What does a Customer Support FTE own?
A Customer Support FTE owns first-line resolution inside the helpdesk you already run. It is accountable for resolving tickets it can answer from your knowledge base, fast and with sources, and for escalating everything else with enough context that the human picking it up does not start from zero.
It does not own the whole queue. Refunds, account changes, complaints, and anything with a person's money or feelings at stake are routed to a human, with the FTE's summary attached.
How does it answer a ticket?
The Customer Support FTE reads the ticket, searches the connected knowledge base and any approved documentation, and drafts a reply that cites the specific article or policy it used. If it cannot find a source, it does not guess; the ticket is escalated with a note that the knowledge base has no answer.
Citing sources is what makes replies reviewable. A manager approving a draft can see in seconds whether the FTE used the right article, and a customer receiving the reply can follow the link.
What happens with escalations?
An escalation arrives to the human team as a summary rather than a raw ticket: what the customer asked, what the FTE found, why it did not resolve, and a suggested next step. The human starts from the FTE's work rather than from scratch.
Escalation categories are set in the remit and can be tightened at any time. Anything involving money, account access, a complaint, or strong emotion escalates by default, regardless of whether the knowledge base has an answer.
- Resolved: answered from a cited source, logged with the article used
- Escalated: summary, findings, reason, suggested next step attached
- Gap flagged: no source found, knowledge base item proposed
- Always human: refunds, account changes, complaints, distress
What can earn autonomy?
The Customer Support FTE starts at Level 0, drafting every reply for approval. The first class usually promoted is the high-volume, low-risk category, such as password reset instructions or shipping status, where the approval history shows a long run of unchanged drafts.
Autonomy is per category, not blanket. A category can be promoted to auto-send while every other category stays gated, and the manager can step any category back with one change. Anything touching money or accounts is never promoted.
How does it learn?
The Customer Support FTE learns by recording what resolved each ticket and by flagging gaps. When several tickets escalate for the same missing answer, it proposes a knowledge base article for approval, so the next ticket of that type resolves at first line.
Its diagnostic missions work the same way: why is ticket volume rising, answered with ranked hypotheses from the ticket data, such as a product change generating confusion or a policy that customers keep misreading, each with evidence attached.
What does a Customer Support FTE cost, and what does the support lead see?
A Customer Support FTE consumes tokens for reading tickets, searching the knowledge base, drafting replies, writing escalation summaries, proposing articles, and its brief, bounded by a per-FTE budget and visible daily; approving replies costs nothing. Because ticket volume moves with the business, the cost follows tickets actually handled rather than a fixed headcount, and usage per role sets the tokens behind each first-line resolution against the baseline cost of resolving it by hand.
The support lead sees each morning the replies waiting for approval with their cited sources, the escalations that reached the team with summaries, the categories with recurring gaps and the proposed articles for them, and spend against budget. In the Performance tab they see first-line resolution rate, median first response, escalation volume by category, approval rate with send-back reasons, and usage, and once a month the FTE's own report. The volume diagnostic mission answers why tickets rose with ranked hypotheses from the ticket data, which is the question a support review usually answers with a guess.