Keeping Asset Data Honest Through Moves, Swaps and Leavers

Tuesday, 11:30, at Stonebridge Insurance. Imogen is working ticket 3318, a claims handler whose laptop will not charge. The inventory sheet says the handler, Marta, has LT-0212. The laptop on Marta's desk is LT-0188. LT-0212, according to the sheet, is also assigned to Daniel in underwriting, who left eight months ago. LT-0188 is in the sheet as "spare". A second search finds LT-0212 in a drawer in the underwriting office, flat, with Daniel's lanyard still wrapped round it. Nobody did anything wrong. Daniel left, and his manager put the laptop in a drawer. Marta's old machine died and somebody gave her the spare. Three of those four events were handled without a ticket. The sheet was accurate the day it was written. That day was fourteen months ago.

Asset drift is the gap that opens between what the inventory sheet says about a device and reality. That means where the device actually is, who actually has it, and what state it is in. It is not caused by carelessness. It is caused by time. Devices move for reasons that have nothing to do with the desk. The sheet only learns about the moves that produced a ticket. This article is part of our Asset Data Inside Tickets series. It is about the two habits that keep the sheet honest. One is updating the row on every ticket that touches a device. The other is checking the sheet against the building once a quarter. The sheet itself, and the tag it is keyed on, are built in Linking Devices Without an Asset System.

It covers why asset data rots from the day it is written. Next come the three drifts, moves, swaps and leavers, and how each gets past the desk. We cover the update-on-ticket habit and what "touches a device" means, then the events that must become tickets even when nobody is asking for help. Next come the quarterly reconcile against reality, what to do with each kind of difference it finds, and how to measure whether the sheet is honest. A short war story follows, then the reconcile checklist to copy.

Asset Data Rots From the Day You Write It

The inventory sheet records the world as it was on the day each row was last touched, and the world does not wait. A laptop is assigned to a person, and the row says so. The person changes desks, teams, roles and eventually employers, and the row says so until someone changes it. Nothing in the sheet decays by itself; it is the building that moves. This is why a sheet built carefully in a single heroic week is often the least accurate one on the desk a year later. The week produced rows that were all true at once. The year produced no habit for keeping them true.

The sheet learns only through tickets, which is both its strength and the whole problem. Every device event that produced a ticket left a trace the desk could use to update the row. That could be a fault, a swap, a new starter's setup. And a desk with the habit uses that trace to update the row. Every device event that did not produce a ticket left no trace at all. That could be a manager lending a spare, a leaver's laptop into a drawer, a desk move organised by facilities. And the sheet does not know it happened. The desk cannot record what it never saw. So honesty has two parts: use every ticket you do see, and go looking, on a schedule, for what you did not.

Accept a certain rate of drift as normal and plan for it rather than being surprised by it. Take an organisation where a tenth of the devices change hands or place in a year, which is a quiet one. That moves twenty devices out of two hundred. If half of those moves happen without a ticket the sheet is ten rows wrong by December with everybody behaving sensibly. Ten wrong rows are ten wrong walks, ten wrong warranty lookups and one very confusing afternoon with a lanyard. The drift is not a failure of the sheet. It is the reason the sheet needs a quarter-day a quarter.

The Three Drifts: Moves, Swaps and Leavers

Nearly every wrong row is wrong for one of three reasons, and naming them is most of the way to catching them. A move is a device that changed place or person without changing state. For example, the laptop went with its owner to another floor. Or the desktop stayed at the desk through a reshuffle and has a new occupant. Or the printer was carried to the other end of the office. A swap is a device replaced by another. Examples are a faulty laptop exchanged for a spare, a loaner issued for a week and kept, a monitor taken from an empty desk. A leaver is a person who left, taking a row with them. The device is in a drawer, a cupboard, a manager's car, or occasionally a different country. And the sheet still shows a name that no longer appears on the payroll.

The table sets out how each one gets past the desk, which is the useful part. Each has a ticket that should have caught it and a reason the ticket was never raised. Read the last column as a list of the conversations the desk needs to have with other departments once, so that the ticket exists next time.

Leavers are the drift that costs most, because the device is usually fine and usually idle. That means the organisation is buying a new laptop for a starter while a working one sits in a drawer under a lanyard. Leavers are also the drift with the clearest fix, because every leaver already generates paperwork somewhere. The desk needs only to be on the list of people who receive it. The other two drifts need the habit and the reconcile. Leavers mostly need HR, the human resources team, to send one more e-mail.

The Update-on-Ticket Habit

The update-on-ticket habit is one rule: a ticket that touched a device does not close until the device's row says what the ticket did. "Touched" means changed the device's person, place, state or identity. That could be a swap, a loan, a repair that took it away and brought it back. It could be a reimage that changed the hostname, a move the technician carried out, a disposal. A ticket that only fixed something on the device, a driver, a setting, a stuck print job, touched nothing the sheet records. It needs no update beyond the tag already being on it. The distinction is quick to make. The point of making the distinction is that technicians update rows on the tickets where rows changed, not on every ticket. Updating on every ticket would be ignored within a week.

Put the rule on the close checklist, as one line, so that it is a step rather than a memory. The desk's close already has a shape, described in What a Proper Close Looks Like: a resolution note, a confirmation, the right status. "Device row updated, or no device change" is one more line, answered yes or not-applicable. It costs the technician about thirty seconds on the tickets where it applies. Those thirty seconds are the cheapest the sheet will ever be kept honest for. The technician is holding the facts, the device and the ticket all at once, and will never hold them together again.

Update the fields that changed and the last-verified date, and write the ticket number in the notes column. That way the row can say why it believes what it believes. "LT-0188: assigned to Marta, in use, verified on ticket 3318" is a row with a witness. If you run QuerySolv, the ticket's event history shows who did what on the ticket and when. So a row that cites a ticket number is a row whose claim can be checked a year later by opening the ticket. In any tool, the test is that every row changed this quarter can name the ticket that changed it. The ones that cannot are the ones the reconcile should look at first.

The Events That Must Become Tickets

Most drift happens in events nobody thought of as support. So the fix is to make those events produce a ticket whether or not anyone needs help. A new starter needs a device, and the ticket that sets up their account should name the tag they were issued. A leaver returns a device. The ticket that disables their account should have a line "device returned: tag, condition, where it went". That line is filled by the technician who received it, not by HR. A loan is a ticket with a return date. A loan without a ticket is a swap the sheet does not know about, and a loan without a return date is a gift. A move that the desk learns about from facilities or a manager is a ticket raised by the desk itself, two lines long. That way the row change has a witness. And a disposal, the device binned, recycled or sold, is a ticket that ends with the row set to retired. The data-wiping step is also recorded on that ticket, which the organisation will one day be asked to prove.

These are requests rather than faults. A desk that already separates the two, as set out in Requests, Approvals and Fulfilment, has a place for them. Each needs a short form or a template with the tag field on it and a line for where the device went. None of them needs an approval beyond what the organisation already requires for the starter or the leaver. The desk is not adding bureaucracy; it is adding one field to paperwork that exists. The request to HR and facilities is also short: "copy the desk mailbox on every starter, leaver and desk move, please". And the mail becomes a ticket in the usual way.

The leaver ticket deserves one more sentence, because it is where the money is. The desk should receive the device, not the manager. The ticket should not close until the desk has it in hand and the row says "spare" with its location. A leaver whose laptop is "with their manager" has a laptop in a drawer. And a drawer is where asset data goes to become fiction. Managers are not hoarding; they are busy, and the laptop is the least of the things the leaver left behind. The desk collects it because the desk is the only party that will ever look for it again.

The Quarterly Reconcile Against Reality

The reconcile is a scheduled comparison of the sheet against the world, done once a quarter, in a morning, by one named person. Its purpose is to find the drift that had no ticket. There are three sources of truth to compare, and the trick is to use all three rather than walk the whole building. The first is the building itself: a walk through one floor or one department with the sheet open, reading tags off devices and checking the row. The second is the login record, wherever the desk has one: which person last signed in to which hostname, and when. This is the cheapest way to find a laptop that has moved with its owner or a desktop whose occupant changed. The third is HR's list of starters and leavers for the quarter, read against the assigned-to column, which finds every leaver still holding a row.

Walk a sample, not the fleet. A quarter of the building each quarter means the whole organisation is physically verified once a year. A last-verified date on each row shows which quarter each device was last seen. Choose the sample by location rather than randomly, because walking a floor is efficient and sampling forty devices spread across four sites is not. Then use the login record for everything you did not walk. A hostname that has not been signed in to for ninety days is a device that is idle, lost, or in a drawer. A hostname whose last user does not match the row is a move or a swap. Twenty minutes with that list finds more drift than a day of walking.

Read the HR list last and slowly. Every leaver in the quarter should either have no device against their name or a leaver ticket that says where it went. Every starter should have a tag against their name that the sheet shows as in use. The gaps are the drifts that happened without the desk, and each becomes a two-line ticket now. That way the row change has a witness even if the event did not. The reconcile is boring, and I say that as someone who has done perhaps forty of them. It is also the only reason the sheet is still worth opening in its second year.

What to Do With the Differences

Every difference the reconcile finds falls into one of five kinds, and each has a fixed response. That way the morning is a procedure and not a series of small investigations. Moved: the device exists and works, and only its person or place is wrong. Update the row, cite the reconcile in the notes, and ask what moved it so the ticket exists next time. Swapped: the device in front of you is not the one the row expected, and the expected one is somewhere else. Update both rows, find the second device, and look for the fault ticket that should have recorded the swap. Retired without a ticket: the device is gone and nobody can say where. Mark it retired with "disposed, unrecorded" in the notes, and raise the wiping question with whoever disposed of it. Idle: the device exists, is assigned to a leaver or a drawer, and works. Collect it, make it a spare, and reissue it before anyone buys a new one. Missing: the row says in use, the login record says silent, and nobody has seen it. This is the only kind that needs a process beyond the row.

A missing device gets a ticket of its own, a fixed set of questions asked in order, and a deadline for declaring it lost. Ask the assigned person, their manager, and the location; check the login record for the last sign-in and where it came from. Check the spares cupboard and the repair shelf, because a surprising share of missing devices are in the room. Then, after a stated period, mark the row "lost" with the date. Tell whoever owns the fixed-asset register and whoever owns data security, and move on. A missing laptop is a data question before it is a money question. And the desk's job is to make sure the right people know in the same week, not to spend a month searching alone.

Count the differences by kind and keep the counts, because the pattern across quarters says where the desk is leaking. A quarter with six leaver rows wrong is a quarter in which HR did not copy the desk. A quarter with six swaps unrecorded is a quarter in which the close checklist was skipped. The reconcile finds the drift; the counts say which habit to fix. And the habit is cheaper to fix than the next reconcile is to run.

Measuring Honesty

Three numbers, reported quarterly, say whether the sheet can be trusted, and none of them needs more than a filter. The first is the verified share: the proportion of rows with a last-verified date inside the past twelve months. It should climb towards all of them as the quarterly walks go round. The second is the drift rate. It is the proportion of rows checked this quarter that were wrong in any way. It says how fast the building is moving relative to the desk's habits, and should fall as the update-on-ticket habit takes hold. The third is the matched rate from earlier in this series. It is the share of hardware tickets this quarter carrying a tag the sheet recognises. It says whether the sheet and the queue still agree about what exists.

Resist adding a fourth. The temptation is a dashboard with device counts by type, by age, by location and by warranty state. The sheet can produce all of them, and none say whether it is honest. Those are reports about devices. The three above are reports about the sheet. And a sheet that is a tenth wrong produces device reports that are a tenth wrong and look exactly like the right ones. (A wrong number has never once looked wrong.) Report the honesty numbers first, and then the device numbers with the honesty numbers beside them. That way whoever reads the warranty report knows how much to believe it.

A Short War Story: The Drawer at Stonebridge Insurance

At Stonebridge Insurance, the desk lead, Rafael, ran the first quarterly reconcile fourteen months after the inventory sheet was built. He found that thirty-one of the two hundred and six rows were wrong. Nine of them were assigned to people who had left. Eight of the nine leavers' laptops were in drawers and cupboards in their old departments, flat and working. Together they were worth about the same as the four new laptops the desk had bought for starters that quarter. A ninth, assigned to a regional broker who had left in the spring, could not be found by anyone. It had to be reported to the data protection officer as unaccounted for, which took a week of e-mails and produced a policy. Imogen collected the eight and imaged them as spares. The desk bought no laptops the following quarter. HR agreed to copy the desk mailbox on every leaver. The leaver ticket gained the line "device received by the desk" that must be ticked before it closes. Rafael counted what the first year's drift had cost. That was four laptops bought unnecessarily and perhaps a day of technician time spent walking to the wrong desks. It also included one week explaining a lost device that had probably been in a car. A row that names a person who has left is not a record; it is a rumour with a column.

The Reconcile Checklist

The checklist below is the quarterly morning as a procedure. It covers preparation the day before, the three sources compared in order, and the five kinds of difference with their fixed responses. It finishes with the three honesty numbers recorded at the end. One named person runs it; the whole desk sees the result. Keep each quarter's completed copy, because the fourth one is the first that can show a trend. A trend is what turns "the sheet is roughly right" into a number someone can defend.

QUARTERLY ASSET RECONCILE — QUARTER: ________   RUN BY: ______________

BEFORE (the day before, 20 minutes)
  [ ] Sheet filtered to this quarter's sample: floor / department ______________
  [ ] Login record exported: hostname, last user, last sign-in date
  [ ] HR list for the quarter obtained: starters ____  leavers ____  moves ____
  [ ] Spares cupboard and repair shelf listed by tag

THE WALK (the sample only)
  For each device seen:   tag matches row [ ]   person/location matches [ ]
                          status matches [ ]    last verified set to today [ ]
  Devices seen with no row:  ______   (add rows; note "found on reconcile")
  Rows with no device seen:  ______   (carry to "missing" below)

THE LOGIN RECORD (the whole fleet)
  [ ] Hostnames silent 90+ days listed: ______   (idle, lost, or in a drawer)
  [ ] Hostnames whose last user differs from the row: ______   (moves/swaps)

THE HR LIST
  [ ] Every leaver: no device in assigned-to, or leaver ticket cites return
  [ ] Every starter: tag in assigned-to, status in use
  [ ] Every move: location updated

DIFFERENCES FOUND, BY KIND
  Moved ______   Swapped ______   Retired unrecorded ______
  Idle (collect and make spare) ______   Missing ______
  For each: row updated [ ]   reason noted with "reconcile Q__" [ ]
            two-line ticket raised where the event had none [ ]
  Missing devices: ticket raised [ ]   owners informed by ________ [ ]

HONESTY NUMBERS (record each quarter)
  Verified share (rows verified within 12 months): ____%
  Drift rate (rows checked that were wrong): ____%
  Matched rate (hardware tickets this quarter with a known tag): ____%
  Which habit leaked this quarter: ______________________________

Stonebridge's second reconcile found eleven wrong rows against the first quarter's thirty-one. The third found six, four of them moves that facilities had organised over a weekend. Rafael's response to the fourth was to ask facilities to copy the desk on weekend moves. They do, in an e-mail that becomes a ticket that Imogen closes in about a minute by updating a row. The sheet is not perfect and will not be. It is honest to within a handful of rows. It can say which rows it is unsure of. And it has not sent anyone to the wrong desk since the spring. For a spreadsheet, that is a distinguished career.

Frequently Asked Questions