How to move a process in-house without it collapsing the month the agency leaves.
Written for Teams who want to run their own marketing and have tried once before.
The bottleneck
Handover usually means a training session and a document. Three months later the process has stopped, because nobody accounted for the fact that the client team has other jobs and no habit of doing this one.
Capability transfers when the receiving team does the work while support is still available, not when they are shown how it is done.
Why it happens, five levels down
- The in-house process stops a few months after handover.Why? It was never actually run by the client team.
- They were trained but did not practise.Why? Handover was scheduled as an event at the end.
- It was treated as a final deliverable.Why? The commercial model rewards delivery, not transfer.
- Transfer is not what the engagement was sold on.Why? Nobody defined independence as a success measure.
- Independence was never a deliverable.Root cause. Make it one and the engagement is designed differently from day one.
The diagnostic
Run this before planning any work. It tells you whether this is your constraint or whether something else is.
- Ask who will do this work and how many hours a week they genuinely have.
- Check whether approval chains would block the cadence you are proposing.
- Confirm the team has access to every tool in their own names.
- Check whether any documentation exists that someone could follow without asking a question.
- Identify what happens when the trained person leaves.
- Ask the team what they think their job is. Compare with what you assumed.
The build sequence
| Stage | Work | Timing |
|---|---|---|
| Assess | Skills, available hours, tools, approvals and the minimum viable rhythm. | Week 1 |
| Document | Write the workflow so a new person could follow it unaided. | Weeks 2–3 |
| Practise | Client team does the work; agency reviews. Not the other way round. | Weeks 4–10 |
| Transfer | Agency moves to review only, then to on-request. | Weeks 10–12 |
| Measure | Track independence: what they complete without help. | Ongoing |
KPI watchlist
| Metric | Source | Read it as |
|---|---|---|
| Publishing or delivery consistency | Their own output log | Whether the habit holds |
| Tasks completed without help | Review log | Actual independence |
| Time per task | Their record | Whether the process is realistic |
| Process survival at six months | Follow-up | The real success measure |
Record a baseline and a date range for each before any work begins. Reconstructing a baseline afterwards is usually impossible.
Four ways this goes wrong
- Handover as a single training session.
- Documentation written by someone who assumes context.
- Training one person with no backup.
- Measuring handover by deliverables shipped rather than work they now do.
Design the handover at the start
Independence is a deliverable or it does not happen.
