Recovery After a Bad Stretch on the Help Desk

Monday, 09:05, at Bluewater Bank. The mail outage ended on Thursday night after three days. The director sent a two-line e-mail on Friday that said "well done, team". This morning the calendar says nothing happened at all. Imogen, who runs the desk of five, can see otherwise. Luis was still replying to tickets at seven last night, according to the timestamps. Ana answered a chat message at 22:14 on Saturday. The queue holds a hundred and forty tickets that arrived during the outage and were never touched. Every one of the five technicians has come in this morning moving at the speed of Wednesday, which was the worst day. Nobody has told them it is over.

A bad stretch is a bounded period of heavier-than-sustainable load with a cause you can name. It might be the outage week, the backlog dig-out, the month of covering a colleague's leave, or the migration weekend. Recovery is the deliberate, scheduled period afterwards in which the load is lighter than normal. The people who carried the stretch can then come back to their normal level rather than carrying the stretch's weight into the next one. The first word in that definition is the one that matters. Recovery is deliberate. It does not happen by itself. A desk that waits for it to happen by itself gets the pattern described in What Burnout Actually Looks Like on a Help Desk. This article is part of our Helpdesk Burnout series, and it is about the fortnight after the bad fortnight.

It covers the three bad stretches a desk gets, and why recovery does not happen on its own. Then come declaring the end, time off that is actually taken, the lighter rota, and the conversation that names it. It covers when a quieter week is not enough, and stopping the stretch recurring. It ends with a short war story and the after-a-bad-stretch plan to copy, in that order.

The Three Bad Stretches a Desk Gets

The outage week is the loud one. Something many people depend on stops, the desk becomes the incident, and for a few days everyone works long hours under pressure, with the whole organisation watching. The running of the incident itself has its own series, Major Incident Management, and this article does not repeat it. What matters here is what the outage week leaves behind: a backlog of ordinary tickets nobody touched. It leaves five people who have been running on urgency for three days, and a sense across the organisation that the desk's job is now done. It is not. The desk's job has just become the hundred and forty tickets.

The backlog dig-out is the quiet one. A queue that has grown for months is finally attacked, usually because somebody senior noticed the number. The desk spends two or three weeks working through hundreds of old tickets. Each is a small, stale, slightly embarrassing conversation with a requester who gave up some time ago. There is no adrenaline in a dig-out. There is only the grind, and the grind is in some ways harder to recover from than the outage. Nobody sends an e-mail saying well done when a queue drops from four hundred to ninety. The method is in The Dig-Out Plan; the recovery afterwards is the same as for an outage, and more often skipped.

The month of leave cover is the one nobody counts. A colleague is away for four weeks, planned or not, and the rest of the desk absorbs their share. Nothing is on fire. Every day is simply a third heavier than it should be, for twenty working days. Because no single day is bad the stretch is never declared as one. The colleague comes back, and the load returns to normal. The people who carried it are expected to be normal too, immediately, having had no week in which they were. Planning the cover is in Vacation Coverage Without the Pile. Planning the recovery from the cover is almost never done, and this is the article that says to do it.

The table puts the three side by side with the fourth that most desks meet eventually, the migration weekend. That is when a system is replaced over two or three days outside normal hours. Each takes something slightly different from people and leaves something slightly different behind. The plan at the end of this article has a line for each.

Why Recovery Does Not Happen by Itself

The queue returns to normal and the people do not, and the gap between those two is where the damage happens. By the Monday after the outage, the ticket volume is back to what it was, so from the outside the desk looks recovered. The manager above the desk, who measures the desk by its volume, moves on. Inside the desk, five people are still carrying three days of long hours and a hundred and forty tickets they feel responsible for. They are carrying a habit of answering chat at ten at night that took three days to form and will not break on its own. The ticket history knows this before the manager does. If you run QuerySolv, every reply carries its timestamp in the event history. That is how Imogen knew about Saturday night without asking anyone.

The second reason is that praise is cheaper than time, and praise is what the stretch usually gets. "Well done, team" costs nothing and is sincerely meant. It also closes the subject, because once the thanks has been given the organisation considers the matter dealt with. A technician who then asks for a day off is asking for something extra rather than something owed. A thank-you is right. A thank-you with a day attached is a different thing, and the day is the part that recovers anyone.

The third reason is that the stretch does not announce its end. An outage has a moment when the service comes back. A dig-out has a morning when the queue is finally short. A month of cover has the day the colleague returns. None of those moments tells the people to stop. The urgency that got them through the stretch is still running, because urgency does not know the stretch is over. It burns for another week or two on nothing, which is the week the mistakes start. Urgency is a useful thing to borrow. It is an expensive thing to forget to return.

Declaring the End

The stretch ends when someone says it has ended, in writing, to the whole desk, and until then it has not. This sounds like a formality and is the most useful single act in this article. "The outage is over as of Thursday night. The hundred and forty tickets from it are a backlog, which we will handle as a dig-out with its own plan, not as this week's load. From today the roster is back to normal, and from Monday it is lighter than normal for everyone who worked the outage, as follows." Three sentences, sent on the Friday, and five people know that it is permitted to stop.

The declaration does two concrete things. It separates the stretch's leftover work from the ordinary queue, so that nobody treats a hundred and forty stale tickets as a normal Monday to be cleared by Tuesday. The leftover is a dig-out, with a taking-stock step first, and it is planned across weeks rather than attacked across a weekend. And it marks the moment from which the recovery period runs, so that "a lighter fortnight" has a start date and therefore an end date. That stops it quietly becoming "whenever things calm down", which is a date that has never arrived on any desk I have worked on.

Who declares it is whoever declared the start, where there was one, and the desk's manager where there was not. For the month of leave cover, where nobody declared a start, the manager declares both at once, on the day the colleague returns. The message is "the last four weeks were a stretch, you carried it, and here is the fortnight that follows". Saying it late is far better than not saying it. Saying it is the whole point.

Time Off That Is Actually Taken

Each person who carried the stretch gets a day off within a fortnight of its end, booked on a named date, covered on the roster, and taken. Not offered. Not "take a day when it's quiet", because it is never quiet and the offer is a way of giving nothing while appearing to give something. The manager books the dates with each person in the declaration week and puts them on the roster so the cover is visible. A technician who turns up on their day off is treated as a problem to be sent home rather than a dedication to be admired.

The day off is not the same as the leave the person was already owed. The two should not be confused, because confusing them is how carried-over leave becomes a number on a spreadsheet that nobody ever takes. If somebody worked a weekend, what they are owed for it in time or in pay varies by place and by contract. The right answer is to check both before promising anything. The specific case of after-hours work is in Compensation and Recovery Time. The recovery day in this article is simpler than that. It is a day the desk gives because the desk needs the person back at full strength. The desk gives it because it is in the desk's own interest.

Some people will refuse the day, and the ones most likely to refuse are the ones who most need it. "I'm fine, honestly, I'd rather get on with the backlog" is said by the person who answered chat at 22:14 on Saturday. Do not accept the refusal as information about their state; accept it as information about their habits, which the stretch has just reinforced. Book the day anyway, make the cover visible, and say that the desk needs them rested more than it needs them on Tuesday. The person who takes the day reluctantly comes back on Wednesday noticeably better. The person who does not take it is still at Wednesday-of-the-outage speed in three weeks' time.

The Lighter Rota

For the two or three weeks after the end, the people who carried the stretch hold fewer queue-duty blocks than the roster normally gives them. During that period, the desk starts no new projects. The queue-duty rotation from Sustainable Pacing makes this mechanical. Move two of each tired person's blocks to the people who did not work the stretch, or to the manager. Write the moves on the roster as the recovery plan rather than as swaps. Where everybody worked the stretch, which is the usual case on a small desk, the lighter rota means something else. In that case, the manager holds more blocks, the project work pauses, and the clear stretches that remain are spent on nothing in particular.

The projects pause out loud. A project paused quietly is a project that the technician feels guilty about, and guilt is load. The manager tells the people waiting on the project, in writing, that it has moved by a fortnight because the desk is recovering from the outage. The same written message says that the fortnight is deliberate. Most people waiting on a project accept a fortnight with a reason far better than they accept a fortnight of silence. The ones who do not are a subject for the expectations series in this library, not for this article.

The service-level targets stay as they are. The desk accepts that a few may be missed in the recovery fortnight and says so in advance. A lighter rota with no cost is not a lighter rota. The honest line belongs in the same message as the declaration: "response times may move a little for two weeks while the desk recovers from the outage, and here is what we expect". That makes a missed target in the recovery fortnight a planned cost rather than a surprise. A manager above the desk who is told that in advance usually says "understood". One who is told after a missed target usually says something else.

The Conversation That Names It

Within the recovery fortnight the desk has two conversations, and they are different conversations. The first is the incident review, which is about the systems: what broke, why, how it was found, and what changes so it does not break the same way again. That review is a standard part of closing an incident and the shape of it is in Closing the Incident and the Review. It is necessary, it is about the technology, and it is not the conversation this section means.

The second conversation is about the people. It has one job: to say, out loud, in front of the desk, that the stretch was hard and that these named people carried it. It must say what the desk is doing about it. "That was three days none of us wants again. Luis, you held the phone through all of Wednesday. Ana, you were on chat until late on Saturday and I have asked you not to do that again. The desk needs you more on Monday than on Saturday night. Everyone's recovery day is on the roster." That is twenty seconds. The alternative is the Monday where nothing is said and everyone is expected to be normal. That teaches the desk that stretches are invisible and that carrying them is simply what the job is.

Then the one-to-ones, in the second week, one question each: how are you, after that? Not "how is the backlog", which is a question about the work; the other one. Listen to the answer for longer than feels comfortable, and take the answer as information about the desk rather than about the person. If the answer is "fine, actually, the day off helped", believe it and check again in a month. If the answer is something else, the next section is about that. The one-to-one that asks and listens is a manager's standing duty rather than a recovery measure. The questions for it are in The Manager's Responsibilities. The recovery fortnight is simply when it matters most.

When a Quieter Week Is Not Enough

Most people come back from a bad stretch with a day off, a lighter fortnight and a conversation that named it, and are themselves again by the end of it. Some do not. The test is the same one the first article in this series uses for the difference between a bad fortnight and a pattern. Are the signs still there after the cause has gone and the recovery has been given? Consider a technician still writing one-word notes, still at the desk at seven, still flat in the huddle four weeks after the outage. That is two weeks after the recovery period ended. That technician is not still recovering from the outage. The outage has become the thing that made something longer visible.

At that point a manager's job is to say so, gently, privately, and in general terms, and then to stop. "You carried a lot during the outage and I don't think the fortnight was enough. If you're finding it hard to come back from, it might be worth talking to your doctor, or to the support programme the organisation has. I'll make the time for either." That is the whole speech. A manager is not qualified to say what is wrong, should not try, and should not press for details the person does not offer. What the manager can do is make it easy: keep the rota light and make the time. Treat the matter as a private one between the person and whoever they choose to talk to.

Two things a manager does not do. The manager does not diagnose, out loud or in their own head, because a manager's theory about a technician's state is both unqualified and unhelpful. The technician will hear it as a judgement whatever words it is dressed in. And the manager does not share it, with the team or with their own manager, beyond "I've adjusted the rota for a few weeks". The exceptions are if the person asks them to or the organisation's own process requires it. The rules about absence and adjustments vary by place and by contract. Where they come into play, the right move is to check both and to involve HR when the person agrees or the process demands. That is a different conversation from the one in this section.

Stopping the Stretch Recurring

A bad stretch that recurs on a schedule is not a stretch. It is the desk's normal load wearing a disguise, and recovery alone will not keep up with it. Count them. A desk that has had an outage week, a dig-out and two months of leave cover in a single year has spent a third of the year in a stretch or recovering from one. The plan at the end of this article has a line for the count because the count is the argument. Three stretches in a year is a staffing or a stability finding, and the case for either is made with the desk's own figures in Justifying Resources With Ticket Data.

Each kind of stretch has its own prevention. The outage week's prevention is in the incident review's action list. The desk's job is to make sure the actions have owners and dates rather than good intentions. The dig-out's prevention is the weekly hygiene that stops the mountain re-forming, in Keeping the Backlog From Rebuilding. A desk that digs out twice has not learned the first time. The month of leave cover is prevented by planning cover as a stretch from the start. That plan puts the lighter fortnight afterwards on the roster before the colleague leaves. This avoids discovering four weeks later that it was heavy.

The migration weekend is the one that can be designed. A weekend of work followed by a Monday of consequences is the shape only if the desk chooses it. A migration can be done on a Friday with the Monday after as a recovery day for everyone who worked it. The teething tickets are then taken by the people who did not. That is the same migration with the recovery built in. The desk rarely chooses the weekend. It can usually choose what Monday looks like.

A Short War Story: The Migration Weekend at Northgate Retail

At Northgate Retail the desk of four replaced the till system over a weekend, working from Friday evening to Sunday night. On the Monday morning the operations manager, Carys, thanked them in the stand-up and asked for the usual queue to be back to normal by Wednesday. The stores were already raising tickets about the new tills. Nobody took a day off, because nobody was offered one. By the second week a technician called Femi, who had done most of the Sunday, was making the kind of careless mistakes he never made. One was a configuration change applied to the live store group instead of the test group. That took the tills in six stores down for two hours on a Saturday. Femi went off work for five weeks shortly afterwards, and a second technician resigned in the gap. The desk ran at half strength through the busiest quarter of the year, covered partly by a contractor and partly by Carys herself at the phone. In the review she priced the weekend honestly: the migration itself had cost a weekend. The Monday that pretended it had not happened had cost two technicians, a store outage and a quarter. The lesson she wrote was that the recovery day is the cheapest line in any project plan. It is the only one nobody ever puts in.

The After-a-Bad-Stretch Plan

The plan is a single page, filled in on the day the stretch is declared over, by the person declaring it, and kept with the roster until the review date. It names the stretch, separates its leftovers from the ordinary queue, and books a day off for each person who carried it. It lightens the rota for a fixed period, schedules the two conversations, and sets a date to check that recovery actually happened. The last section is the count, and it is the part to bring to the next conversation about staffing. Fill it in even when the stretch was "only" a month of leave cover. Especially then.

AFTER-A-BAD-STRETCH PLAN — ONE PAGE
The stretch: ______________________   Kind: outage / dig-out / leave cover / migration / other
Started: ________   Ended: ________   Declared over by: ______________ on ________
Declaration sent to the whole desk:  [ ]   (what ended, what is left, what is lighter)

WHO CARRIED IT
  Name ______________  Long days: __  Weekend/evening hours: __  Recovery day booked: ________
  Name ______________  Long days: __  Weekend/evening hours: __  Recovery day booked: ________
  Name ______________  Long days: __  Weekend/evening hours: __  Recovery day booked: ________
  Name ______________  Long days: __  Weekend/evening hours: __  Recovery day booked: ________
  Each recovery day is on the roster with named cover:  [ ]

WHAT THE STRETCH LEFT BEHIND
  Tickets untouched during it: ____   Handled as a dig-out with its own plan:  [ ]   Owner: ________
  Promises made during it that still stand: ______________________________
  Projects paused, and who was told, in writing: _________________________

THE LIGHTER PERIOD   From ________ to ________ (two to three weeks)
  Duty blocks moved from the people above to: ______________ / ______________ / manager
  New projects started in this period: none
  Service-level line sent to management in advance:  [ ]   Date: ________

THE TWO CONVERSATIONS
  Incident / stretch review (the systems): held on ________   Actions have owners and dates:  [ ]
  The conversation that names it (the people): held on ________   Who was thanked by name: ____________
  One-to-ones in week two, one question each (how are they, after that?): ____ / ____ / ____ / ____

THE CHECK   Date, four weeks after the end: ________
  Early-signs checklist run for each person who carried it:  [ ]
  Anyone still showing the signs after the lighter period: ______________
  Pointed, privately and in general terms, to a doctor or the support programme:  [ ]

THE COUNT
  Stretches on this desk in the last twelve months: ____   Kinds: ______________________
  Weeks spent in a stretch or recovering from one: ____
  Recurrence fix for this one: ______________________   Owner: ________   Due: ________

Imogen filled the plan in on the Monday morning and sent the declaration at half past nine. The hundred and forty tickets became a dig-out with a three-week plan. Luis took the Thursday off, and Ana took the Friday. The chat message at 22:14 was mentioned once, kindly, with a request. At the four-week check everybody was back to normal except the queue, which was at ninety-one and falling, and which had never been the point.

Frequently Asked Questions