Measuring and Reducing Printer Tickets Per Device

Friday, 15:50, at Larkhill Academy, a secondary school with a desk of two. Grace is writing the term's report for the business manager and has reached the line about printers. She knows what the staffroom thinks: the staffroom printer is the worst device in the building, it is always broken, and it should be replaced. That view has been expressed to her in the staffroom, by e-mail, and once in the car park. She also has the ticket export open, and the export is not sure. The staffroom device has the most tickets. It also prints more pages in a week than the art block prints in a term. The art block's device, which nobody complains about because nobody expects anything of it, has a ticket for every few hundred pages it manages. Grace has to write one sentence about which printer to replace, and the folklore and the export disagree.

The disagreement is the subject of this article. Measuring printer tickets means counting them per device, over a quarter, against the pages each device printed. That way, the busiest device is not mistaken for the worst one. The quiet device that fails every time it is asked to work is not mistaken for a good one. The count then drives four decisions a desk can actually take: keep a device, watch it, move it, or retire it. This article is part of our Printer Support series. It turns the one-off census from the first article into a standing quarterly scorecard.

It covers the one number, tickets per device per month, and getting the device onto every ticket. Next come the second cut, tickets per thousand pages; the lemon device; and the location hotspot. Then come the quarterly review in one hour and declaring a device dead. After that comes what reduced the count on a fictional desk and what did not. Last come a short war story and the printer scorecard to copy.

The One Number: Tickets Per Device Per Month

The number that drives printer decisions is tickets per device per month. It is every printer ticket in the period, attributed to the device it was about, divided by the months in the period. It is read as a list with the worst at the top. Not tickets by category, which says what kind of fault the fleet has and nothing about where; not tickets per technician, which says nothing useful about printers at all; and not the total, which is the number the first article counted and which cannot be acted on. A device is a thing the desk can replace, move, or send an engineer to. A category is not. Count the thing you can act on.

The period is a quarter, and the reason is arithmetic. A small desk's devices generate a handful of tickets a month each, and a handful is noise. A device with two tickets one month and five the next has not got worse. It has had a damp week. Three months smooths that enough to read; two quarters side by side is enough to say whether a change worked. Judge nothing on a month. The rules for counting consistently, deciding which date puts a ticket in the period and what counts as one ticket, are in Counting Tickets Properly and apply here unchanged.

Two counting rules are specific to printers. First, a "nobody can print" ticket that turns out to be a stuck job on the server's queue is attributed to the device whose queue it was, because that is where the hours went, even though the device was innocent. The resolution note's phrase, "stuck head job", is what separates it later. Second, every walk-up counts: the toner changed in passing, the jam cleared on the way to a meeting, the "can you just look at it" in the corridor. Printers are the device people most often fix in person. A per-device count that includes only the formal tickets will flatter the devices nearest the desk and condemn the ones furthest away. Log the corridor, as a one-line ticket afterwards with the device named.

Getting the Device Onto Every Ticket

None of this works unless every printer ticket says which printer, and on most desks, at the start, few of them do. The fix is in three places. On the device, the label with the queue name, which the on-printer card carries in its first section, so the requester has a name to type. On the form, there is a field asking for that name. The first article added it to Dana's form at Meridian Parts, and it closes a quarter of tickets in the first reply. And at close, the technician fills in the device if the requester did not, as part of the resolution note. This is in the same spirit as categorising at close. The person who fixed it knows which device it was, and the count only needs one of the two people to say.

The field is the thing. Most small-team help desks have no asset module, and QuerySolv is one of them. The device label lives in a field on the department's form and in the resolution note. The quarterly count is a filter on that field. That is enough. A desk does not need a register of every device with its serial number and purchase date to count tickets per printer. It needs a consistent label on every ticket. The ways to attach a device to a ticket without buying anything are in Linking Devices Without an Asset System. Where the labels are inconsistent, "staffroom", "Staff Room printer" and "SR-MONO" for the same device, the first quarter's job is to make them one, by hand if necessary. Then put the one on the card.

The Second Cut: Tickets Per Thousand Pages

The staffroom device at Larkhill had the most tickets because it did the most work. A count that stops at tickets per device will replace the busiest printer in the building every time. That is an expensive way of discovering that the new one is busy too. The second cut is tickets per thousand pages: the device's tickets in the period divided by the pages it printed, in thousands. Pages come from the counter the device prints on its own test page, which the triage flow asked technicians to record on every visit. They can also come from the managed-print supplier's monthly report or from the server's own accounting where the tool keeps it. Record the count at the start and end of each quarter for every device, which is one walk round the building twice a quarter, or one report.

Here is Grace's quarter, with every number invented for the example. Eleven devices, thirty-eight printer tickets. The staffroom device: fourteen tickets, which was the top of the first list, on about sixty thousand pages, which is roughly one ticket for every four thousand pages. The art block device: nine tickets on about six thousand pages, one ticket for every seven hundred. The other nine devices: between zero and four tickets each, on eight to fifteen thousand pages, somewhere around one ticket for every five thousand pages. On the first cut the staffroom was the problem. On the second cut the staffroom was among the most reliable devices in the school. The art block device, which nobody had ever complained about, was failing six times as often for the work it did. Both cuts are true. Only the second one is a decision.

Read the two together and the fleet sorts itself into four kinds of device. High tickets and high pages is a busy device doing honest work, and the question is whether it needs a twin beside it. High tickets and low pages is a lemon or a bad location. Low tickets and high pages is the device you wish they all were. Low tickets and low pages is a device nobody uses, which the fleet standard may not need at all. The scorecard at the end of this article has a column for each cut, and a verdict column that depends on both.

The Lemon Device

A lemon is a single device whose tickets per thousand pages sit far above every other device of its kind for two quarters running, after paper and storage have been ruled out. The last clause matters, because the commonest "lemon" is a sound device fed from a damp cupboard. The physical article's test, a fresh dry ream and a torch, is cheaper than an engineer and much cheaper than a replacement. A device that is still at the top of the list after the paper has been fixed, the rollers replaced and an engineer has visited is a lemon. The right response to a lemon is not a fourth repair. It is a date.

Tell a lemon unit from a lemon model, because the fix is different. If one device of a model fails and its three siblings do not, the unit is the problem and it is replaced with another of the same model. If every device of a model sits near the top of the list, the model is the problem and the fleet standard should not be choosing it. That is a different conversation with a different person. The method for cutting the count by model as well as by unit, and for telling a model's reputation from its record, is in Device and Model Hotspots. This article borrows it rather than repeating it. Larkhill's art block device was a unit: its model's two siblings in the humanities block were among the best in the school.

The Location Hotspot

Sometimes the device is fine and the place is not. A device by an external door breathes damp air and collects wet umbrellas. A device in a corridor gets knocked, unplugged by the cleaner, and used by everyone who passes, which is everyone. A device beside the team that prints labels all day meets a label on its hot roller once a month. A device in a cupboard overheats. The sign of a location hotspot is that the tickets stay when the device is replaced. The second-floor printer at Tollgate Foods in the physical article had been replaced twice. Both replacements jammed on the last day of the month. That was because the fault was in a cupboard where the statement paper lived and in a tray set for the wrong stock. A hotspot is a place that generates tickets whatever is put in it. The idea in general, with how to find one in the ticket data across devices, applications and groups, is in What a Hotspot Is.

The test is a swap. Move the suspect device somewhere quiet and dry and put a known-good device in its place, for a quarter, and watch where the tickets go. If they follow the device, it is a lemon. If they stay with the place, it is a hotspot. In that case, the fix is to the place: the paper store, the door, the power, the tray settings, or the card. In the end, the fix may be the decision to put the printer somewhere else. A swap costs an hour of the desk's time and answers a question that three engineer visits did not. Printers, unlike people, do not mind being moved to see how they behave.

The Quarterly Review in One Hour

The scorecard is filled in once a quarter and read in one meeting of one hour. The meeting includes the desk lead and whoever changes the toner. If the fleet is managed, the supplier's monthly reports are to hand. The first twenty minutes are the two cuts: tickets per device, tickets per thousand pages, worst at the top, and the four-way sort. The next twenty are the three questions asked of each device in the top third of the list. Has the paper and storage been ruled out? Has a service visit happened and what did it find? And is this the device or the place? The last twenty are the verdicts, one per device, with an owner and a date for each that is not "keep". The verdict table below is what the hour produces.

Keep the review to the hour by refusing to diagnose in it. A device that needs a torch gets a technician's name and a date, not twenty minutes of theories around a table. The meeting's job is to produce the verdict column, and the verdict column's job is to be read next quarter against what happened. The quarterly rhythm for hotspot reviews in general, which this one can sit inside rather than beside, is in The Quarterly Hotspot Review. A desk with two people does both in the same hour.

Declaring a Device Dead

A printer gets repaired forever if nobody adds up the repairs, because each visit, on its own, is cheaper than a replacement and reasonable to approve. The rule that ends it is written down in advance, in hours, so that the decision is arithmetic rather than courage. Here is Larkhill's rule, with the figures as the fictional desk chose them and not as a standard. Replacing a general-purpose device costs the desk somewhere between three and five hours: unboxing, the queue and the driver on the server, mapping, the card, the test, and removing the old one. A device is declared dead when, for two quarters running, the desk's hours on its tickets and its engineer visits exceed that. This applies after a fresh ream, a roller check and one service visit have failed to bring it down. It is also declared dead when the supplier has visited three times for the same fault. It is also declared dead when a repair would cost more than the device is worth, which the supplier will say. Any one of the three is enough.

Write the rule into the fleet standard and apply it without a meeting. The general version of the decision, for laptops and desktops as well as printers, is in Swap, Repair or Retire. Printers differ only in that the hours are easier to count, because every visit is a walk. The hours are also the sentence for the budget line. "This device cost the desk eleven hours last quarter and nine the quarter before; a replacement costs four hours and the device" is a request that gets approved. "The staffroom hates it" is not, however true both are.

Retire means remove. A retired printer that is moved to a quiet office to "see out its days" takes its ticket count with it and becomes a personal desk printer with a history. That is the one kind the fleet standard specifically forbids. It goes to the supplier, the recycling route or the skip, with its queue deleted from the server the same day and a note on the ticket that closed it. If you run QuerySolv, that ticket's complete event history holds every visit, every reopen and every note from the device's life. That way, the decision to retire it reads as a record rather than a verdict. In any tool, the device's tickets are its biography, and it is worth linking the last one to the first.

What Reduced the Count, and What Did Not

Three quarters after her first scorecard, Grace had the honest record below, every number invented and the pattern the useful part. The art block device was retired at the end of the first quarter and its replacement, the standard model, has had two tickets since. The paper store for the main office moved from a cupboard by the fire exit to a shelf in the office. That block's jam tickets fell from seven in the quarter to two. The staffroom device got a twin, a second standard mono device on the same bench. The staffroom device's own tickets fell by about half while its pages barely changed. That is what a device doing honest work does when the work is shared. Central queues went in across the school in the second quarter, and the "cannot print" tickets fell most of all. The school's printer tickets went from thirty-eight in the first quarter to seventeen in the fourth.

Two things did not move. Scanning tickets stayed at three or four a quarter, because the devices' accounts had no owner until the third quarter. Then the queue standard gave them one and the expiry went on a calendar. And the humanities block's two devices, which were already the best in the school, stayed the best, because there was nothing to fix. A count that fell everywhere would have been suspicious. A count that fell where the fixes were applied and stayed flat where they were not is the count telling you it is honest, and the fixes telling you which ones worked.

Small counts need the usual care. Seventeen tickets across eleven devices is a number where one bad week moves a device two places on the list. So read the direction over two or three quarters and the verdicts on the per-thousand-pages cut. Say in the report that the counts are small. Never cut printer tickets by technician. The device is the unit, and the technician who takes more printer tickets is usually the one whose desk is nearer the stairs. Report the two counts and the hours in one line to the business manager, in the shape the quarterly review article describes, with the verdicts attached. That way, the sentence about the staffroom printer is a sentence with a number in it and not a car-park conversation.

A Short War Story: The Seven Repairs at Carrow Veterinary Group

Carrow Veterinary Group was a chain of six veterinary practices with a desk of two at the main surgery. The reception printer at the busiest practice was repaired seven times in one year under a service contract. Each visit was approved by the practice manager because each was cheaper than a new device and each engineer found something real to replace. Each failure also meant a morning of reception printing invoices and vaccination records by hand on the practice's consulting-room device. Nobody added the mornings up, because the mornings were the practice's and the repairs were the desk's. The two columns never met. A technician called Ewan started a per-device scorecard in the spring. He found the reception device had thirty-one tickets and seven visits in the year. He found that the desk alone had spent about twenty-six hours on it. The practice had spent at least as many printing by hand. He wrote the three-visit rule into the standard and cited it. The device was replaced within the month for about four hours of his time. The lesson he put at the top of the scorecard was one line: a device gets repaired forever if nobody adds up the visits, and the person who adds them up is the person who ends it.

The Printer Scorecard

The scorecard below is Grace's, filled in once a quarter, one row per device, with the two cuts, the pattern, the service visits, the desk's hours and the verdict. Section A is the period and the fleet's middle values, which every device is read against. B is the device table, worst at the top. C is the verdicts carried forward with owners and dates. D is the honest line about what changed and what did not. Keep every quarter's sheet, because the third sheet is the one that shows whether the verdicts worked. Attach the latest to the fleet standard's annual review. Nothing on it is a joke, because the verdict column retires printers.

THE PRINTER SCORECARD — ONE QUARTER
Desk: ______________________   Period: ____________ to ____________   Filled in by: ______________   Reviewed on: ____________

A. THE FLEET THIS QUARTER
   Devices in service ______   Printer tickets, all devices ______   (walk-ups logged after the fact: ______)
   Fleet middle value: tickets per device per month ______     tickets per thousand pages ______
   Desk hours on printer tickets and visits, all devices ______ h

B. BY DEVICE (worst first on tickets per thousand pages; one row each)
   Label ____________ Location ____________ Model ______ Pages ______ Tickets ____ /month ____ /1000pp ____ Top category ________ Visits __ Hours __ Last verdict ______
   Label ____________ Location ____________ Model ______ Pages ______ Tickets ____ /month ____ /1000pp ____ Top category ________ Visits __ Hours __ Last verdict ______
   Label ____________ Location ____________ Model ______ Pages ______ Tickets ____ /month ____ /1000pp ____ Top category ________ Visits __ Hours __ Last verdict ______
   Label ____________ Location ____________ Model ______ Pages ______ Tickets ____ /month ____ /1000pp ____ Top category ________ Visits __ Hours __ Last verdict ______
   (continue for every device; devices with no tickets are listed too)

C. VERDICTS THIS QUARTER (every device in the top third, and any device with a visit)
   Device ____________  Verdict: keep / watch / move / twin / retire   Paper and storage ruled out: yes / no
      Reason: ______________________________________   Owner ____________   By ________
   Device ____________  Verdict: keep / watch / move / twin / retire   Paper and storage ruled out: yes / no
      Reason: ______________________________________   Owner ____________   By ________
   Device ____________  Verdict: keep / watch / move / twin / retire   Paper and storage ruled out: yes / no
      Reason: ______________________________________   Owner ____________   By ________

   The dead-device rule in force: desk hours above ______ h for two quarters after a ream, rollers and one visit;
   or ______ visits for one fault; or repair cost above the supplier's replacement threshold. Any one applies.

D. WHAT CHANGED SINCE LAST QUARTER
   Verdicts carried out: ______________________________________________________________
   Counts that fell, and the fix they followed: __________________________________________
   Counts that did not move, and why: __________________________________________________
   One line for the quarterly report: "____ printer tickets across ____ devices, about ____ desk hours; ____ retired, ____ moved."

Grace's sentence to the business manager that Friday said that the art block device would be replaced, that the staffroom device would get a twin, and that the staffroom device itself was among the most reliable in the school for the work it did. The staffroom took the second half of that sentence badly for about a week, until the twin arrived and the queue at the bench disappeared. The art block, which had never complained, got a printer that worked, and did not notice, which is the best review a printer can hope for.

Frequently Asked Questions