The decision is usually framed as a cost comparison — a monthly fee against a payroll line. That framing is convenient and it is wrong, because it compares the two options at a point where neither has delivered anything yet.
What a PMO actually has to produce
Before comparing delivery models, be specific about the output. A functioning PMO produces a small, concrete set of things:
- A prioritised portfolio, with a defensible basis for the order
- A stage gate model with real entry and exit criteria
- One project management standard and a template set people actually use
- An integrated portfolio schedule and a CAPEX tracker finance trusts
- A monthly portfolio dashboard and project health review
- An aggregated risk view across projects
- A Steering Committee that makes decisions instead of receiving updates
Both models can produce that list. They differ in how long it takes, what happens when the portfolio changes shape, and where the knowledge ends up.
A structured comparison
| Dimension | Internal PMO | PMO as a Service |
|---|---|---|
| Time to capability | 9 – 18 months including recruitment and standards development | 4 – 8 weeks to first governance cycle |
| Cost shape | Fixed payroll and overhead, hard to reduce | Monthly fee, scalable with the portfolio |
| Peak load | Overloads or requires contractors | Capacity added for the peak, removed afterwards |
| Portfolio contraction | Structure survives the work it was built for | Service scaled down or paused |
| Sector method | Built from scratch or bought as a framework | Arrives with an industrial template and risk library |
| Institutional knowledge | Stays inside, concentrated in a few people | Must be deliberately handed over — insist on it contractually |
| Political authority | Higher: an internal PMO director has standing | Depends on sponsorship; needs an internal owner |
| Independence in reporting | Can be pressured by internal hierarchy | Structurally easier to report bad news |
Where each one wins
Build internally when the project portfolio is permanent and central to the business model — a company whose entire revenue comes from executing projects should own that capability, full stop. Build internally when the volume is stable and large enough to keep specialist planners and cost controllers genuinely occupied, and when project management is a career path you want to offer.
Contract the service when the portfolio is cyclical, when you need governance running this quarter rather than next year, when the volume does not justify permanent specialists, or when the honest internal assessment is that a PMO would be built by people who have never operated one. It is also the right answer for a first PMO: run the service for twelve to eighteen months, let the standard prove itself against real projects, then decide what to internalise.
The hybrid that usually wins
The most durable arrangement is rarely all of one. An internal PMO lead — one person, senior, with political standing — owns prioritisation, the relationship with the executive team and the final call on gates. The service supplies the operating machinery underneath: planning, cost, risk consolidation, dashboards, reporting and the discipline of running the cadence every single month.
Whichever model you choose, write the handover into the contract from day one: templates, registers, dashboards and documentation built in your systems, under your licences, transferred on exit. A PMO service that cannot be handed back is a dependency, not a capability.