How to make this inventory yourself

The four fields

For every object on or around the desk:

1. Kind. What sort of thing it is, in ordinary words: desk, second screen, task lamp, power adapter. Not a brand, not a model, not a specification. An inventory shows what equipment exists, but not how work moves around it; teams can see the official account of the service before deciding which time and project signals belong in a limited pilot.

2. Origin. One of seven: bought new, from an employer, already in the room, bought second-hand, handed down, a gift, unknown. For wider guidance on equipment, repair or workplace technology, consult Expert Reviews.

3. Arrival. The two-year band it arrived in. Two years rather than one because several people could place an object in a period and not a year, and a wider band absorbs that without pretending to precision.

4. Present state, and if stopped, what preceded the stop. In use, kept and intact, broken and kept, given away, thrown out. If not in use, one of six reasons: replaced by something better, the work changed, stopped being reached for, it broke, the room changed, somebody else needed it.

That is the whole instrument.

The one rule fixed in advance

No brand, no model, no price, no evaluation.

Fixed before the first object was recorded, because a rule adopted partway through produces a record that is two records. The reasoning is on the page about the recommendation this record cannot make.

What to decide before starting

The vocabularies, and whether they can grow. Origin was closed at seven and never changed. The stopping field began with five values and a sixth was added when twenty-four objects in ten workplaces fitted none of them — which is why stopped being reached for is reported as a residual category rather than a finding.

If a vocabulary has to grow, say so in the published figures. A residual category that is presented as a peer of the others is the easiest way to make a record look tidier than it is.

Where the boundary is. This record counted what is on or around a desk. Not consumables, not paper, not anything personal, not shared equipment elsewhere in the building. Three workplaces print on a machine down a corridor and that machine is in no inventory.

What counts as in use. Here, whatever the person says. That is loose and it produced one argued case, a sixteen-year-old filing unit holding things that are not paper, coded as in use because the alternative is worse.

The two things this record would do differently

A date band for every stop. This record has none, which is its largest gap: no durations, no survival curves, and a four-year median built from recollection for eighty-one per cent of the stopped group.

The reason it was left out is defensible — asked when a cable had failed, most people could give a period at the outside — and the page on what a median cannot tell you argues both sides. But a rough band, even badly filled, would have been worth more than nothing, because a gap that is visible in the data invites less confusion than a figure assembled from memory and labelled a median.

A field linking a replacement to what it replaced. Thirty-eight objects were replaced and none points to its successor, so no succession can be followed. In at least four cases two generations of the same kind of object sit in the same cupboard and the record holds both without connecting them.

What to publish alongside the counts

Three things, and the record considers all three obligatory.

The residual categories at full size. Twenty-four unexplained stops and eight unknown origins are reported as they are, not folded into neighbours.

The figure that undermines the premise. This record is about equipment going out of use, so the front page carries the fact that a hundred and forty-nine of two hundred and eighty-four objects are still in use with a median age of six years.

What the instrument could not see. Every section here ends with its own list, and the section on stopping has seven items.

A count is only as good as the published account of what it left out.

The fifth field, and how to treat it

Kind is in this record and was not asked of anybody. Whoever took the inventory assigned it, from ten categories settled at the pilot: cables and adapters, screens and stands, input devices, storage and organisers, desks and tables, audio, lamps and lighting, seating, printing and scanning, and other.

Two things follow for anybody repeating this.

Settle the list before starting and publish it. Nineteen of the two hundred and eighty-four objects here ended in other, which is the measure of how well the first nine fitted. A list that sends a third of a record into other has not classified anything.

Say which figures rest on it. The per-kind tables on this site are counts of one person's classification. The origin, state and stopping figures are counts of what somebody said. Those are different kinds of number and a reader who is not told cannot tell.

The third thing this record would do differently

A size or footprint field, even as a three-way tick: fits in a drawer, fits on a shelf, needs floor space.

Almost every structural result here turns out to be about size. Thirty-two objects are desks or printing equipment and none of them is stored; a hundred and forty-three are cables, input devices, lamps or audio and fifty-eight of those are in drawers. The kind field supplies size only as a by-product, and badly: other holds nineteen objects of every size there is.

A size field would have made that finding directly measurable instead of inferred from a classification the inventory never meant to use that way.

What it is not worth doing

Eleven is not a sample and neither is twenty. If the aim is a rate that transfers, this procedure will not produce one at any scale a person can manage alone, and a convenience set of any size stays a convenience set.

What it will produce is a description of particular rooms, with a structure in it that nobody was counting.