The Monthly First Contact Resolution Review

The first Monday of the month, 08:30, at Acme. Jonah runs a desk of four and has the counting rules sheet open beside last month's export. The figure for the month is 176 of 402 tickets, a little under half. The month before it was 214 of 410, a little over. It has fallen by the width of a good week. His first instinct, which he has learned to notice and distrust, is to write an e-mail to the team. It begins "I've noticed first contact resolution has dropped" and ends with the word "focus". He does not send it. He has watched that e-mail produce, over the following month, three technicians closing faster and nobody knowing why the figure fell in the first place. He has watched the figure go back up for reasons that were not improvements.

Instead he books thirty minutes with the team for Wednesday and starts the review. The monthly first contact resolution review is a short, fixed routine for reading the month's figure with the numbers that keep it honest. It answers three questions: what moved, exactly; why; and what one thing changes next month. It produces one action with an owner and a sentence for the page that goes upward. It does not produce a league table, an e-mail about focus, or a target. This article is part of our First Contact Resolution series, and it is the last in it. It assumes the counting rules from Measuring First Contact Resolution Honestly. It assumes the per-category table from First Contact Resolution by Category. It also assumes the paired-metric check from The Limits of First Contact Resolution. It shows how they are used together in half an hour a month.

It covers what the review is for and who is in the room, then the checks before the number is read. Reading the number with its pairs, the three questions in turn, and the actions and their owners follow. Next comes writing the review down and passing one sentence upward. A worked review at a fictional desk, a short war story, and the review sheet to copy follow, in that order.

What the Review Is For

The review exists to turn a number into one decision, and everything about its shape follows from that. It is thirty minutes, because a longer meeting about one figure becomes a meeting about everything. It is the desk lead and the technicians, because they are the people who will do the one thing that comes out of it. The twenty tickets the review reads were also theirs. It is not the leadership meeting, which gets one sentence from it. The one-page report that carries that sentence upward, and what else goes on the page, belongs to The Monthly Report They Will Read. This review is where the sentence is written.

It is a review, not a scoreboard, and the difference is in what the room looks at. A scoreboard shows a figure and invites a feeling about it. A review shows a figure beside the figures that explain it, then opens tickets. Nobody leaves a good review feeling praised or blamed. They leave knowing which twelve tickets could not be finished last month and what would have let them be finished. That is more useful and considerably less dramatic. If the review ever becomes a place where people feel judged by the number, the number will start being managed in the room's favour within a month. The war story below describes that with names.

The review also has a fixed place in the month. The figure is final two weeks after the month ends, when every reopen window has closed. So the review sits in the third week. That is late enough that the number is real, and early enough that the one change can land before the next count. A review held on the first of the month is reviewing tickets whose reopen windows are still open, and will be corrected by the following week's reopens. (It will not be thanked for the correction.)

Before the Number: Same Rules, Same Window

The first two minutes of the review are spent not reading the figure, and they save the other twenty-eight. Check that the counting rules sheet has not changed: same reopen window, same date rule, same denominator with the same two exclusions, same walk-up logging. If any line changed, the figure is not comparable with last month's and the review's first question is already answered. Then check the volume. A month with 402 tickets against one with 410 is comparable. A month with 260 is a different month, and a share drawn from it moves more easily. Then glance at the per-category mix against last quarter. A month that was light on password tickets and heavy on hardware will have a lower figure for reasons that have nothing to do with the first line.

These are the three ways a figure moves without anything happening at the desk: the rules drifted, the volume changed, or the mix shifted. Each takes a minute to rule out and each, when it is the cause, ends the review early with an honest sentence. Jonah's rules had not changed and his volume was level. His mix had shifted, a little, and he noted it for the first question rather than letting it answer the whole review on its own.

Reading the Number With Its Pairs

The figure is then read beside its partners, on one page, and never before them. The reopen rate says whether the fixes held. The satisfaction score says whether the requesters agreed. The escalation rate and resolution time by category say whether the first line is handing off when it should. The pairing idea belongs to Reading Metrics Together. The particular readings for this number, with a grid of what each combination usually means, are in the previous article. The review borrows the grid and applies it in one sentence: "the figure went down, reopens went down, satisfaction went up". That is a very different month from "the figure went down, reopens went up, satisfaction fell".

Jonah's month read as the first of those: the figure down, reopens down slightly, satisfaction up a little, escalations up. From the grid, that is a desk taking the second contact when the problem needs it, or handing off more than before. It is not a bad month until the tickets say otherwise. The e-mail about focus would have been written to a desk that had, as far as the pairs could tell, done the right thing more often. This is the moment the review earns its thirty minutes.

Question One: What Moved, Exactly

A desk-wide figure almost never moves everywhere at once. It moves in one or two slices and the rest hold steady, and the first question is to find the slice. Cut the month's figure by category first, against the per-category expectation table, and look for the row that left its range. Then by channel, because a desk whose chat figure fell and whose e-mail figure held has a chat problem, not a desk problem. Then by week, because a fall that began in the third week began with something that happened in the third week. The answer to "what moved" is a row, a channel or a week, and usually only one of them.

Only then, and carefully, by technician. The per-person cut is read for coaching and never for ranking. It is not read for anyone with fewer than forty tickets of a comparable kind in the month, because below that the figure is two unlucky tickets wide. At Acme that rule excludes one of the four technicians most months, which is fine; the cut is a diagnostic, not a census. When one person's figure has moved and the others' have not, the question is not "what happened to you". It is "which of your tickets could not be finished, and what stopped you". The answer is nearly always a new kind of ticket landing in one queue. People's figures move when their work changes. They rarely move because the person did.

Jonah's cut found the slice in two minutes. Application errors had fallen from somewhere around half to about one in four. Every other category sat inside its range, the channels were level, and the fall began in the second week. One of the four technicians, Wen, had a lower figure than usual. She had also taken most of the application tickets, because she sat nearest the finance team. The slice was a category and a fortnight. The person was where the slice had landed.

Question Two: Why

The second question is answered by reading tickets, not by discussing them. Take the slice from question one and open twenty of its tickets that were not resolved on first contact, at random, with the room watching. Write one word beside each: knowledge, permission, tool, information, or mix, meaning a kind of ticket that could never have been first contact. Twenty is enough for a pattern to show and few enough to read in ten minutes. If the word is the same for twelve of the twenty, the review has its answer. If the words scatter, the honest answer is "we do not know yet" and the action is to read forty more before the next review. That is a perfectly good action.

The reading goes quickly when every handoff and reopen is on the ticket itself, as a line in its history. In that case, "who had this, and when did it change hands" is a glance rather than a reconstruction from memory and e-mail. QuerySolv keeps that history on every ticket, as most tools do. A desk whose tool does not will find question two takes most of the half hour. The reading is also where the review's tone is set. The technician whose ticket is on the screen should hear "what would have let you finish this one?" and nothing else. The ticket is the subject. The person is the witness.

Jonah's twenty were nineteen "knowledge" and one "mix". A new finance application had gone live in the second week. Nobody on the desk had seen it before, and every ticket about it had gone to the finance team's own application administrator. That administrator fixed most of them by resetting a user profile the desk could have reset had it known the steps. Wen had escalated eleven of them, each with a note saying so. The figure had fallen because a kind of ticket arrived that the first line had never been given the means to finish. That is the most common reason a first contact figure falls and the one the e-mail about focus would have missed entirely. (Focus is not a runbook. It has never once reset a profile.)

Question Three: What One Thing Changes Next Month

The third question has a one-word budget, and the word is "one". A review that produces five actions produces none, because five changes in a month cannot be told apart in the next count. The desk, which has a queue to run, will do the easiest of them and feel it has done the review. One change, with an owner and a date before the next count, is a thing that can be done and whose effect can be seen. The change should attack the cause from question two and nothing else. That could be a runbook for a knowledge gap, a right requested for a permission gap, a template or script for a tool gap, or a form field for an information gap. For a mix shift, it is a sentence on the page saying so and no change at all.

The change is also written so that the next review can tell whether it landed. "Write the runbook" is a wish. "Runbook for the finance application's profile reset, written by Wen with the application administrator, published to the agents-only knowledge base by the twentieth, and the first line given the right to perform the reset" is a change. Next month's review can open the application-error row and see whether it climbed back towards its range. If it did, the sentence writes itself. If it did not, the next review reads twenty more tickets and finds out why, which is also a result.

Three things are not the one change, and should be refused when they are proposed, which they will be. An e-mail to the team about the figure is not a change; it is a mood. A target for next month is not a change; it is a request for the behaviours the fifth article in this series lists. A new metric is not a change either, however tempting it is to measure the problem harder. The review has found a cause. The change removes the cause. Everything else is a way of feeling busy about it.

The Actions and Their Owners

The review opens by closing last month's action before it reads this month's figure. This small discipline is most of what makes the routine work. Did the runbook get written? Did the right get granted? If yes, the current figure is read in that light. If no, the action carries over and this month gets no new one. An action that did not happen in a month will not happen alongside a second. One carry-over is ordinary. Two in a row is a sign the action was too big or the owner too busy, and the review's job is to shrink it, not to add to it.

The action record is three lines: the change, the owner by name, and the date. Next month, one of three words follows: done, carried, or dropped, with a reason. Keep the lines on the review sheet rather than in a separate tracker. That way, the history of the figure and the history of what was done about it sit on the same pages. A year of review sheets read together is the best account many desks will ever have of why their first contact resolution is where it is. It was written fifteen minutes at a time by people who were mostly trying to get back to the queue.

Writing It Down and Passing One Sentence Up

The review sheet is the record, and it is filled in during the review, not afterwards from memory. The figure with its counts, the pairs, the slice, the cause word and the ticket numbers read, the one change with its owner and date, and the sentence for the page. The sentence is the only part that leaves the desk. It is written in the form the previous article set: the figure, the pairs, what it means, and what changes. Jonah's was "first contact resolution 176 of 402, a little under half, down from a little over; reopens down, satisfaction up; the fall is entirely in application errors after the finance system went live and the first line had no runbook for it; runbook and reset right in place by the twentieth." A director who reads that knows what happened and what is being done. A director who reads "FCR fell to 44 percent" knows only that something is wrong and that nobody has said what.

The sentence goes on the one-page monthly report, beside the other desk figures and their sentences. The full sheet stays with the desk for anyone who asks. The team's dashboard, which in QuerySolv as in most tools shows what is waiting and what is overdue right now, is not where this lives. The dashboard is for the present and the review is for the month that just ended. A dashboard that tries to carry a monthly figure with names beside it becomes the wall at Pinecrest in the story below. The reasons dashboards should motivate rather than monitor, and how to keep one honest, are in Dashboards That Motivate Instead of Surveil. The review sheet is a document. Documents are harder to resent.

A Worked Review at Acme

Here is Jonah's review in full, with Acme's numbers invented for this article. The figure: 176 of 402, a little under half, from 214 of 410. The rules check: unchanged. Volume: level. Mix: application errors up from about one in twelve of the month's tickets to about one in six, which accounted for part of the fall but not most of it. Pairs: reopens down a touch, satisfaction up a touch, escalations up by about thirty tickets, nearly all in application errors. The grid's reading: the first line is handing off a kind of ticket it cannot finish, and doing so correctly. Question one: the application-error row, from the second week, landing mostly on Wen. Question two: twenty tickets read, nineteen "knowledge", a new finance application with a profile-reset fix the desk had never been shown. Question three: one runbook and one right, owner Wen with the application administrator, by the twentieth.

The following month's review opened by closing the action: done on the eighteenth. The figure was 205 of 396, a little over half. The application-error row had climbed back to about two in five, within its range for a category the desk had only just learned. Reopens were flat, satisfaction flat, escalations down by about the same thirty. The sentence for the page said so in two lines. The one change for that month was something else entirely, because the review had finished with the finance application and moved on. The whole exercise cost Acme two half-hours and a runbook. The e-mail about focus would have cost less and bought nothing.

A Short War Story: The Wall at Pinecrest Schools

At Pinecrest Schools, a trust of six schools with a desk of five, the IT manager Amara started a monthly first contact resolution review in good faith. After the third one, Amara moved its output to a whiteboard in the office so the team could "see progress": five names, five figures, updated monthly. The following month a technician called Ben, who covered the smallest school, had handled twelve tickets of a comparable kind. He showed a figure of about one in three against his colleagues' one in two. He was spoken to about it, gently, by a deputy head who had wandered past the board. Within two months two of the five technicians had worked out the fastest way up the board. It was to let hard tickets sit in the shared queue until someone else claimed them. The oldest unowned ticket in the queue went from a day to nine days. The desk-wide figure rose while the trust's head of operations began asking why the schools were complaining more. The cost was a quarter of avoided tickets, a technician who asked for a transfer, and a review that had to be rebuilt as a document after the board was wiped. A review that puts names beside a number on a wall stops being a review. It becomes a game, and the people who are good at the game are not the people who are good at the job.

The Review Sheet

The sheet below is one page per month, filled in during the thirty minutes and kept with the previous months'. Section A closes last month's action. Section B is the checks before the figure. Section C is the figure with its pairs and the grid reading. Sections D to F are the three questions, with the ticket numbers read written down so that anyone can check the cause. Section G is the one change and the sentence for the page. If a section cannot be filled in, write why; a sheet with "not read, volume below floor" in section C is a complete sheet. The sheet is the review's whole output, and a year of them is the desk's memory.

FIRST CONTACT RESOLUTION — MONTHLY REVIEW SHEET
Desk: ______________________   Month reviewed: ____________   Review held on: ____________
Present: ______________________________________________

A. LAST MONTH'S ACTION
   The change: ______________________________________   Owner: __________   Due: __________
   Outcome: done / carried / dropped     Reason if not done: ______________________________

B. BEFORE THE NUMBER
   [ ] Counting rules sheet unchanged (window ____ days, by ______ date, same exclusions)
   [ ] Volume this month ____ tickets, last month ____ (comparable: yes / no)
   [ ] Mix against last quarter: rows that changed share ______________________________
   [ ] Every reopen window in the month has closed

C. THE FIGURE WITH ITS PAIRS
   First contact resolution: ____ of ____ (about ____)     Last month: ____ of ____ (about ____)
   Mix-adjusted at last quarter's mix: about ____
   Reopen rate ____ (last ____)   Satisfaction ____ from ____ ratings (last ____)
   Escalation rate ____ (last ____)   Resolution time, median, most-escalated categories ____ / ____ / ____
   Grid reading, one line: ________________________________________________________

D. QUESTION ONE — WHAT MOVED, EXACTLY
   By category: row(s) outside expectation ______________________________________
   By channel: ______________   By week: ______________
   By technician (coaching only; not read below ____ comparable tickets): ______________

E. QUESTION TWO — WHY
   Twenty tickets from the slice, not resolved on first contact:
   ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____
   One word each, tallied:  knowledge ____  permission ____  tool ____  information ____  mix ____
   The cause in one sentence (or "not yet known; read forty more"): ______________________________

F. QUESTION THREE — WHAT ONE THING CHANGES
   The change (specific enough to be checked next month): ______________________________
   Attacks which cause: ______________   Owner: __________   Done by: __________
   Refused this month: [ ] an e-mail about the figure  [ ] a target  [ ] a new metric  [ ] a second action

G. THE SENTENCE FOR THE PAGE
   "First contact resolution ____ of ____ (about ____), from ____; reopens ____, satisfaction ____;
    the move is in ______________ because ______________; by the ____ we will ______________."

Jonah keeps the sheets in a folder, one a month. The folder is now the first thing he hands a new technician in their second week, because it explains the desk better than anything else he has. It gives every month's figure, what moved it, which tickets showed it, and what the desk did about it. Wen's runbook is on the third sheet. The e-mail about focus is on none of them, and the desk is better for the absence. First contact resolution is the requester's favourite number. This is how a desk keeps it honest: half an hour a month, one change at a time, with the tickets open and the names off the wall.

Frequently Asked Questions