Ninety Days to Cutover: Running a Transition Without Burning Your Best People

Ninety Days to Cutover: Running a Transition Without Burning Your Best People

Transition is the phase everybody underestimates. The contract gets negotiated for four months and the knowledge transfer gets six weeks, which is roughly the inverse of where the risk actually sits.

Here is the thing nobody tells you at signature.

The people who will carry your transition are the same people currently holding your operation together. There is no separate transition team. There are your three best agents, your one supervisor who knows why the exceptions exist, and a process document last updated in 2023 by somebody who has left.

What follows is a working sequence, counted from cutover.

Week minus twelve: find out what you actually do

The object here is not the process map but the process. Those diverge in every operation that has been running longer than a year, and the divergence is where your transition risk lives.

Sit with the queue for a full cycle — a week minimum, a month if the work has monthly rhythms — and write down every decision point, including the ones that are not in any document. The address-format workaround. The customer type everyone knows to escalate. The three accounts that are handled differently for reasons nobody can now explain.

This is the single highest-value activity in the entire transition and it is almost always compressed, because it produces no visible artifact until the end and feels like administration.

A practical test: hand your draft documentation to somebody from a different department and ask them to work three cases from it. Wherever they stop and ask a question, you have found a gap that would otherwise surface in week two of live running as a quality failure.

See also: How to Decide Which Vacancies Deserve Your Attention

Week minus ten: decide what is moving

Scope drifts in both directions during transition, and both directions cost money.

Scope that expands — “while we’re at it, they could also handle the email queue” — arrives without a corresponding extension to the timeline or the training. Scope that quietly contracts, usually because a system access turned out to be harder than expected, leaves work stranded with a team that has already been told it was leaving.

Experienced providers of Philippine outsourcing services will ask for this list before they quote, because a scope expressed as a department name is one they will be renegotiating in month four. Write it down as contact types and work items instead. Then write down what is explicitly not moving, which is the half that prevents the argument in month four.

Week minus eight: settle the authority schedule

The question this answers is what the receiving team may decide, and up to what value.

This document is short, it takes an afternoon, and programs that skip it spend their first quarter escalating everything because nobody wants to be the person who guessed. The symptom is a service level that looks fine and a client-side manager whose calendar has filled with approvals.

Set thresholds by transaction type. Name who approves above them. Agree how a threshold gets changed, because it will need changing once real volume arrives and the first estimates turn out to be wrong in both directions.

Week minus six: train for the exception, not the flow

Standard induction covers the happy path well. The happy path is also the part most likely to be automated within eighteen months, which makes it the least valuable thing to spend classroom hours on.

Build the curriculum around what goes wrong. The partial failure. The customer who is already angry from a previous contact. The case where the system says one thing and the document says another. Experienced providers will push for this themselves; less experienced ones will deliver the induction they already have and let the floor discover the rest.

Reporting on how AI is reshaping outsourcing services and their staffing describes a shift that makes this more urgent than it was: as routine contacts resolve without a person, the work landing on a new agent in their opening week is tougher than it would have been five years ago. Training designed for the old mix produces people who are confident on work that no longer reaches them.

Week minus four: run the systems, not the slides

Access provisioning is the most reliably late item in any transition, and it fails quietly because it fails one person at a time rather than all at once.

Test it properly. Have every named agent log into every system they will need, complete one real transaction end to end, and record how long it took. You will find at least one application where the offshore team has read access and needs write, and you would rather find it now than at nine o’clock on cutover morning.

Run the escalation path the same way. Send a test escalation through and see where it lands. The most common avoidable failure in week one is an escalation route that was configured correctly and connected to a distribution list nobody monitors.

Week minus two: agree what week one means

Cutover week will not resemble steady state, and the plan should say so in writing rather than leaving everyone to discover it.

Volume at a deliberately reduced level. Every interaction reviewed, not sampled. Daily calibration between the two quality teams rather than weekly. A named person on both sides who can make a decision without convening anybody.

Set the service level for week one at something achievable and say out loud that it is a ramp target rather than the contractual one. Teams held to a steady-state number on day three learn to protect the number instead of learning the work.

Cutover: go when the gates are met

Volume should step up on evidence rather than on the calendar, which sounds obvious and is routinely overridden by a contract date.

Define the gates in advance — quality score on reviewed interactions, escalation accuracy, documented competence on specified transaction types — and hold to them. A program that slips a fortnight because a gate was not met costs a fortnight. A program that goes to full volume on schedule with a team that is not ready costs a quarter, and the repair is conducted in public, in front of the customer.

Week plus four: the review that does not happen

Four weeks after cutover, while everyone can still remember, write down what the documentation missed.

Every transition surfaces knowledge that existed only in somebody’s head. That knowledge is now in a chat thread, a hallway conversation and three email replies, and within a month it will be gone again. Capturing it is an afternoon’s work, and it decides whether your knowledge base reflects the operation or the operation as somebody imagined it during a workshop in week minus eleven.

What actually goes wrong

It is almost never capability. In most failed transitions the receiving team was perfectly able to do the work.

What fails is that the work was never fully described, the authority was never settled, and the timeline was set by a contract date rather than by readiness. Those three are decided before anybody takes a call, which is why transition is a planning discipline wearing an operational costume — and why the programs that run smoothly in year two were almost always the ones that looked slow in week minus ten.