How to Triage a Call That Came Through Three People
Why this matters
A residential customer describes their own problem to you directly. A portfolio work order has usually crossed three people before it reaches your dispatcher: the tenant told the on-site porter, the porter told the assistant manager, the assistant manager typed six words into a system. What arrives is a label with the observation stripped out of it.
That is not carelessness. Every person in the chain is compressing to save the next person's time, and compression drops the mechanism and keeps the category. The cost lands entirely on you, as the wrong parts on the truck, the wrong trade dispatched, occasionally the wrong unit, and a severity call made from a word rather than from a fact. This procedure is about rebuilding enough of the original observation to dispatch correctly, in one phone call, before the truck rolls.
What a handoff actually removes
Three things go, in this order, and knowing the order tells you what to ask for.
The mechanism goes first. "It is blowing but not cold and there is water on the closet floor" becomes "no cooling." The second half was the diagnostic content and it did not survive the first retelling, because the porter did not know it mattered.
The location precision goes second. "The closet in the hallway" becomes "the unit." Whoever types the order works from the unit number, because that is the field on the form.
The timeline goes third. "It started yesterday and got worse this morning" becomes nothing at all. Nobody in the chain has a field for it, and it is the single most useful thing for severity.
What survives every handoff is the category word, the unit number, and the tenant's emotional register. Two of those three are unreliable and the third is not evidence.
Step 1: Reconstruct the chain before you read the content
Ask one question of whoever handed you the order: who actually saw this, and who told you.
You are establishing handoff count, not assigning blame, and the answer changes what you do next. A direct report from the person who saw it can be dispatched on. A three-handoff report is a rumor about a repair, and treating it as fact is how you send a technician with the wrong kit.
Record the handoff count on the order. It costs a field and it lets you measure, later, whether your accuracy problem is in your triage or in theirs.
If anything in the chain mentions a gas smell, a sounding CO alarm, a burning smell, arcing, or water at energized equipment, stop the reconstruction and act on it immediately. Instruct whoever is at the property: for a gas odor or a CO alarm, everyone leaves the unit and the building at once, no switch touched, no light operated, no phone used inside, call the gas utility and the fire service from outside at a distance. For a burning smell or arcing, leave the area and do not touch the device; kill the circuit only if the panel is dry and reachable without passing the affected area. For water at energized equipment, keep everyone out of the wet area and de-energize from a dry location. A hazard that reached you through three people is older than a hazard reported directly, which makes it more urgent, not less.
Step 2: Separate observation from interpretation in the text you were given
Read the order text and mark each phrase as one or the other.
"Water on the floor" is an observation. "Leaking water heater" is an interpretation, and it may be someone's third-hand guess. "Breaker keeps tripping" is an observation. "Bad breaker" is an interpretation, and it is wrong more often than it is right.
Dispatch on observations. Carry interpretations as hypotheses and never let one determine the parts you load, because an interpretation that turns out wrong costs you the whole visit.
The tell for an order with nothing usable in it: only interpretation, no observation. "AC not working." "Plumbing issue." "Electrical problem in 4B." Those are not reports, they are routing labels, and they are the ones that need a callback.
Step 3: Reach the observer, or price the uncertainty
The gate, so a dispatcher can apply it without judgment:
Call the observer before dispatch when the order passed through 2 or more handoffs AND either the reported condition involves water, a smell, heat or loss of heat, or a second unit, OR the order text contains no observation at all and only a routing label. One call, three questions, capped at 5 minutes. If the observer cannot be reached on one attempt, dispatch with a diagnostic-only scope and say so on the order and to the manager.
The three questions, in this order:
- What did you see or hear or smell, in your own words? Not what is wrong. What was there.
- When did it start, and is it worse now than when it started? This is the severity question and it is the one nobody asks.
- Is anything wet, and is anyone else affected? These two travel together because they are the accrual questions, and a tenant who says "only a little" to the first has still told you it is wet.
Five minutes is the cap for a reason. Past five minutes you are diagnosing over the phone with an untrained observer, and the accuracy stops improving while the tenant starts forming expectations about scope that you did not authorize.
"Diagnostic-only scope" is the honest fallback, not a failure. Tell the manager on the order: we could not reach the tenant, we are coming to diagnose, parts may require a second visit. A manager told this in advance treats a return trip as expected. A manager told afterward treats it as a failure.
Step 4: Verify the target three ways before you commit hours
A wrong unit is the most expensive triage failure because you can work on the wrong thing for an hour before anyone notices. On a phased or renovated property the label, the map, and the reality disagree routinely.
Verify by address and unit number, by the physical thing described, and by an occupant confirmation on arrival. If two of the three agree and one does not, resolve it before you open anything. There is a sibling card devoted to exactly this failure; use it rather than re-deriving the method here.
Step 5: Classify severity from the observation, not from the label
The word "emergency" travels through the chain unchanged and unverified, and so does the word "minor." Both are supplied by whoever was most or least alarmed, not by the mechanism.
Classify on what accrues while you wait: water in motion, a safety condition, a total loss of an essential service in an occupied dwelling, or anything affecting a unit other than the reporting one. The severity card in this category covers the full rule; the point here is that you cannot apply it to a label, only to an observation, which is why Step 3 comes before it.
Step 6: Write back up the chain in the words you will use on the invoice
Close the loop with one message to the manager, in the same language the invoice will carry. Found, done, next, and what remains open.
Two things this buys. The manager can forward it upward without translating, which is the actual service you are selling. And the words on the invoice will match the words in the record, so when the order is queried three months later by someone who was not there, the two documents agree. Invoices get disputed when the report says "cooling issue" and the invoice says something else entirely, even when both are true.
Worked example: forty orders, then forty more
Illustrative counts from one portfolio.
The order that started it. Text read: "AC not working, unit 4B." Handoff count, once reconstructed: three. The tenant told the on-site porter, the porter told the assistant manager, the assistant manager entered the order.
The callback took four minutes and produced this: the unit was blowing air but not cold, it started the previous evening, it was worse in the morning, and there was water on the floor of the hallway closet that the tenant had mopped twice. None of that reached the order text.
The water changed the dispatch: different kit, a different first check, and a severity that was no longer next-week routine because the water was recurring rather than a one-off spill. It also meant the technician arrived expecting a condition rather than a category, which is the difference between a diagnosis and a search.
The measurement. The shop compared order text against what the technician actually found on 40 consecutive work orders from that portfolio. 14 of the 40, which is 35 percent, carried a material difference: 5 changed the parts needed, 4 changed the trade or system involved, 3 turned out to be a different unit or a different location within the unit, and 2 changed the severity classification. Those four categories account for all 14.
The change. The three-question callback was applied to every order with 2 or more handoffs, which was 23 of the 40 in the next sample. Each callback averaged about 0.1 hours, so 2.3 hours of office time across the sample.
The next 40. 6 of 40 carried a material difference, which is 15 percent, down 20 percentage points. The mismatches that cost a visit, meaning parts, trade, or location, fell from 12 to 4, so 8 were avoided. At roughly 0.9 hours of travel and wait per wasted attempt, that is about 7.2 hours recovered.
Compare like with like: 2.3 hours of unbilled office time spent against about 7.2 hours of unbilled field time avoided, so roughly 3.1 hours of shop clock recovered for each hour spent. That is a shop-clock comparison, not a revenue return. What the shop keeps is the margin on whatever it sold into the freed field hours, which is a separate question.
Running the rule against its own founding case. The 4B order had 3 handoffs, clearing the first condition. But on the text as received, "AC not working, unit 4B" mentioned no water, no smell, no heat, and no second unit, so the original wording of the second condition would have said no callback, and the water would have been discovered on site. The rule only catches this case because of the clause covering an order that contains no observation at all and only a routing label. That clause was added after this order, not before it, and it is the reason the rule reads the way it does above. A gate written only around named conditions will always miss the orders that were compressed hardest, which are exactly the ones that need the call.
What flips this
- A property with an on-site maintenance person who inspects before reporting. Handoff count is high but the observer is competent, and the order text is usually better than a tenant's direct report. Reduce the callback rule to orders where their text is a label only.
- Orders that arrive through a portal with structured fields. Some portals force a description field, which preserves the observation. Check whether yours does before assuming the compression happened.
- An emergency. Do not run the ladder. Act on the hazard, dispatch, and reconstruct afterward.
- A single owner with a handful of units. The chain is one person and they usually saw it themselves. Skip to Step 2.
How to verify you got this right
- Add a handoff-count field to your intake and populate it for two weeks. If you cannot tell how many people a report crossed, you cannot apply any rule about it.
- Take 20 completed orders and compare the order text against the technician's findings. Count material differences and sort them into parts, trade, location, and severity. The distribution tells you which question in the ladder you are skipping.
- Read your last 10 order texts and mark observation against interpretation. If most contain no observation at all, the callback gate should be firing on most of your work and probably is not.
- Compare the words on three invoices against the words in the matching completion reports. Divergence there is where disputes start, months later, with someone who was never on the job.
References
- See related: The Wrong Unit Reported on a Multi-Unit Property
- See related: The Emergency That Is Only an Emergency to the Tenant
- See related: The Emergency Call Triage SOP
- See related: How to Schedule Around Occupied Units
- Trade-standard practice for work-order intake and dispatch documentation