On 27 May it closed the acquisition of Ezra AI Labs and shipped Voice AI, a two-way screening interview rather than a one-way recording. The plan ladder was rebuilt in the same window: Essential / Core / Advanced / Expert is gone, replaced by three tiers — Core, Plus, Pro.

Why it shows up in recruiting stacks

• Structured hiring is the schema, not a module. Scorecards, interview kits and source-of-hire are first-class objects with their own reporting, so interviewer discipline survives a hiring manager who does not want it. That is what keeps a 40-req quarter auditable.
• 500+ pre-built integrations. Whatever assessment, scheduling, background-check or sourcing tool wins your evaluation, it ships a certified Greenhouse connector first — CodeSignal, Checkr, GoodTime, Gem and Metaview all do. Connector coverage is the one axis where Greenhouse still beats every ATS in the segment.
• Agent access with an admin ceiling. Greenhouse MCP is on Core, Plus and Pro. A Site Admin enables scopes under Dev Center → MCP Access, and users cannot approve more than the admin allowed; an org that never configures it gets read-only scopes by default. It reads and writes, works with Claude, ChatGPT, Glean, Copilot Studio, Amazon Q, Google Antigravity and Grok, and accepts any client doing OAuth 2.0 with PKCE. Every call is permission-aware and audited — the answer to the CSV-export shadow integrations TA teams otherwise build on their own.

Pricing

Three plans, all quote-only. Greenhouse states pricing is “customized based on your company’s hiring needs” and names the plan, hiring volume, organizational complexity, and required features as the inputs.

• Core — sourcing and talent CRM, structured interview kits, scheduling, reporting and analytics, SSO, the AI Notetaker, and Real Talent matching.
• Plus — adds AI-powered report filters, the Business Intelligence Connector, Sourcing Automation, candidate texting, and the fraud and spam detection half of Real Talent.
• Pro — adds unlimited CRM events, an audit log, enterprise data configuration and security, a developer sandbox with sync, and enterprise Real Talent.

What teams pay is a different number from what the tier list implies. Vendr’s marketplace data, drawn from 873 purchases, puts the median Greenhouse contract at $26,587/year against a range of $10,137 to $74,492, with buyers negotiating an average 16% off the first quote. Buyer-reported 2026 procurement trackers add roughly $10–30K of implementation and 8–15% annual renewal escalation — resale-benchmark estimates, not vendor figures, so treat them as the shape of the deal rather than the deal.

Two tier boundaries decide more evaluations than the headline price does. SSO sits in Core, so security review does not push you up a tier — the opposite of Ashby, which charges $100/month for it on Foundations. Fraud and spam detection sits in Plus. If application fraud is the reason you are shopping, Core does not solve it, and Plus is the quote to ask for on the first call.

Voice AI is sold outside the plan ladder. Greenhouse lists it as available for purchase in the US and internationally, in English plus 10 more languages, with monthly bias audits by Warden AI. Budget it as a line item, not a tier upgrade.

Best for

TA teams at 200–2,000-employee B2B companies with a dedicated recruiting-ops function, where the real problem is interviewer discipline across dozens of hiring managers, and where the integration surface — assessments, background checks, scheduling, HRIS — is wide enough that connector coverage decides the evaluation.

Not for

Companies under 50 employees making fewer than 20 hires a year. Headcount-indexed, quote-only pricing burns more evaluation time than the process discipline returns at that size, and Workable and Ashby publish ladders you can price in an afternoon. High-volume hourly hiring belongs with Paradox or Fountain. Staffing and agency recruiters belong with Bullhorn — Greenhouse’s data model is shaped for in-house TA and carries neither placements nor client billing.

Versus the alternatives

• Workday — the other half of the enterprise segment, and the pick when recruiting has to sit inside the HCM system of record and share one worker object with payroll and performance. Greenhouse wins on interviewer workflow and connector breadth; Workday wins the moment HRIS consolidation is the deciding vote.
• Ashby — the fastest-growing entrant here and the strongest argument against Greenhouse below 500 employees, because ATS, sourcing, CRM, scheduling and analytics arrive on one bill with a published price ladder under 100 employees. Pick Ashby when native analytics and a single vendor decide it; pick Greenhouse when you are buying assessment, background-check and scheduling tools separately and need them all supported on day one.
• Lever — the cheaper structured-hiring alternative, also quote-only, now under Employ Inc. Pick Lever when the ATS-plus-CRM merge is the point and the budget will not clear $25K; pick Greenhouse when approval chains, permissioning and audit requirements come from a public-company compliance function.

Watch-outs

• Changing MCP scopes invalidates every active MCP token in the org. Everyone reconnects, and any agent workflow someone built stops mid-run. Guard: batch scope changes into one announced maintenance window instead of tuning them request by request.
• MCP is open beta, not GA. Tool surfaces and behavior can move without a deprecation window. Guard: keep the MCP path off hiring-manager-facing surfaces, and pin an integration test to the tool schemas until Greenhouse names a GA date.
• Two of the six AI features announced on 10 June 2026 have not shipped. The Job Kickoff Agent and the Candidate Insights Agent were both dated Q3 2026 and are still listed as expected; the Notetaker is GA, and Chart Builder and MCP are in open beta. Guard: score the deal on what is GA today and write the two agents into the contract as a milestone, not an assumption.
• Renewal escalation compounds against a headcount you plan to grow. Buyers report 8–15% annual uplift on a contract already indexed to company size, so a team that doubles headcount without changing hiring volume pays roughly twice for the same 40 reqs. Guard: negotiate a multi-year rate at signing, model the year-two headcount rather than today’s, and open from Vendr’s 16% average discount instead of the first quote.
• Analytics stops short of BI. Chart Builder answers plain-text questions and AI Report Insights summarizes a dashboard, but the Business Intelligence Connector — the path into a warehouse — starts at Plus. Guard: if a data team already owns hiring reporting, scope Plus from the first call rather than discovering the connector gate at renewal.