COLUMN

In-house PMO training starts,
but the send-back decision still sits outside

You want to stop the outsourced monthly fee and put a PMO in-house. You buy training. Next month, the same send-back is still there. What you cut first is not the lecture: who knows last time, is there a single approver, and who will use this plan for what.

Separate what can be handed over, starting from the send-back →

2026.09.16 #InHousePMO #Internalization #Training

Written by Kazuo Yutani, Representative of Metamorphose LLC. As a PMO, I have watched plans get sent back and then handed over to in-house teams.

The order for building an in-house PMO isn’t set by how much people know

You want to bring down the monthly fee paid outside. You want a PMO in-house. The first purchase is a training course. The vocabulary lines up. Templates get distributed. Next month, the plan comes back anyway.

What you line up first is who receives last time’s history, a single final approver, and what the deliverable is for. That doesn’t mean nothing else matters. But seat someone in a role where these three are missing, and the send-backs tend to survive even a correct process.

Three common ways in-house PMO training gets introduced — and doesn’t take

Common approach	What is still there next month
Line up certifications and vocabulary first	Meeting language grows. The number of decisions cut often doesn’t
Distribute one company-wide WBS template	The floor can’t find its own row. The approver loses the storyline
The external PMO leaves and hands over a document set	Why it stopped last time never arrives. Only surface updates remain

The third one tends to happen the moment internalization is announced. The handover pack is thick. But if “who do I ask to learn what happened last time” is blank, the new PMO fills it in by imagination.

Three things to hand an in-house PMO before any training

Before the lecture, check whether these three can be handed over. If they can, fix a live plan together. If they can’t, fill them in-house first. What can’t be filled is what stays outside.

Someone who knows last time: can they be asked directly? Are they still on the project?
A single final approver: whose approval makes it final? If it is still a committee, the person being developed gets caught in the middle.
Who will use this plan, for what: can you say “if this is achieved, it succeeded”?

Hand over from the unit — a team or a phase — where all three can be answered on the spot. Don’t try to internalize everything at once. It is the same cut as deciding whether to add an external PMO by headcount.

In-house PMO OJT can start from one live send-back

Passing is not “attended the course.” The benchmark is that the person can produce one plan where the next decision to cut can be identified inside the meeting. One plan doesn’t mean done. If questions aim at gaps in explanation, the scope or the reader is off.

The plan you fix together doesn’t have to be a new project. Start with the one plan that is being sent back right now. Reproducing, in an in-house person’s hands, the same pattern used to stop a plan being sent back — that is close to the first pass mark for internalization. It is not the only definition of internalization.

Internalize from the unit where all three can be answered

Don’t hand everything over at once. A unit where approval is split, or where last time’s history can’t be reached, is cut first or stays outside for now. Hand over from where the three can be answered.

“Run it in-house only from next year” can be the goal. It shouldn’t be the first benchmark. The first benchmark is one person, one unit, cutting the send-back axis by themselves. Not necessarily within 30 days.

CONTACT

We look at whether the plan being sent back right now has a unit you can hand in-house.

Is what’s missing knowledge, or a recipient and an approver? That is where the conversation starts. That single line is the entry point to our Metamon diagnostic. No plan document or training curriculum needed. One line on the form (a team name is enough if you’re not sure).