Keeping the Backlog From Rebuilding After the Dig-Out

The Monday after the dig-out, the count was 71. The chart had come off the wall. And the team at Northgate Retail's desk wanted to go for lunch. Ravi wanted to as well, and went. But before he did, he opened the created-versus-solved chart for the previous four weeks and looked at the gap rather than the lines. Created: 162, 158, 171, 166. Solved, not counting the dig-out's closes: 155, 151, 160, 157. The gap was still there, nine or ten tickets a week, exactly as it had been before the fortnight. The pile was gone. The thing that had built the pile was eating a sandwich in the kitchen, unchanged. It would have four hundred tickets back on the desk by the following autumn.

Every desk that has dug out once knows this moment, and most of them handle it by not looking at the chart. The dig-out is satisfying, visible and finite. The work that stops the pile rebuilding is none of those things: it is small, weekly, and never finished. And it is the part of this series that matters most. A desk that only knows how to dig out will be digging out every year for the rest of its life.

Rebuilding is what a backlog does when the dig-out removed the tickets but not what caused the pile. That could be the gap between inflow and outflow, or the missing closing rules, or the capacity shortfall. The pile returns at the same rate it formed, and nobody notices for the same reasons nobody noticed the first time. This article is part of our Backlog Recovery series and it is the one that starts the Monday after The Dig-Out Plan ends.

It covers why the Monday after is the dangerous day. Next come inflow control, which is removing demand rather than refusing it, and closing discipline, meaning rules with dates. We cover the fifteen-minute weekly aging check, then the capacity truth and the arithmetic behind it. Next come early-warning thresholds that trigger action, the one-afternoon mini dig-out, and what management hears monthly. We finish with the weekly health check to copy.

The Monday After

A dig-out changes the count and nothing else. Inflow is the same, because the organisation did not change. Capacity is the same, because the team did not grow. The habits are the same, because two weeks of blocks is not long enough to form new ones, and the blocks were about the old tickets anyway. The only thing that is different on the Monday after is that the pile is small, which means the gap is invisible again. That is exactly the condition under which the first pile formed.

The gap is easy to find and hard to feel. Pull four weeks of created and solved, ignoring the dig-out's closes, and subtract. If the result is positive, the backlog is growing by that number every week, starting now. At that rate it will be back at its old size in roughly the time it took to build the first time. Ten a week is five hundred a year. The arithmetic is not complicated. It is just that nobody does it on the Monday after, because the Monday after is for lunch.

Four things stop the rebuild, and a desk needs all four. The first is less inflow of tickets that should never have been tickets. The second is closing rules that stop tickets living in waiting statuses. The third is a weekly check that catches the pile at twenty instead of four hundred. The fourth is an honest number for capacity that the desk and its management both accept. The rest of this article is those four, then the thresholds that tell you when to act, then the small act itself.

Inflow Control Is Not Saying No

Inflow control means reducing the number of tickets that arrive, by removing the reasons they are raised rather than by refusing them. A desk that refuses tickets has not reduced demand. It has pushed demand into the corridor and the direct message, where it is still work and no longer counted. A desk that removes the reason a ticket is raised has reduced the work itself. The difference is the whole subject, and the techniques are the ordinary ones, applied with the backlog in mind.

Start with the repeats. Pull three months of closed tickets and sort by subject or category. The top five kinds will account for a surprising share of inflow, and most of them are the same question asked by different people. Each one has one of three cures. A knowledge article the requester can find before they ask, which cuts the question. A fix to the underlying cause, which cuts the problem; finding and working those is the subject of Finding the Repeat Tickets. Or there is a change to a form or a process somewhere else in the organisation, which cuts the request at its source. An example is a joiner form that asks for the right access the first time. Ten tickets a week removed at the source is the whole gap at most small desks, and it is cheaper than a person.

Then the duplicates and the bounces. A requester who raises the same problem twice because the first ticket went quiet is inflow that the desk created. A ticket that bounces between two departments three times before anyone works it is one ticket counted as three touches, and the bounces are time. Both are fixed at intake: merge the duplicate and reply once, and route to the right place first time. Routing to the right place is a question covered elsewhere in this library. Neither needs a tool change. Both need the triager to look before they assign.

And the walk-ups and messages that become tickets late, after the work is done, or never. These are not inflow the desk wants to reduce; they are inflow the desk wants to see. Untracked work is capacity spent with no record. When the capacity conversation happens the record is the only evidence. Log them, even in one line. A desk that is "too busy to log the walk-ups" is too busy because of the walk-ups, and cannot prove it.

Closing Discipline: Rules With Dates

Closing discipline is a small set of written rules that say when a ticket in a waiting status is closed, by whom, and with what message. That way no ticket can live in Awaiting Reply or On Hold indefinitely. The inventory in the previous articles found most of the pile in exactly those statuses. That was not because anyone chose to put it there, but because nothing said when it should leave. The rules below are the ones that work; the numbers are a starting point, and yours may differ. But every rule must have a number, because a rule without a number is a hope.

Awaiting Reply: the desk asks its question. It chases once after three working days, and chases a second time after six with a closing date in the message. The desk closes the ticket on the tenth working day with a reply that invites the requester to re-raise by replying. Three touches, two weeks, every time, written down. The chase wording that gets replies rather than silence is a craft of its own. And the aging and follow-up series owns it; the backlog rule is only that the cadence exists and ends.

Resolved: when the desk has fixed something and told the requester, the ticket closes automatically or by hand after a fixed number of working days without a reply. Five is common, and the resolving reply says so. In tools that have both statuses, Resolved is the desk's word and Closed is the requester's silence or confirmation. In tools that have one, the resolving reply carries the same sentence and the ticket closes on the date. Either way, a resolved ticket has a closing date the moment it is resolved.

On Hold: every ticket On Hold names what it is waiting for and who outside the desk owns that thing. It also names a review date no more than two weeks away. On the review date someone chases the outside party and sets a new date. Or they tell the requester honestly that the wait is indefinite and ask whether they want to keep the ticket open. On Hold without a review date is a cupboard. Check too whether your tool's SLA clock pauses in a waiting status; some do, some do not. A team that believes the clock is paused when it is not will park tickets and then watch them breach.

Projects, changes and anything with more than a handful of steps leave the queue. A ticket that is really a project never closes on a ticket's timescale and drags the aging numbers with it. So it is recorded wherever the desk tracks that kind of work, with a date, and the ticket closes with a reply pointing there. And one rule over all the others: every close has a reply. The status change tells the requester nothing in many tools and nothing useful in the rest. And a close without a reply is where the next bankruptcy's complaints come from.

The Weekly Aging Check

The weekly aging check is fifteen minutes, on the same day each week. In that time, one person reads five numbers and the oldest ticket, and acts on anything that has crossed a line. It is the single habit that would have prevented the first mountain. And it is the only part of this article that has to happen every week without exception. Do it on Monday morning, before the live queue has had a chance to look urgent. Have the report or the spreadsheet open and the health check sheet beside it.

The five numbers start with open tickets over thirty days and open tickets over ninety days. Next comes the weekly gap, created minus solved, for last week. Then come the count of unassigned tickets older than one day and the count of tickets in waiting statuses past their rule date. Then read the oldest open ticket, by number and subject, aloud if there is anyone to read it to. Most tools can produce the first two from a backlog-by-age view and the third from a created-versus-solved report. The last two take a filter on status and a look at the dates. Where a tool cannot do one of them, the export-and-count spreadsheet from the inventory does it in five minutes.

Then the test, which is one question asked of every ticket over thirty days: does it have an owner and a date? An owner is a named person on the desk. A date is the next day on which the desk will do something to the ticket, written in the ticket. A ticket over thirty days with both is work in progress, and the check leaves it alone. A ticket over thirty days with either missing gets one of them in the fifteen minutes, or goes on the list for the next mini dig-out. The per-ticket technique for the ones that have gone stale, the reading, the chase, the decision, is the subject of The Stale Ticket Sweep. The weekly check only needs to find them.

Keep the sheets. Twelve weeks of health check sheets is a trend. A trend is what you show management when the capacity conversation comes, which it will. A desk that can say "the over-thirty count has risen from 22 to 61 over nine weeks while inflow rose from 150 to 180 a week" is a desk with a case. A desk that can say "it feels busier" is a desk with a feeling. Feelings are accurate and nobody funds them.

The Capacity Truth

The solved line on a created-versus-solved chart is flat because capacity is flat, and no amount of weekly checking changes that. Capacity is the number of tickets the desk can close in a week with the people and hours it has. It can be estimated on the back of an envelope in five minutes, which is five minutes most desks have never spent. The honest number is lower than anyone expects, and it is the number that the whole backlog conversation eventually comes down to.

Here is the arithmetic for a desk of three at Acme. Three people at thirty-seven and a half hours is 112 hours a week. Take out leave, sickness and training, which average out to about a tenth: 100 hours. Take out meetings, project work, walk-ups that are not logged, and the general tax of being interruptible. On most small desks, that is somewhere between a third and a half of what remains. Call it 45 hours gone, leaving 55 hours actually spent on tickets. If the average ticket takes 25 minutes of desk time across its life, including reading, replying, fixing and closing, 55 hours is about 130 tickets a week. Acme receives 150. The gap is 20 a week, which is a thousand a year, which is the mountain, forecast on an envelope.

The number has only three levers. Reduce inflow, by the inflow control above: 20 a week is five repeat questions with a knowledge article each, or two root causes fixed. Increase the hours spent on tickets, by protecting them: moving four hours a week of project work off the desk is ten tickets. Or add a person, which is the lever everyone reaches for first and the one that takes longest. The case for it, and the honest timing, is in When to Hire the Second Person. There is a fourth option, which is to lengthen the promise. That way the desk commits to less and the backlog becomes an agreed queue rather than a broken one. Lengthening the promise is legitimate as long as it is agreed with the people waiting and not just declared.

Say the number to management in those terms, and say it before the pile returns. "We can close about 130 tickets a week with the time we have and we receive 150; here are the three ways to close the gap and what each costs" is a sentence a manager can decide on. "We're really stretched" is not. The one-page version is in Explaining the Backlog to Management. Most managers, shown the envelope, pick inflow control first, because it is free, and that is usually the right pick.

The Early Warning: Thresholds That Trigger Action

An early-warning threshold is a number which, when the weekly check crosses it, requires a specific action rather than a note. The point of writing thresholds down in advance is that on the week they are crossed, the desk is busy. That is why they were crossed. A busy desk will always decide that this week is an exception. A threshold written on a calm Monday removes the decision. The table below is a set that works for a desk receiving a hundred to two hundred tickets a week. Scale the counts to your inflow and keep the structure.

Amber is for watching, and watching means writing the number on the sheet and reading it again next Monday, nothing more. Red is for acting in the same week, and the action is written in the table so that nobody has to invent one on a bad Monday. Two ambers in different rows are not a red. Three weeks of the same amber is, because a number that is drifting the same way for three weeks is not noise. It is the created line pulling away from the solved line. That is the mechanism from the first article in this series, caught at twenty instead of four hundred.

Resist the urge to add rows. Six numbers is what one person can read in fifteen minutes and act on; twelve is a dashboard, and dashboards get glanced at. If a seventh number matters to your desk, swap it for one of the six rather than adding it, and write down why. The check is small on purpose. Its whole value is that it happens every week, and things that happen every week have to be small.

The Mini Dig-Out

When a red threshold is crossed, the response is a mini dig-out: one afternoon, one or two people, one protected block. It uses the same order of attack as the fortnight, applied to a pile of twenty or forty instead of four hundred. Decisions first, chases second, quick wins, then the real work re-triaged with owners and dates. It ends when the red number is back to amber or the block ends, whichever is first. It is recorded on the health check sheet as having happened. Most desks need one every six to ten weeks. A desk that needs one every fortnight has a capacity problem, and the sheets will show it.

The mini dig-out is the reason the full dig-out never happens again. Four hundred tickets took a year to form because nothing interrupted them; a block every couple of months, triggered by a number, interrupts them at thirty. It costs an afternoon a quarter, roughly, against a fortnight a year. Nobody has ever regretted that trade. Several people have regretted the alternative, usually in a meeting.

Protect the mini dig-out block exactly as the fortnight's blocks were protected, with a cover technician and a manager who says no. A three-hour block that is interrupted twice is an hour, and an hour does not move thirty tickets. If the block cannot be protected, say so on the sheet. A sheet that shows three red weeks and three cancelled blocks is itself the capacity case, made without adjectives.

What Management Hears Monthly

Management hears four numbers once a month and nothing else about the backlog, unless a threshold goes red. The four start with open over thirty days, now and a month ago. Then come the weekly gap averaged over the month, the oldest ticket's age, and the number of mini dig-outs run. Use one paragraph, the same shape every month, sent on the same day. That way the manager learns to read it in thirty seconds and notices when the numbers move. The month the gap turns positive is the month to add a sentence about capacity, and the sentence has the envelope arithmetic in it.

Resist the quarterly backlog presentation. A desk that reports monthly in four numbers never needs one. A desk that gives one has usually waited until the numbers were bad enough to be interesting. The best backlog report is boring, arrives on time, and says the same four things with slightly different numbers. The month it is not boring, everybody already knows why, because the report has been teaching them to read it.

A Short War Story: The Second Mountain at Harbourview Clinic

Harbourview Clinic's two-person desk dug out a backlog of 380 tickets in a fortnight, properly, with blocks and a chart and a careful bankruptcy. The desk finished at 60, which the clinic manager called excellent. Beatriz, who had led the dig-out, meant to start a weekly check the following Monday. But the following Monday was busy, and the one after was busier. The check never became a habit because nothing was yet wrong. Eleven months later the count was 410, and the oldest ticket was from the week after the dig-out. The clinic manager asked, not unkindly, whether the desk intended to do this every year. The second fortnight cost the same as the first, plus a manager who no longer believed the number would stay down. There was also a team who had learned that digging out was something that happened to them annually. The weekly check that would have prevented it would have taken fifteen minutes on roughly forty-five Mondays. The lesson Beatriz gives now is that the dig-out fixed the pile, and only the Monday check fixes the gap.

The check started the Monday after the second dig-out and has not been missed since. The desk has run four mini dig-outs in two years and no full ones. And the clinic manager gets four numbers on the first of each month, which she says she reads on her phone in the car park.

The Weekly Health Check

The sheet below is the fifteen minutes on one page. Fill it in on the same day each week, from the same report or spreadsheet, by the same person where possible. Keep the sheets in order. The threshold columns are the table above; write your desk's own numbers into them once and leave them. The last two lines are the only ones that require a decision, and the decision is already written in the thresholds.

WEEKLY BACKLOG HEALTH CHECK — ONE PAGE
Desk: __________________  Week beginning: __________  Checked by: ____________

THE SIX NUMBERS                     This week   Last week   Amber at   Red at
  Open over 30 days                 ______      ______      ______     ______
  Open over 90 days                 ______      ______      ______     ______
  Weekly gap (created - solved)     ______      ______      ______     ______
  Unassigned older than one day     ______      ______      ______     ______
  Waiting statuses past rule date   ______      ______      ______     ______
  Oldest open ticket (days)         ______      ______      ______     ______

THE OLDEST TICKET
  Number: ________  Subject: ______________________________
  Owner: ____________  Next action date: ________  Decided today? [ ] yes

OVER-30 TEST
  Tickets over 30 days: ____   With owner AND date: ____   Fixed today: ____
  Still missing one or both (for the mini dig-out): ______ ______ ______ ______

CLOSING RULES THIS WEEK
  Awaiting Reply closed by rule: ____   Resolved closed by rule: ____
  On Hold reviewed on date: ____ of ____

INFLOW
  Top repeat subject this week: ______________________  Count: ____
  Inflow-control action open: ______________________  Owner: ________

VERDICT (circle one)
  all green  /  amber: watch ____________  /  RED: ____________

ACTION THIS WEEK (red only)
  Mini dig-out block booked:  day ________  time ________  cover: ________
  Note to management sent?  [ ] yes  [ ] not needed

A desk of one fills in the same sheet and strikes the cover line; the mini dig-out block is a closed door and a forwarded phone. A desk with several departments fills in one sheet per department, because the gap in one is not the gap in another. It then fills in one line summarising all of them. What does not change is the day, the fifteen minutes, and the rule that a red number gets an action in the same week.

Ravi's desk did the first check the Monday after lunch. The gap was ten a week, which was red by the table's third row, and the action was to pull the repeats. Three subjects accounted for fourteen tickets a week between them; two got a knowledge article and one got a fix to the joiner form. The gap was two a week by the end of the month and negative by the end of the next. The count a year later was 48. There was no second dig-out, and there has not been one since. Nothing about the desk changed except fifteen minutes on a Monday. That turned out to be the whole difference between a desk that digs out and a desk that does not need to.

Frequently Asked Questions