How to Build an Equipment Inventory a Facility Will Use

Why this matters

Facilities departments buy equipment inventories, and most of what they buy is never used. The pattern is consistent: a vendor walks the building, produces a technically excellent spreadsheet in the vendor's own format with the vendor's own identifiers, hands it over, and it sits in a shared drive because nobody has the days it would take to re-key it into the system that actually schedules work. The building's information did not improve. An inventory is used only when it lands in the customer's numbering, in the customer's fields, at a level of detail their next decision needs. Everything below serves that.

The other reason to get this right is that an inventory survey is the cheapest way into an institutional account. It is small, it is low risk for the buyer, it puts you in every mechanical room in the building, and the shop that produced the asset list is the shop that knows the building when the service agreement comes up.

Build it in three passes, not one

The instinct is to capture everything in one sweep. Do not. Three passes at increasing resolution, each of which is useful on its own, is faster in total and it protects the project from stalling halfway with nothing deliverable. This is the spine of the whole procedure, and the reason for it is that each pass answers a different customer question and the customer's willingness to fund the next pass depends on seeing the last one.

Pass one, existence. Every asset, tagged, located, typed, counted. Answers "what do we have and where." Fast, roughly 3 to 6 minutes per asset in accessible spaces. Pass two, condition and data plate. Nameplate capture, age, condition rating, access notes. Answers "what state is it in." Roughly 8 to 15 minutes per asset, more where a panel or a guard has to come off. Pass three, criticality and dependency. What each asset serves, what fails when it fails, whether a spare or a bypass exists. Answers "what do we fix first when two things break at once." Slowest and most valuable, and it requires the facility's people, not just yours.

A department that only ever funds pass one still ends up with a usable list. A department handed all three at once frequently cannot absorb any of them.

Pass one: existence, in their numbering

Before you count anything, get the customer's numbering scheme and their CMMS export, even if it is incomplete. Two things you are extracting: how they format an asset tag, and which fields their system requires on import. If their tag reads 22-RTU-07, yours reads 22-RTU-07, not RTU-7 and not Unit 7. If their system requires a location code from a controlled list, you use their list.

Skip this and every row you produce needs translating, which is a day of somebody's time they do not have, and the reliable outcome is that the file is never imported.

Then walk. Capture five fields and nothing else: tag, asset type, location code, a photo, and whether the tag was found on the equipment or assigned by you. That last flag matters more than it looks. It separates verified assets from ones you inferred, and it tells the department exactly which rows to check.

Where an asset has no tag, do not invent a number in their scheme. Record it with a temporary marker and a photograph, and let the facility assign the permanent number, because a number you invent may collide with one they have reserved.

Pass two: condition, and the discipline of a defined scale

Condition is where inventories go soft. "Good" and "fair" from two technicians are not the same measurement, and a list with 200 subjective ratings cannot be used to prioritize anything. Define the scale before anyone walks, in observable terms, and put the definitions on the field form rather than in a memo.

A workable five-point scale, defined by what you can see rather than by adjective:

The point of definitions in observable terms is that two technicians converge. Test that before the survey runs: send two people to rate the same ten assets independently and compare. If they disagree on more than two of ten, the scale is not defined tightly enough yet, and you fix the scale rather than telling people to be more careful.

Age is the other soft field. Take it from the data plate where a manufacture date or a date code exists, and mark it as estimated where you took it from a building record or from appearance. An estimated age used in a replacement model without the estimate flag is how a department ends up defending a capital request built on somebody's guess.

Safety at this pass, because it is the pass that reaches for data plates. Where a plate sits inside an electrical enclosure, the enclosure stays closed unless the work is being done de-energized and locked out per 29 CFR 1910.333(b)(2) with the live-dead-live proving sequence in NFPA 70E-2021, 120.5, by a qualified person. Reading a label is not a reason to open a live panel. Where a plate sits on a machine with stored energy, a compressor, an accumulator, a spring-loaded damper, isolate and release the stored energy under 29 CFR 1910.147 before reaching past a guard. Where the plate is above a ceiling in an occupied care area, do not lift the tile without the facility's containment class assigned for that location, because dust released above an occupied ceiling lands on people who are not part of your project. Where a plate is behind pipe insulation of unknown composition, do not disturb it; report the location and let the facility's own program handle assessment, since disturbing suspect thermal system insulation is an airborne exposure that a glove does not address and that needs a respiratory program under 29 CFR 1910.134 and the facility's asbestos survey under 29 CFR 1910.1001 before anyone cuts into it.

Pass three: criticality, which you cannot do alone

Criticality is a facility judgment, not a technical one. A small pump that serves a research freezer outranks a large air handler that serves a corridor, and no data plate says so. Sit with the facilities engineer and the operations person for the space and rate each asset on two axes: consequence of failure, and time to restore.

Consequence is about who is affected and how fast: nobody notices for a week, occupants uncomfortable within a day, an operation stops, a regulated environment is out of compliance, a life safety function is impaired. Time to restore is about parts availability and access: on the shelf, days, weeks, long lead.

The assets that need attention are the ones high on both. That is usually a small fraction of the list, and identifying it is the single output a facilities director will actually take to their board.

Worked example: a 240-asset municipal survey

A shop is engaged to inventory a municipal complex. The department expects "a few hundred" assets.

Pass one finds 240 assets across four buildings. At an average 4.5 minutes each, that is 18.0 hours of capture, plus roughly 6.0 hours of travel between buildings and mechanical room access waits, so about 24.0 hours. Of the 240, 61 had no legible tag, which is about 25 percent, and every one is flagged as assigned-by-us rather than found.

That 25 percent figure is the first deliverable finding, before any condition data exists. A department that believed its list was current now has a number attached to how wrong it is, and the number came out of pass one alone.

Pass two covers 240 assets at an average 11 minutes each, or 44.0 hours, plus about 8.0 hours where guards, access panels or ladders were involved, so about 52.0 hours. Condition comes back with 18 assets at rating 5 and 34 at rating 4, so 52 of 240 assets, about 22 percent, sit at rating 4 or worse.

Pass three covers only the 52 assets at rating 4 or 5, with the facilities engineer present, at roughly 20 minutes each including discussion, so about 17.3 hours, of which the engineer's own time is roughly half. Cross-rating consequence against time to restore puts 9 of those 52 in the high-consequence, long-lead quadrant.

Total survey effort is about 93.3 hours across the three passes. Note where it went: pass three, the most valuable output, is about 19 percent of the hours. Pass two is about 56 percent, and it is the pass a department is most tempted to cut.

Now the finding that pays for the survey. Those 9 assets are 3.75 percent of the 240, and they are the ones where a failure means an operation stops and the part is a long lead. Against a department that had been replacing equipment in the order things broke, a prioritized list of 9 is an actionable capital plan. The other 231 assets are not urgent; saying so with evidence is as useful as the list of 9, because it lets the department stop worrying about them.

The failure mode to avoid here is delivering the 52 rather than the 9. A list of 52 degraded assets reads as a demand for a large budget and gets deferred whole. A list of 9, with the 52 behind it as supporting detail, is a decision somebody can make this quarter.

One caution on the numbers above. The per-asset minute figures are this shop's own averages across accessible mechanical spaces in a low-rise complex. A high-rise, a hospital with escort and containment requirements, or a plant with a permit system will run longer per asset, sometimes substantially, and the ratio between the passes moves too because access overhead loads pass two hardest. Re-time your own first building rather than carrying these figures into a different kind of site.

How to verify the inventory will actually get used

One test, run before you invoice: ask the facility to import a sample of 20 rows into their live system and tell you what failed. Not to review the spreadsheet, to import it. A review passes almost anything; an import fails on the real defects, a location code outside their controlled list, a tag format their field rejects, an asset type that does not exist in their taxonomy.

If all 20 import clean, deliver the rest. If they do not, you have found the defect while the file is still small enough to fix, and you have done it before the department formed an opinion about whether the deliverable works.

Second test, at ninety days: ask whether any work order has been raised against an asset your survey created. If none has, the inventory is not in use, and the reason is worth knowing while you can still fix it.

References

• 29 CFR 1910.333(b)(2), electrical lockout and verification; NFPA 70E-2021, 120.5, live-dead-live proving; 29 CFR 1910.147 for mechanical and stored energy isolation
• 29 CFR 1910.1001, asbestos, including duties around suspect thermal system insulation; 29 CFR 1910.134, respiratory protection program requirements
• See related: Asset Tags and Equipment Lists and Why They Matter to You
• See related: What a Maintenance Management System Does With Your Notes