Running an open-source CRM: the honest experience

We run our pipeline on Relaticle, an open-source CRM we self-host, and the honest summary after living with it is this: the licence is free, the ownership is total, and the running cost is real — paid in maintenance attention rather than per-seat fees. For a firm with technical capability in-house or on tap, the trade is clearly favourable. For a firm with neither, a mid-market SaaS CRM remains the safer default, and pretending otherwise would be selling you my preferences.

Why did we choose open source at all?

Three reasons, in decreasing order of importance. Ownership: the CRM is our commercial system of record — every deal, contact and conversation trail — and we wanted that database on infrastructure we control, exportable at will, with no vendor able to change the pricing or the terms underneath it. Cost shape: SaaS CRM pricing scales per seat, so headcount growth silently inflates the tooling bill; a self-hosted instance costs the same whether three people use it or fifteen. Malleability: we build systems for clients, and running an open platform ourselves means custom fields, pipelines and API integrations are engineering decisions, not feature requests into a vendor's void.

What we did not choose it for is the reporting layer alone — any competent CRM can feed the self-compiling instrument panel described in The MD Dashboard Blueprint. Open source changes who owns the machine, not what the machine must do.

What does it actually cost to run?

The truthful accounting has three lines. Hosting: modest — a small server, typically single-digit pounds per month at our scale. Maintenance: the real line. Updates need applying, backups need running and — crucially — need testing, and the occasional breakage lands on us rather than on a vendor's status page. Averaged out, it is measured in hours per month, not per week, but those hours are non-negotiable; skip them and you are one failed disk from losing the company's memory. Capability: someone must be able to do the above. We are a systems firm, so this is Tuesday; a 12-person consultancy with no technical staff would be renting that capability or gambling.

Against that sits the absent line: per-seat licence fees, which for a growing team on mainstream SaaS pricing commonly run to thousands of pounds a year. The full feature-by-feature and cost comparison is laid out in open-source CRM vs HubSpot; the short version is that the money you save is converted into responsibility, not eliminated.

How does the day-to-day compare with SaaS?

Mostly indistinguishable, which is itself the finding. Contacts, companies, deals, pipelines, tasks, notes — the working surface of a CRM is commodity functionality now, and Relaticle covers it cleanly. The differences appear at the edges. Integrations: SaaS platforms have marketplaces; we have an API and build what we need, which suits us and would frustrate a team expecting one-click connections. Polish: open-source projects improve in visible increments, and occasionally an edge case is simply ours to work around. Support: there is no account manager; there is the project's community, the codebase, and our own competence.

One under-reported advantage: because we own the schema, the CRM bends to the business rather than the reverse. Custom pipeline structures that would be awkward in rigid SaaS tiers are straightforward — a point that matters for odd-shaped sales motions like the two-sided pipeline recruiters run, described in outbound for recruiters, where candidates and clients need parallel tracking.

Who should follow us — and who shouldn't?

The mechanism for deciding is short. When your firm has in-house technical capability, or a systems partner who carries maintenance for you, then open source converts licence fees into owned infrastructure and is worth it. When your data-ownership or privacy requirements make third-party hosting uncomfortable, then it moves from option to default. When neither applies — no technical capability, no partner, no unusual requirements — then choose boring SaaS, because a CRM that breaks with nobody to fix it costs more than any licence. And whichever side you land on, the deciding factor for the CRM project overall remains adoption, not software: the habits and definitions travel with you either way.

What would I do differently?

Two things. I would write the backup-restore drill down as a procedure on day one rather than week three — a backup you have never restored is a hope, not a backup. And I would resist customisation for longer: the temptation with an open platform is to bend it early and often, when the discipline that actually pays is running plain-vanilla until the process itself has settled. The CRM, however owned, is still only the substrate. What compounds is the numbers discipline built on top — knowing what a client is worth over their lifetime, the arithmetic set out in lifetime value: the number that justifies the system, and reading the machine on a fixed rhythm of daily glance, weekly review, monthly close. Those work identically on any CRM. Ownership just means nobody can take the substrate away.

---

Next step: the Growth System Audit — £450, seven days, credited against any build — maps where your growth system leaks, including whether your CRM setup fits your firm, and what to build first.