What does first-line resolution measure?
The share of tickets in one category that were resolved at the first line, meaning by a Customer Support Digital FTE's reply from the knowledge base with a citation, without escalation, and without the customer reopening the ticket within the window the team sets. On ftes.ai it is computed per category per cycle in the Performance tab and compared with the baseline from the helpdesk's own history for the same category. It is the primary outcome measure for a support role and the denominator's source for that role's cost per outcome.
It measures answering, not avoiding. That distinction is the reason the platform does not show a deflection rate anywhere.
How is it computed, with the inputs left blank?
Category: [category]. Cycle: [date] to [date]. Tickets received in the category: [N]. Replied by the FTE with a citation, approved or auto-sent within bounds: [C]. Of those, reopened by the customer within [days]: [O]. Escalated to a person, on the always-escalate list or as a knowledge gap: [E]. Resolved at first line: [C] minus [O]. First-line resolution: ([C] minus [O]) divided by [N]. Baseline: the same category's first-line resolution from the helpdesk for the two cycles before the pilot, [rate] and [rate], computed the same way if the helpdesk allows, or labelled as approximate if it does not.
Escalations count in the denominator and not the numerator; they are honest non-resolutions and are measured separately by summary quality.
- Numerator: cited replies not reopened in the window
- Denominator: all tickets received in the category
- Escalations: in the denominator, measured separately
Why per category and not for the whole queue?
Because a knowledge base is strong in some categories and thin in others, and the whole-queue number averages a category the FTE resolves well with one it should not touch yet. Per-category resolution tells the head of support which categories to roll out next, which articles to approve from the FTE's proposals, and which category has earned the Promotion Standard's consideration. It also matches how the always-escalate list works: some categories are entirely escalated by design, and including them in a whole-queue rate would only lower a number that was never meant to include them.
The rollout playbook for first-line support adds one category at a time for the same reason.
Why not deflection rate?
Deflection rate counts tickets that did not reach a person, whatever happened to the customer, and rewards menus, loops, and closing tickets without an answer. First-line resolution counts tickets answered with a cited reply that the customer did not reopen, and treats an escalation as a non-resolution that is measured for its quality rather than punished. A support leader optimising deflection builds the bot customers hate; one optimising first-line resolution per category builds a knowledge base and an escalation path. The myth piece on AI support making customers angry sets out the difference from the customer's side.
The metric shapes the behaviour. Choose the one whose behaviour you want.
How does it mislead?
When the reopen window is too short, so a customer who replies a week later with the same problem is counted as resolved; set the window from the team's own history. When categories are drawn too widely, so a resolved order-status ticket and an unresolved returns dispute sit in one number; draw them as the knowledge base is organised. When measured only in a quiet month, since first-line resolution that holds in a quiet month and collapses in a peak has not been measured; the peak-season planning guide in the library schedules the measurement through the season. And when a rise coincides with a fall in escalation quality, which means tickets that should have escalated were answered; read both.
Satisfaction is a separate measure, shown only where the helpdesk surveys it and marked unmeasured otherwise.
How is it applied?
Record the baseline per category from the helpdesk before the pilot. Choose the first category by the First-Function Matrix. Read first-line resolution weekly for that category alongside escalation quality and the send-back reasons. At the monthly review, write the per-category rate against the baseline in the note, decide the next category, and use the rate with the approval record as inputs to the Promotion Standard for that category. Measure through the first peak before drawing any conclusion the head of support would repeat to a board.
The head-of-support persona piece lists this as the first of the three measures to track from week one.