Software

Adoption of open-source CRM is growing on explicit total-cost grounds — no licence fees, self-hosting, data control — and AI features now ship in the open-source tier as standard rather than as a paid add-on.

Open-source ERP is a $5.31B market in 2026 growing at 9.66% CAGR, with adoption up 32% in 2026 driven explicitly by rising proprietary licence costs. AI has changed the shortlist: open-source options now appear on evaluations that would previously have gone straight to SAP or NetSuite.

This is the one generic category where open source does NOT set the price to zero. Every open-source option is strong on employee records and weak or absent on multi-jurisdiction payroll — the part with a real moat. That asymmetry is the study's central finding.

The open-source project management tier is projected to reach $3.2B by 2031 at 9.5% CAGR, and GitLab, OpenProject and Taiga now ship managed cloud editions — removing self-hosting effort, the last practical reason to pay.

The open-source tier here is thinner than in CRM or ERP, but it does not need depth: at the small-team end an ATS is a database, a form and a job-board integration, and OpenCATS covers all three for nothing.

No open-source CLM of consequence exists, and the reason is liability rather than capability. The artefact is an executed legal record with retention, audit and privilege duties attached, so self-hosting it transfers risk to the buyer instead of saving them a licence fee — the inverse of the calculation that makes open-source CRM attractive. Where the free tier is missing, the incumbent keeps its pricing power.

No meaningful open-source CLM exists. The artefact is an executed legal record with audit and retention duties attached, and self-hosting that is a liability rather than a saving — the opposite of the calculation that makes open-source CRM attractive.

In content management, open source is the incumbent rather than the alternative. WordPress alone runs most sites with a detectable CMS, Drupal carries much of the government and university tier, and the headless generation (Strapi, Payload, Directus) launched as free or source-available software. A paid entrant therefore competes with a free product in every architecture.

The open-source tier here is real and it is the reason the floor is zero. A posture scan is a config read against a public rule set, which a community can maintain — unlike a supplier network or a card programme. The paid tier sells triage, graph context and audit evidence, not the scan.

This is the clearest open-source price floor in the generic set. Moodle and Open edX run national education systems, so the free tier is not a toy — the paid vendors sell administration, reporting and compliance evidence rather than the ability to deliver a course.

No open-source CPQ of consequence. A configuration and pricing engine is close to worthless without the product catalogue and commercial policy behind it, and those are the customer's own confidential data rather than anything a community can ship. The absence is a statement about where the value sits, not about the difficulty of the software.

No open-source CPQ of consequence. A configuration engine is worth little without the product catalogue and pricing policy behind it, and those are the customer's private data rather than something a community can ship.

An asset repository is a file store with metadata and rights attached, which is exactly the shape a community project can carry — so the free tier here is credible, unlike procurement or expense.

A document repository is the shape a community project can build: a file store with versions, metadata and permissions. As with DAM, the free tier here is credible. The paid tiers split three ways: the Microsoft 365 default, enterprise ECM (OpenText, Hyland), and vertical leaders in legal (iManage, NetDocuments) and regulated quality documents (M-Files).

No open-source EAM of consequence exists. ERP suites such as Odoo and ERPNext include a basic maintenance module, but nothing open-source carries the regulated asset register, risk-based inspection or reliability analytics that a utility or refinery buys EAM for. Where the free tier is missing, the incumbents keep their pricing power.

EAM is a consolidated category of suite and industrial-group products. The enterprise tier is IBM Maximo, SAP and IFS, with Octave, Oracle and the infrastructure-software groups (Bentley, Trimble) behind them, and APM vendors (GE Vernova, AVEVA) layered on top. Every major independent has been acquired.

There is no open-source tier, for structural reasons. Being the employer of record means owning a registered entity, a payroll tax account and benefit contracts in each country, and those cannot be forked. Without a free tier the pressure on price comes from venture-funded rivals and suite bundling instead, and Deel's profitability shows that pressure has not yet broken the leader's economics.

No open-source EOR exists, and none can. The product is a local legal entity in each country with payroll, tax and benefits registrations attached. That is a balance-sheet and licensing asset, not code. Open-source payroll engines exist inside ERP suites, but they do not make anyone the employer of record.

No open-source tier, and interchange is the reason. The economics of this category depend on issuing a card and earning on the spend; a community project cannot issue one. That is also why the commercial price is falling toward zero without any free tier existing — the floor is being set by payments revenue, not by open source.

No open-source tier, and interchange is why. A community project cannot issue a card, and without the card the economics of this category do not work at all.

Open-source GRC is real but competes on a different axis: the paid tier sells an auditor-acceptable evidence trail and the integrations that populate it, not the control framework, which is public.

The open-source tier here is unusually strong — five mature wiki and knowledge platforms with permissive or copyleft licences — which is why the paid tier has moved upmarket into AI-assisted enterprise search rather than defending the document store.

Three layers, one buyer. Recommendation and skills graph (Degreed, EdCast, Sana), skills intelligence and internal mobility (Eightfold AI, Gloat, Fuel50, Workera), and the content the layer recommends (Go1, Coursera, LinkedIn Learning, Pluralsight, Skillsoft). The CHRO buys all three, increasingly from the HR suite that holds the employee record.

This category has no open-source tier, and the absence is structural rather than an omission. The thing being bought is a live supplier network carrying catalogue, tax and payment detail for thousands of counterparties. A community can host software; it cannot host the other side of a trading relationship. Compare CRM, LMS and DAM in this same set, where the artefact is a record, a course or a file and the free tier is strong.

No credible open-source tier exists here, and the reason is instructive: the value is a live supplier network with payment and tax data attached, which is not something a community can host into being. Compare CRM and LMS, where the artefact is a record or a course and the free tier is strong.

This category has no open-source tier at all — and that absence is the finding rather than an omission. CRM, ERP, project management, HRM and ATS each have a credible free substitute that sets a price floor; sales engagement and revenue intelligence do not, because the product is a data network effect over call recordings, CRM history and buyer signals rather than a schema anyone can reimplement. The nearest adjacents are Mautic (GPL-3.0) for marketing automation and SuiteCRM (AGPL-3.0) for the CRM record beneath it, and neither substitutes.

This category has no open-source tier at all — and that absence is the finding rather than an omission. CRM, ERP, project management, HRM and ATS each have a credible free substitute that sets a price floor; sales engagement and revenue intelligence do not, because the product is a data network effect over call recordings, CRM history and buyer signals rather than a schema anyone can reimplement. The nearest adjacents are Mautic (GPL-3.0) for marketing automation and SuiteCRM (AGPL-3.0) for the CRM record beneath it, and neither substitutes.

This category has no open-source tier at all — and that absence is the finding rather than an omission. CRM, ERP, project management, HRM and ATS each have a credible free substitute that sets a price floor; sales engagement and revenue intelligence do not, because the product is a data network effect over call recordings, CRM history and buyer signals rather than a schema anyone can reimplement. The nearest adjacents are Mautic (GPL-3.0) for marketing automation and SuiteCRM (AGPL-3.0) for the CRM record beneath it, and neither substitutes.

This category has no open-source tier at all — and that absence is the finding rather than an omission. CRM, ERP, project management, HRM and ATS each have a credible free substitute that sets a price floor; sales engagement and revenue intelligence do not, because the product is a data network effect over call recordings, CRM history and buyer signals rather than a schema anyone can reimplement. The nearest adjacents are Mautic (GPL-3.0) for marketing automation and SuiteCRM (AGPL-3.0) for the CRM record beneath it, and neither substitutes.

No open-source tier exists, and it cannot. The product is not software, it is a continuously re-verified corpus of people and companies. Code can be given away; the cost here is acquisition, verification and decay — contact data rots at roughly a third a year — so there is no volunteer-maintained alternative and there will not be one. The nearest things are public registers: open corporate registries, OpenCorporates, and government business registers. They carry firmographics but not people, which is precisely the gap the paid field sells into.

Read this category as two layers rather than one market. Layer one owns records — ZoomInfo, Apollo, Cognism, Lusha, People Data Labs — and is commoditising, because waterfall enrichment made any single supplier replaceable. Layer two routes between them — Clay, FullEnrich, BetterContact, LeadMagic — owns no data, carries no acquisition cost, and is where the valuations have gone. An entrant choosing layer one is buying the deflating half.

Two layers, both crowded. Team chat is held by the suite bundles (Teams, Slack, Google Chat, Zoom Team Chat), with four mature open-source servers under them. Customer web chat is older. It ran from the 2000s live-chat widget to the 2016 scripted chatbot to today's LLM agent billed per resolution, and the same vendors (Intercom, Zendesk, Text, Tidio) carried their installed bases through each wave.

WFM has three tiers, and each tier has a different owner. UKG holds the enterprise end, Kronos's installed base of pay rules, union agreements and timeclocks, and it is still buying. Payroll suites (ADP, Paychex–Paycor, Workday, Dayforce) sell time as a module of the payroll the buyer already runs. A crowded SMB tier of venture-funded apps sells to the operator, priced per user or per location, and three of them have free plans.

The open tier is real, and in this category it is the challenger rather than a hobby. A workflow engine is code that calls APIs on a schedule or a trigger, which a community can build and a technical team can host. n8n turned that into a $2.5B company by publishing its source under a licence that lets any business run it internally for free.