Firing an Employee Without Handing Them a Claim

Why this matters

By the time an owner is reading anything about firing someone, the decision is usually made. The tech has been a problem for months, the crew knows it, and the only question left feels like scheduling. That is the wrong question, and the cost of asking it is that a termination which was substantively correct becomes a claim that is expensive to defend and unpleasant to lose.

The right question is what has to be true before you do it, and the answer is not the same in every case. Three situations present in a small shop and they need different things from you. Getting the branch wrong is how a shop spends four weeks it did not need to spend, or fires on the spot when it should have suspended, or walks into the one case where a page like this cannot help. None of this is legal advice about your own facts; the state layer decides a good deal of it.

The question that decides it is not "can I"

Under at-will you almost always can. That is the wrong axis and a sibling card owns why. See related: At-Will Employment and the Exceptions That Actually Bite.

The axis that matters is what a stranger would be able to see, months later, about why this happened. Work the branch by asking two things in order. First, what kind of case is this: a pattern that built up, a single incident, or a decision arriving on the heels of something the law protects. Second, does the thing that kind of case runs on already exist in writing. A pattern case runs on a documented history. An incident case runs on an investigation. A case with recent protected activity runs on a decision that can be shown to predate the activity. Those are three different documents and they are not interchangeable.

Branch A: the performance case that has been building

This is the common one. Someone has been slow, sloppy, rude to customers or unreliable for months.

The criterion is the file, not the behaviour. Open the personnel file and ask whether it contains dated entries naming specific behaviour against a standard the person was told about, written at the time. See related: Documenting Performance Before You Ever Need To. Then sort:

• The record exists. Dated write-ups, a communicated standard, a chance to correct, and other employees held to the same bar. Proceed. The meeting is execution. See related: The Termination Conversation SOP.
• The record does not exist, and the behaviour is ongoing. The honest answer is usually to spend four more weeks building it rather than to fire today. This is the branch owners hate and it is right more often than not, for a reason that has nothing to do with fairness to the employee: four weeks of dated record converts a case you would settle into a case you would win early and cheaply, and it costs you four weeks of a performance you were already tolerating. Write a plan with a measurable standard, a date, and a review point, hand it over, and work it honestly. If the person corrects, you keep a tech. If they do not, you now have the case.
• The record does not exist and the behaviour has stopped. You cannot build a record on a problem that is no longer happening. Either the problem was fixed, in which case the case is over, or your real complaint is something else and you have not named it yet.

What would change this. If the performance problem is severity rather than pattern (one job so badly done it cost you a commercial account), it is not this branch. It is branch B, because the thing to establish is what happened on that job.

Branch B: one serious incident, established or not

Theft, a threat, violence, impairment on a site, falsifying a time card or an inspection sign-off, a safety violation that could have killed somebody. Here waiting is the wrong answer, and so is firing in the parking lot.

The criterion is whether the conduct is established or only alleged. If you watched it happen, or the evidence is documentary and unambiguous (the time card says six hours on a job the GPS log puts him at for ninety minutes), the conduct is established and you can act.

If it is reported rather than observed, the move that preserves every option is suspension pending investigation, with pay. Paying through the suspension is what keeps it from being a de facto termination, and the cost of a few days of wages is trivial against the cost of firing on a report that turns out to be wrong. Then run the investigation and let the finding, not the report, be the basis. See related: Running a Workplace Investigation That Holds Up.

The document you need on this branch is the investigation file, and no amount of write-up history substitutes for it. A shop that fires for theft with a thick performance folder and no investigation has documented the wrong thing: the folder proves he was a mediocre tech, which is not the reason given.

Where this branch has a real hazard. Impairment, a threat, or a person who has to be removed from a site now is a safety event before it is an employment decision. Get the person away from vehicles, tools and other people first, do not let an impaired person drive a company truck or their own car off the lot, and call whoever you would call for a customer in the same state. The employment decision keeps until everybody is safe.

Branch C: something protected happened in the last few weeks

The person has complained about discrimination or harassment, raised a wage or safety issue, asked for an accommodation, filed a workers comp claim, taken leave, or talked with a coworker about pay. This does not make them unfireable, and believing it does is its own trap, because a shop that freezes ends up with an untouchable employee and a crew that notices.

What it does is move the bar. Timing alone can carry an inference, so a termination weeks after protected activity has to be explicable on facts that predate it. Sort on three things:

• When was the decision actually made, and can you show it. A decision documented before the protected activity is a different case from the same decision made after it. If the plan, the final warning, or the layoff list is dated earlier, say so and be able to prove the date.
• How close is it. Days is worse than weeks, weeks worse than months. There is no safe interval, and anyone who tells you there is has invented one.
• Who knew. If the owner making the call genuinely did not know about the complaint, that helps, but only if the foreman who pushed for the firing did not know either. A biased supervisor's influence on an unbiased decision-maker can be imputed to the employer (Staub v. Proctor Hospital, 2011), so "I did not know" is not the end of the question.

This is the branch where you stop and call an employment lawyer before the meeting, not after. Walk in with the personnel file, every dated write-up, the date and content of the protected activity, who knew about it and when, and the file of anyone else who did the same thing and kept their job.

When two branches are live at once

Most real cases are A and C together, because a performance conversation is often what prompts the complaint in the first place. That combination is not a reason to abandon the performance case; it is a reason to run it visibly. Keep doing the thing you had already written down, on the schedule you had already set, and change nothing about the person's hours, route or overtime while it runs. The protection is that the plan predates the complaint and was worked unchanged. Altering the plan after the complaint, in either direction, is the move that creates the second claim.

A worked case: the four weeks nobody wanted to spend

A twelve-person residential shop wants to let a service tech go for quality. The number is real. Over the last twelve weeks he completed 96 service calls and generated 9 callbacks coded to workmanship, which is 9.4 percent of his own completed calls. The other four techs completed 384 calls over the same twelve weeks and generated 12 workmanship callbacks, which is 3.1 percent of theirs. His rate is about three times the rest of the crew's, on the same measure and the same per-call basis.

The file holds two entries. One reads "spoke to him about quality" with a date. The other reads "attitude". That is branch A with no record, so the answer is four more weeks.

The plan is written and dated: no more than 1 workmanship-coded callback in his next 32 completed calls, which at his run rate of 8 calls a week is four weeks, and which is the 3.1 percent the rest of the crew is already running rather than a number invented to be unreachable. It names the coding source so there is no argument later about what counts. It sets a review at two weeks.

In week two he tells the owner the overtime is not being spread evenly. An oral internal complaint about pay can be protected activity (the Supreme Court held in Kasten v. Saint-Gobain, 2011, that an oral complaint counts when it is clear enough to put the employer on notice). Branch C is now live in the middle of branch A. The owner changes nothing: same plan, same review date, same overtime rotation he was already running, and he writes down that he considered whether to change anything and chose not to.

At four weeks the tech has completed 31 calls, one short of the 32 the plan named because of a sick day, and produced 4 workmanship callbacks, which is 12.9 percent. The gate fails on the numerator before the denominator closes: 4 callbacks against a ceiling of 1, so the short week does not rescue it and there is no argument about the denominator to be had.

The termination is now defensible on a record that opens before the complaint, states a number, and reports a result that misses it fourfold. The four weeks cost the shop one tech's below-standard output for a month. They bought a dated plan, a mid-point note, a measured outcome and a documented decision not to retaliate.

The failure mode, for contrast. The same owner fires on the day of the overtime complaint, with the same two-line file. The stated reason is quality. The only quality document in existence is dated the week of the firing, the comparator rates were never written down, and the complaint is three days upstream. Everything he said about the callbacks is true, and none of it is in a form anybody outside the shop can check.

The line where this stops being your call

Three signals mean the next move is a lawyer rather than a page: a live agency charge or a lawyer's letter already in hand, a termination that would follow protected activity by days rather than months, and any case where the person is the only one of something in your shop and the comparator you would point to does not exist. In all three, the thing you take to the meeting with counsel is documents, not your account of events.

References

• U.S. Equal Employment Opportunity Commission, guidance on retaliation and on consistent treatment of comparators
• Kasten v. Saint-Gobain Performance Plastics Corp. (2011), oral complaints under the FLSA anti-retaliation provision, 29 U.S.C. 215(a)(3)
• Staub v. Proctor Hospital (2011), influence of a biased supervisor on a decision-maker
• See related: Documenting Performance Before You Ever Need To, Running a Workplace Investigation That Holds Up, The Termination Conversation SOP
• See related: Fire This Employee or Give Another Chance: Decision Tree, for whether to fire at all, and How to Fire an Employee the Right Way for the mechanics once you have decided