The Fundamental Attribution Error With Customers and Crew
Why this matters
"He's lazy." "She's cheap." "They just don't care." Every shop says these things, and some of them are true. Most are not, and the cost of the wrong ones is specific: a good tech written off and eventually lost, a customer marked difficult who would have bought if anyone had asked the right question, a fix aimed at a person when the problem was the route, the stock list or the quote. The habit that prevents it is cheap: before you explain behavior by character, check whether the situation explains it.
What the error is
The fundamental attribution error is the tendency to explain other people's behavior by their character and to underweight the situation they were in. Lee Ross named it in 1977, building on an experiment by Edward Jones and Victor Harris in which people read essays that writers had been assigned to write, knew they had been assigned, and still judged the writers to hold the views in them.
Its mirror image is the actor-observer asymmetry: when you are late, it was traffic; when a tech is late, he is unreliable. You see your own situation in full and other people's barely at all, so the same behavior gets two different explanations depending on whose it is.
In a shop the error is amplified by distance. The owner sees a tech's numbers on a screen and a customer's decline as a single line on a ticket. The situation that produced both happened somewhere the owner was not.
Worked example: the "slow" tech
A six-tech heating and cooling shop. The owner pulls a month of completed tickets and sees one tech, call him Marcus, averaging 2.1 jobs a day against the rest of the crew's 3.4. The owner's read: Marcus is coasting.
Before the conversation, the service manager breaks the day into its parts for a typical 9-hour shift.
Drive time. Marcus covers the rural north zone. His GPS logs average 3.2 hours of driving a day against the crew's 1.5. That leaves him 9 - 3.2 = 5.8 hours off the road, against 9 - 1.5 = 7.5 for everyone else.
Parts runs. Marcus's van carries the older stock list, from before the shop standardized on the newer equipment line. He makes counter trips most days, logged at 0.9 hours a day, against the crew's 0.3. That brings his working time to 5.8 - 0.9 = 4.9 hours, against 7.5 - 0.3 = 7.2.
Time per job. Marcus spends 4.9 / 2.1 = about 2.33 hours per job; the crew spends 7.2 / 3.4 = about 2.12. He takes about 10 percent longer per job on site.
Ticket mix. Dispatch had quietly been sending Marcus the no-heat diagnostics, because he is the best diagnostician. Sixty percent of his tickets were diagnostics against about a quarter of the crew's. A diagnostic runs longer than a maintenance visit by nature.
So the 2.1 against 3.4 that read as laziness decomposes into a route with about twice the driving, a stock list that costs him three times the counter time, and a ticket mix the shop chose for him. The remaining 10 percent difference per job on site is the kind of gap a mix of 60 percent diagnostics against 25 percent would produce on its own; to know for sure, compare his diagnostic time against the crew's diagnostic time, job type to job type. On this evidence Marcus is not the slow tech. He is doing the shop's hardest work in its worst territory, and the shop's own dispatch and stock decisions put him there.
The fix was not a performance conversation. It was restocking his van to the current list, which removes most of the 0.6-hour daily gap in counter time, and tracking diagnostic tickets separately so the comparison stops punishing the person who takes them.
Note what would have happened otherwise. A talk about "picking up the pace" would have told the shop's best diagnostician that his reward for taking the hard calls was being called lazy. Techs leave over exactly that.
Worked example: the "cheap" customer
The office flags a customer who has declined a system replacement three times in two years and paid for a repair each time. The ticket note reads "price shopper, won't buy." Before the account is marked low-priority, the office lead calls.
OFFICE: Hi, it's Dana from the shop. I'm following
up on the repair last week. Everything still
running all right?
CUSTOMER: It's running. For now.
OFFICE: The tech noted you'd looked at replacement a
couple of times. I wanted to ask what got in
the way, so we stop pitching you something
that doesn't fit.
CUSTOMER: Honestly? It's not my house. I rent. The
owner pays for repairs, but he won't sign
off on a new system, and I can't.
OFFICE: That helps a lot. Would it be useful if we
sent the owner the age, the repair history
and the quote directly?
CUSTOMER: Please. I've been trying to tell him.
The behavior that looked like cheapness was a tenant with no authority to buy. Three replacement quotes had been presented to the one person in the transaction who could not say yes. The shop sent the history to the property owner with the repair count and the system age, and the next decline came with a reason the shop could work with.
Notice how the question was asked: "what got in the way" assumes there was a reason and invites it. "Why didn't you go ahead?" puts the customer on the defensive and gets a shrug.
Other situations that routinely sit behind a "cheap" label: a quote presented as one option with no lower tier, a household on a fixed income where a repair fits the month and a replacement does not, a customer burned by a previous contractor who now haggles everyone, a decision-maker who was not home when the tech presented. Each has a different fix and none of them is "give up on the customer."
The three questions that separate person from situation
Harold Kelley's covariation idea gives a practical test. Before you settle on a character explanation, ask three things:
- Consensus: would most people in the same situation do the same? If any tech given the north zone and the old stock list would run 2.1 jobs a day, the situation is doing the work.
- Consistency: does this person do it across situations? If Marcus ran slow on the south route too, with a current van and an ordinary ticket mix, that would point at him.
- Distinctiveness: does it happen only here? A tech who is late only on Mondays after the weekend on-call shift is telling you about the on-call schedule.
Character is the right explanation when consensus is low, consistency is high and distinctiveness is low: other people in the same spot behave differently, this person behaves this way everywhere, and it is not tied to one condition. That combination is less common than it feels.
Where to look first in a shop
The situational causes that most often sit under a character label:
| Label | Situation that commonly produces it |
|---|---|
| Slow tech | Route geography, stock list, ticket mix, older truck, a helper who is new |
| Careless tech | No checklist for that job type, rushed by an overbooked board, unclear scope from the office |
| Doesn't care | Pay plan that rewards something else, no feedback on callbacks, never told the result |
| Cheap customer | No authority to buy, single-option quote, fixed income, prior bad contractor |
| Difficult customer | A promise someone made that the tech did not know about, a missed window, a confusing bill |
| Unreliable office | Interrupted every four minutes, no rule for who decides what |
Pay plans deserve their own warning: behavior that looks like a character flaw in a tech is very often a rational response to how the tech is paid. See related: Incentives Explain Behavior Better Than Character.
Failure modes on both sides
Overcorrecting into excuses. The point is to check the situation first, not to rule out character forever. If you fix the route and the stock and the mix, and the numbers do not move, the consistency question now has an answer. Have the performance conversation, with the situational fixes on record as evidence you looked.
Fixing the situation silently. Restocking Marcus's van without telling him why leaves him wondering what changed and leaves the "slow" label sitting in the owner's head. Say it: "Your numbers looked low until we broke them down. The route and the stock were costing you most of it. We fixed the stock." It costs one sentence and resets the relationship.
Comparing raw numbers across unlike situations. Jobs per day across techs with different routes and mixes compares the situations, not the techs. Track a number that holds the situation steady, such as time on site per job type, before reading anything into individual performance.
Labels that travel. "Price shopper" in a ticket note outlives the call. The next person to read it treats the customer that way without ever checking. Write what happened ("declined replacement, tenant, owner authorizes") instead of what you concluded about them.
The habit
When you hear yourself, or anyone in the shop, explain a problem with an adjective about a person, ask one question before acting: what about their situation would make a reasonable person do this? If you can answer it, check the answer against the record. If you cannot, ask them. Either way you will be right more often, and when it really is the person, you will know it on evidence rather than on a feeling about a number.
References
- Ross, L., "The Intuitive Psychologist and His Shortcomings," Advances in Experimental Social Psychology (1977)
- Jones, E. E. and Harris, V. A., "The Attribution of Attitudes," Journal of Experimental Social Psychology (1967)
- Kelley, H. H., work on the covariation model of attribution (1967, 1973)
- See related: Incentives Explain Behavior Better Than Character
- See related: The Tech Kept Recommending Replacement and It Was the Pay Plan