PMO: Workflows
What the PMO reads, what it issues, the continuous sense-predict-intervene loop, and the handoffs with delivery, Finance, HR and the client-facing functions.
Markdown. No sign-up, no email.
The PMO does not deliver anything. It sees across everything, which is a different job and depends entirely on the quality of what it can see.
What arrives#
| From | What | Becomes |
|---|---|---|
| Delivery teams | Task movement, or its absence | The schedule signal |
| Finance | Spend against forecast | Budget variance |
| Engineering | Defects and rework | The quality signal |
| HR and resourcing | Who is assigned where | Resource risk, and single-person critical paths |
| Sales and client-facing | Commitments made | Tracked promises |
| Every project | Risks logged, and risks not logged | The register, plus what the signals imply nobody wrote down |
The last row is where the value is. A risk register records what people thought to write down. The interesting risks are the ones visible in the data and absent from the register: three slipped milestones, one person on the critical path, a dependency nobody acknowledged.
What leaves#
| To | What |
|---|---|
| Project managers | The daily prediction, changes only |
| COO | Portfolio position and anything predicted to miss a client date |
| CEO | Kill and re-scope recommendations, with evidence |
| Finance | Forecast spend, revised |
| Delivery teams | Dependency warnings, addressed to the team that owes the work |
The loop#
The last node closes the loop and is the one usually missing. An intervention nobody measured is indistinguishable from an intervention that did not happen.
Handoff contracts#
From delivery teams. The PMO reads their systems rather than asking them for status. Status meetings that exist to populate a spreadsheet are the thing this model removes. What remains is a conversation about what to do, which is a better use of the same half hour.
To the COO. Anything predicted to miss a contracted date goes up the same day. The COO hearing it from the client is a failure of this function regardless of whose fault the delay was.
With Finance. They own the spend numbers; the PMO owns what the trajectory means for delivery. Two functions, one set of figures, and the figures come from Finance.
With HR. Any change to what a person spends their week doing is theirs, not ours. The PMO identifies a resource risk; it does not resolve it by moving someone.
Cadence#
| Continuous | Signal collection |
| Daily | Prediction refresh, ten-minute manager review |
| Weekly | Portfolio review, unacknowledged dependencies, commitments at risk |
| Monthly | Forecast accuracy, false-alarm rate, intervention effectiveness |
| Quarterly | Baseline discipline: how many projects re-baselined, and why |
The failure this design is built against#
A PMO that produces beautiful weekly reporting nobody reads, describing a state of the world that was true on Thursday, while the project that will miss its date in three weeks looks green because the signal that would have shown it was never in the report.
Every element here, continuous sensing, a forecast with its reasoning attached, a ten-minute daily review, a mandatory response within a day, and measuring whether the intervention worked, exists to make that specific outcome hard to reach.