How to Standardize Across Units That Are Not Standard

Why this matters

A property gets sold to you as one thing: forty-eight units, one owner, one dispatch line. Then you get there and the buildings went up in three phases across eleven years, half the units were renovated by different contractors, and no two mechanical spaces are laid out the same. You cannot carry parts for forty-eight variants, so you carry parts for none of them, and every work order becomes a diagnostic visit followed by a return trip.

Return trips are how portfolio work stops being profitable. They double travel, they double tenant access coordination, and they age the open work order the manager is graded on. The fix is not to make the property standard. It is to sort it into a small number of classes that behave the same from your truck's point of view, and to stock and schedule against those classes instead of against units. This procedure produces one artifact, a class sheet, and everything else follows from it.

Step 1: Classify by what changes your work, not by what is different

The mistake that kills this before it starts is classifying on everything that varies. Unit layout, finish, orientation, age, and a dozen other attributes all differ, and almost none of them change what you do.

Use one rule to decide whether an attribute earns a split:

An attribute earns its own class only when it changes truck stock OR changes expected on-site time by 25 percent or more, measured on a median of at least 5 completed work orders per candidate class. Anything else is a note on the unit record, not a class.

The Boolean is OR, because either alone ruins a first visit. A variant needing a different part means a return trip regardless of how fast the work is. A variant taking half again as long means a blown appointment window regardless of what is on the truck.

The median-of-5 condition is what stops you inventing classes from one memorable job. Below five completed orders you are looking at a technician's bad afternoon, not a pattern.

Attributes that usually earn a split, across trades: the isolation and connection arrangement, the physical access route (crawl space, roof, locked common room, through a tenant's living space), the generation or capacity class of the installed equipment, and the control or wiring scheme. Attributes that usually do not: cosmetic finish, unit number, floor, tenant type, and the year on the building plaque.

Step 2: Survey once, in a batch, and get something for it

Surveying forty-eight units one at a time as work orders arrive takes two years and you will never have a complete picture, because by the time you reach the last one the first has changed.

Do it in a batch. Ask the manager for access to a sample across every phase, wing, or renovation vintage, in one or two scheduled days when you can move without tenant coordination. You do not need every unit. You need enough per candidate class to satisfy the median-of-5 condition, plus every unit that anyone describes as "the odd one."

Do not do this for free and do not try to bill it as a mystery line. Trade it. The manager gets an access map, an equipment age list, and a deferred-items list they can take into a budget conversation, and those three documents are genuinely useful to them. You get a class sheet that makes every subsequent visit cheaper. Both sides are better off and nobody has to pretend the survey was free.

If a vacancy cycle is running, ride it. Vacant units are the cheapest access you will ever get on that property.

Step 3: Write the class sheet

One row per class. The columns are fixed, because the sheet's job is to be readable by a technician who has never been to the property:

• Class label. Short, spoken aloud without embarrassment on a dispatch call.
• Which units. By address and unit number, not by description.
• What defines it. The one or two attributes that earned the split.
• Truck stock to carry. Specific, by generic part type and count.
• Expected on-site time. In hours, from your own completed orders.
• Access notes. Route, key or code, restricted hours, and any isolation device whose location or labeling is wrong.

Keep the class count small. Somewhere between three and six classes covers most mid-size portfolios. If you land above six, you have classified on an attribute that did not earn it, and the sheet has stopped being usable.

Merge rule: any class holding fewer than 3 units folds into its nearest neighbour and is carried as an annotated exception on those unit records. A class of one or two units cannot meet the median-of-5 evidence condition and cannot justify dedicated stock, so it is a note, not a class.

Step 4: Set stock by coverage, not by completeness

You will not stock for every class and you should not try. Set the target as coverage:

Stock the truck to cover the classes that together hold at least 80 percent of the units. Everything else is order-in, with a stated lead expectation you give the manager at the time of diagnosis.

Eighty percent is the working default, not a law. Raise it toward ninety when the portfolio is far from your supply house or when the manager's own metric is open-order age. Lower it toward seventy when the parts in question are bulky, expensive to hold, or slow-moving, because dead stock on a truck is a real cost too. Move it in one step at a time and re-measure a full quarter before moving again.

The reason coverage beats completeness: the last classes are always the smallest and the most expensive to carry, and they are exactly the ones where a scheduled return trip is acceptable to a manager who was told the lead time up front.

Step 5: Put the class at the unit, and verify isolation at the device

A class sheet in the office does nothing for the technician standing in a mechanical room. Two things go on the unit record so they are visible before the truck leaves: the class label and the access route.

The safety piece is not optional on a phased property, because labeling drifts across renovations and the maps drift with it.

• If a shutoff, valve, breaker, or disconnect is unlabeled or does not match the map, do not act on the map. Trace and verify at the device before any work, and correct the record before you leave.
• For electrical work, prove the circuit dead at the point of work with a tester proved on a known live source immediately before and immediately after the test. A mislabeled panel in a phase-two building is the ordinary case, not the exception, and a map inherited from a prior vendor is not a verification.
• Where your employee is servicing equipment and unexpected energization or release of stored energy could injure them, the isolating device is locked and tagged in the de-energized position before work starts, not merely switched off (29 CFR 1910.147 in general industry, 29 CFR 1926.417 where the work is construction and the hazard is an electrical circuit, since 1926.417 covers lockout and tagging of circuits rather than stored energy generally). On a shared-service property this matters more than usual, because a breaker you opened is a breaker another occupant has a reason to close.
• If you smell gas at any point on the property, everyone leaves the area and the building immediately. Do not operate any switch, do not use a light, do not use a phone inside. Call from outside, from a safe distance, and keep others clear until the utility or the responding authority releases the area.
• Where a common isolation device serves more than one occupied unit, notify every affected occupant before you close it and confirm what they lose. On a phased property the served area frequently does not match the unit numbering.

Correcting one mislabeled isolation point per visit, and writing the correction on the class sheet, is the highest-value habit in this whole procedure. It compounds, and it is the part that makes the sheet trustworthy to the next technician.

Step 6: Set the re-survey trigger

The sheet decays. Renovations happen, equipment gets replaced, a wing changes hands. Set the trigger rather than a calendar:

Re-survey the affected units when any of these occur: a renovation or capital project touching the systems you service, a change of management company, or a run of three consecutive work orders in one class where the stocked parts did not fit. The third one is the real signal, and it is why the class needs to be recorded on the completed order.

Worked example: the filled-in class sheet

Illustrative. Forty-eight units across three phases, one portfolio.

The survey ran over two days during a vacancy cycle and covered nineteen units plus every unit anyone had called odd. Five candidate classes came out of it.

Applying the merge rule to the sheet itself. Class E holds 2 units, which is fewer than 3, and with two units it cannot reach a median of 5 completed orders. Under the rule stated in Step 1 and Step 3, E does not stand as a class. Its two units fold into D, the nearest neighbour on access and stock treatment, carried as an annotated exception noting the shared common room and the key held only by the manager. That leaves four classes, with D holding 6 units.

Applying the coverage rule. Classes A and B together hold 35 of 48 units, which is 73 percent, under the 80 percent floor. Adding class C takes it to 42 of 48, which is 88 percent, clearing it. So the truck stocks for A, B, and C. The remaining 6 units, all now in class D, are order-in with a lead time quoted at diagnosis.

What it changed. In the twelve months before the class sheet, the shop completed 33 of 60 work orders on that portfolio on the first visit, which is 55 percent. In the twelve months after, with A, B, and C stocked and the class label on the dispatch, it completed 47 of 60 on the first visit, which is 78 percent. That is a gain of 23 percentage points, and the classes that were not stocked account for most of the remaining shortfall, which is the intended outcome rather than a failure.

The number that mattered to the manager was different and worth knowing: median days from dispatch to close, which moves almost entirely on first-visit completion. That is the figure to put in front of them, because it is the one on their report.

What flips this

• A portfolio mid-way through replacing equipment across all units. Classify anyway, but expect churn, and set the re-survey trigger on the project schedule rather than on failed-fit counts. A sheet rewritten three times in a year is still cheaper than no sheet.
• Very high-value or bulky parts. Coverage above about 70 percent stops paying, because the holding cost of the last classes exceeds the return-trip cost. Move the floor down and be explicit with the manager about lead times.
• Scattered single-family rentals rather than a campus. The defining attribute is usually equipment vintage rather than building phase, and travel dominates, so the stocking case for a marginal class is stronger than its unit count suggests.
• A portfolio you may not hold next year. Survey smaller: cover only the classes needed to reach the coverage floor, skip the full deferred-items list, and do not invest in stock you cannot redeploy.

References

• See related: How to Standardize Tools Across a Crew
• See related: The Wrong Unit Reported on a Multi-Unit Property
• See related: How to Schedule Around Occupied Units
• Trade-standard practice for site survey, equipment inventory, and isolation device labeling
• OSHA 29 CFR 1910.147, control of hazardous energy, for lock and tag on servicing and maintenance by your employees in general industry; 29 CFR 1926.417 for lockout and tagging of circuits where the work is construction