Your Best Performer Should Be a Laboratory, Not a Weapon

Every team has one.

The designer whose files never come back. The developer whose deploys do not need a follow-up. The copywriter whose drafts go out roughly as written. The account manager who has somehow never sent a client the wrong version.

When an organization starts measuring quality, this person becomes visible in the data, and something predictable happens next.

Their number becomes the standard.

Not officially. Nobody writes it down. But it enters the room: Jane manages 0.3, why is everyone else at 1.4? And from that moment the organization has converted its single most valuable source of information into a threat.

This is the ninth article in this series, and it is about what to do with an outlier instead.

First, Verify That They Are Actually Better

Before building anything on a strong number, check that the number means what it appears to mean. There are three ways to look excellent that have nothing to do with being excellent.

Their work might be easier. The previous article, No Rate Without a Denominator, covered this in detail. Someone working on stable, templated, well-understood deliverables will have a lower defect rate than someone doing novel work, and the difference can be larger than any difference in skill.

They might be catching their own defects silently. This is the most common case and the most misleading. A person who finds and fixes fifteen problems in their own work before anyone sees it produces zero findings in the log and enormous value that is completely invisible to the measurement. They are not defect-free. They are the best QA in the building, operating on a sample size of one.

They might not be reporting. If the culture punishes findings, the person with the cleanest record may simply be the one who has learned fastest.

Only the second of those makes them worth studying, and it happens to be the most likely one. Which is fortunate, because it is also the one that generalizes.

The Difference Is Almost Never Talent

When you actually sit with a consistently reliable person and walk through how they work, the finding is nearly always the same, and it is nearly always disappointing to anyone hoping for a story about craft.

They have a private control.

Something small, undocumented, invented by them, and run every single time. It usually takes under two minutes and it usually sounds trivial when described out loud.

They send every email to themselves first and read it on a phone.

They open the staging link in a private window, because they got burned once by a cached session.

They keep a text file of the client's product names with the correct capitalization, and they paste from it rather than typing.

They read the copy backwards, sentence by sentence, when checking for typos, because it stops the brain from autocorrecting.

They never trust a date that arrived in a chat message. They open the source document.

They click every link. Not the important ones. Every one.

None of that is talent. All of it is a habit that compensates for a specific, known failure mode, usually acquired by having been burned by that exact failure mode at some point in their career.

Which means it is transferable, and it is being wasted as long as it lives in one person's head.

Extracting the Control

The extraction is not a survey. Asking someone "what do you do to maintain quality?" produces generic answers, because these habits are so automatic that the people running them do not experience them as procedure.

What works is narrower.

Watch them do a real deliverable, end to end, and note every action that was not in the brief. Those actions are the control set. Most of them will surprise the person doing them when you point them out.

Ask about scars instead of methods. "What has gone wrong for you before?" produces the story, and the story contains the control. People remember the incident far better than the habit it created.

Ask what they check last. The final pass before sending is where the personal controls cluster, because that is where the fear is.

Ask what they never trust. This is the highest-yield question. The answers are specific and immediately actionable: dates from chat, prices from a deck, screenshots of copy, anything forwarded, anything that says "final."

You are looking for the gap between the documented process and what the reliable person actually does. That gap is the deliverable.

Turning a Habit Into a Control

A habit extracted and then written into a wiki page is a habit that has been buried with extra steps.

The point of extraction is to move the control out of the person and into the system, and there is a rough hierarchy of how durable that can be.

Automate it if a machine can. Link checking, spell checking, image dimension validation, staging-link verification, alt text presence, tracking parameter presence. If the reliable person is doing something a script could do, you are paying a professional to be a cron job, and the script does not get tired at six in the evening.

Put it in the artifact if it cannot be automated. A field in the brief that says which document the pricing came from. A required source link next to any date. This is stronger than a checklist because it changes what "finished" looks like rather than adding a step after finishing.

Put it at the boundary if it needs a human. The Trust Chain argued that review effort should concentrate immediately before a crossing. A personal habit that only fires before client delivery belongs there for everyone.

Write it down only as a last resort, and expect the least from it. Documentation is where controls go to be technically available.

The test for whether the extraction worked is simple: the control should now run when the reliable person is on holiday.

What Weaponizing Them Actually Costs

Using the outlier as a standard is not merely unfair. It is expensive in three specific ways, and the costs arrive in a predictable order.

You lose the information first. Once a person's number becomes the bar others are measured against, they acquire an interest in that number. Their private controls become a competitive advantage rather than something to share, and they may not even do this consciously. The most valuable thing they have, the knowledge of exactly how work goes wrong, quietly stops circulating.

You lose the team's trust second. A standard set by an outlier is, by construction, a standard most people cannot meet. A target that most of a team fails does not produce improvement. It produces either resignation or creative reporting, and usually both.

You lose the person last. Being held up as the reason colleagues are being criticized is socially expensive. People in that position tend to either lower their visibility or leave, and both outcomes cost you the practice you never bothered to extract.

The organization ends up with the outlier's number in a slide deck and none of the outlier's methods in its process, which is close to the worst possible trade.

The Reframe

The useful sentence is short.

An outlier is not a target. An outlier is a hypothesis about what your process is missing.

Held that way, a consistently reliable person becomes the most productive thing in a quality program. Every one of their private controls is a defect class your process does not currently defend against, discovered for free, already validated by someone doing real work under real deadlines.

Held the other way, they are a stick, and sticks do not generate information.

The Version That Also Works Downward

There is a mirror image of this that is worth stating, because it follows from the same logic and is less obvious.

The person with the highest defect rate is also a laboratory.

Not a suspect. A laboratory.

If someone genuinely produces more defects than their peers on comparable work, the interesting question is what is different about their situation, not what is wrong with their attention. Different tools. Different intake. Different clients. A different point in the workflow. Missing context that everyone else acquired informally two years ago and nobody remembers transmitting.

Every one of those is a finding about the system, and the person experiencing it has better access to it than anyone else in the building.

That framing only survives if the organization has genuinely separated fallibility from non-compliance, which was the subject of A Blame-Free Culture Is Not an Accountability-Free Culture. Without that separation, "let's study what is different about your situation" is heard as the opening of a disciplinary conversation, and the person will answer accordingly.

What This Buys You

A team where the best practices have been extracted, automated where possible, and installed at the boundaries has a property that a team of individually careful people does not: its quality does not depend on who is working today.

That is the actual goal. Not a roster of exceptional individuals, which no organization can guarantee, but a process that produces the outlier's results from ordinary attention.

The outlier does not become less valuable in that world. They become the person who finds the next control, because they will keep finding them, because that is what they were always doing.

You just stop paying for it once and using it once.