What changes when the desk moves?
An outsourced desk answers everything at first line from a mix of the knowledge base and the agents' accumulated knowledge, escalates what its contract says, and reports on volume and response time. A Customer Support Digital FTE on ftes.ai answers only what the knowledge base answers, with a citation, escalates a fixed list with four-part summaries, and reports first-line resolution per category. The migration moves categories, not the whole queue, and the biggest change is that the knowledge the desk carried in people's heads has to be written down.
The comparison piece on a Support FTE against outsourced support sets out the division; this guide is about the move.
What should you keep with people?
Every category on the always-escalate list: complaints, refunds and credits, account access, legal or safety language, distress, and anything about a specific customer's money. Keep people for the escalations the FTE routes with summaries, and decide whether those people are the outsourced desk's escalation tier or an in-house team; most companies find the migration is the moment to bring the hard cases in-house, because they arrive with context now. Keep the desk running for the categories not yet moved, and keep it through the first peak if the contract allows.
The head-of-support persona piece describes what the in-house team's day becomes.
What moves to the FTE, and in what order?
First, record the baseline per category from the helpdesk: first-line resolution, first response time, and volume, plus the desk's cost per ticket from the contract. Second, choose one high-volume, low-risk category with reasonable knowledge base coverage, confirm its articles, write the escalate list with receivers, and connect the helpdesk with a dedicated scoped account, routing only that category's tickets to the FTE while the desk keeps the rest. Third, run it gated for a month, approving cited replies and approving proposed articles for the gaps. Fourth, at the first review, if the category's record is clean, move the next category; consider auto-send for the first under the Promotion Standard. Fifth, repeat until the desk's remaining scope is the escalate list or nothing.
The thirty-day first-line support playbook runs the first category week by week.
- Baseline per category from the helpdesk and the contract
- One category to the FTE, the desk keeps the rest
- Next category after each clean review
What breaks in the first month?
The knowledge base. The desk's agents knew things the articles do not say, and the FTE escalates every ticket the articles cannot answer as a gap rather than guessing. The first weeks produce a stream of gap escalations, each with a proposed article, and the support lead spends time approving articles that should have existed for years. This is the migration's real work and its durable benefit; by the third week the gaps thin. Tone also breaks: the desk had a house style the Brand Brain did not capture, and the send-back reasons in week one are about it. And the contract may break: some desks charge per ticket and object to categories leaving, so read it before the pilot.
Nothing that breaks is the FTE failing. It is the knowledge base being made honest.
What does the record show after a peak?
First-line resolution per migrated category against the desk's baseline for the same category, measured through the first peak, not a quiet month. First response time including overnight. Escalation quality, the share of summaries that met the standard, which the desk never reported because it did not produce summaries. Knowledge gaps closed, as articles approved and first-line resolution in their categories afterwards. Cost per resolved ticket from the role's usage against the desk's contract rate. And an in-house team, if you built one, whose day starts from summaries.
The metric explainer on first-line resolution describes why it is measured per category and through a peak.
What should you refuse during the migration?
Moving several categories at once, or the whole queue, because the knowledge base cannot absorb the gaps and the team cannot approve the articles. Promoting a category to auto-send during a peak. Letting the desk's escalation tier receive escalations without the summary standard; they should receive summaries too, and the contract should say so. Measuring on a quiet month. And ending the desk's contract before the last category is settled and the first peak has passed, because the desk is the migration's safety net.
The myth piece on AI support making customers angry describes what a rushed migration produces.