What is different about PMing for internal users who cannot churn?

Interviewer's question: "What is different about PMing for internal users who cannot churn?" Brightcore Comfort Systems runs HVAC service calls across one metro area. Callum Bretherton has coordinated dispatch there for five years.

How to answer this, stage by stage

Nobody is grading whether you know that internal users can't churn. They're grading whether you can say what that actually changes about where you look for risk.

Let's learn

FieldMind is a tablet app that suggests likely fixes to a field technician mid-call, based on the unit type, the symptoms entered, and past service history.

Before it, a Brightcore tech would call dispatch or a senior tech for a second opinion on an unfamiliar unit, sometimes waiting ten or fifteen minutes for someone to pick up. With FieldMind, a suggestion appears in under a minute.

The turn: Brightcore didn't have a FieldMind complaint problem. Complaints stayed low the entire time. What it had was a slow, quiet migration of real troubleshooting knowledge out of FieldMind and into private notebooks nobody at dispatch ever saw.

One of these lands in an inbox. The other lands in a glovebox, and nobody at dispatch ever opens it.

At its worst: a whole crew of six techs keeps a shared paper notebook of "what actually works" for tricky rooftop units, updated by hand, carried between trucks, completely unknown to dispatch, who believe FieldMind usage is healthy because the app stays open on every tablet, every shift.

None of this was ever the plan. It's what fills the gap the moment reporting a problem costs more than working around it.

What I would leave alone: FieldMind's suggestions on simple, customer-present jobs, a filter swap, a thermostat reset, don't need this level of concern. A wrong suggestion there gets caught immediately, by the tech or the customer, in seconds.

The lesson: a captive user's silence is not consent. It's usually the sound of someone quietly building a way around you, because leaving was never on the table to begin with.

Now here is the same thing as a story

The short version above is what you'd say defending this position to Brightcore's operations director. Read this one for how the notebook actually started.

Callum Bretherton has coordinated Brightcore's dispatch desk for five years. He can tell which tech is a good fit for a tricky commercial job before he's finished reading the ticket.

For its first several months, FieldMind mostly earned trust the ordinary way: techs used it, it was usually right, and word got around. Techs would even mention it on their job notes, "FieldMind flagged a capacitor issue, confirmed," proud of the fast diagnosis.

Then came a solo call on a commercial rooftop unit, no customer around to double-check anything. FieldMind suggested a fix for a symptom that looked identical to one it had seen a hundred times before, but wasn't. The tech followed it, the fix didn't hold, and the customer called back furious two days later during a heat wave.

The fourth step never shows up in a usage log. It's the only one that actually mattered.

Word of that call spread through the crew within a day, the way bad calls always do. Nobody filed a formal complaint about FieldMind itself, since the tech didn't want to look like he'd blindly followed a suggestion. But three other techs, on their own, quietly started keeping a shared notebook of rooftop-unit symptoms and the fixes that had actually worked, cross-checked against, not instead of, FieldMind.

Here's the decision I'd take back: making "flag this suggestion as wrong" a five-field support ticket instead of a single tap. That felt like reasonable process when the team wanted structured data on every report. It stopped making sense the moment filing a ticket took longer than just quietly not trusting the tool anymore.

Replayed with a one-tap flag: the same rooftop misfire happens. The tech taps "wrong" right there in the app, in three seconds, no form. Dispatch sees three flags on the same symptom pattern within a week, a real, countable signal, and fixes the underlying rule before a fourth tech ever has reason to start a notebook.

I built the ticket-based flagging process because I wanted good, structured data on every failure. It took a furious customer during a heat wave to see that structure nobody actually used told me nothing at all.

PICK, mapped onto one bad callNot a debate about who's harder to please. PICK is what tells you which silence is actually the expensive one.

Any one of these outranks a quiet usage dashboard. All three showed up at Brightcore within a month.

The top-right corner is exactly where a wrong suggestion has nobody around to catch it.

The recap, one line per letter: position is that captive users carry more risk, not less; impact is visible cancellation versus invisible workaround; cost asymmetry is the workaround's real, uncounted cost; and kill criteria is shadow-tool usage spreading between crews.

And if you want to be sure it really works, try it somewhere elseSame four letters, a chain of veterinary clinics instead of an HVAC crew. Nothing about the two jobs is alike.

Oakhollow Animal Hospital Group gave its vet techs a triage-note assistant that drafts intake summaries from a quick verbal description. Priya Rajan has worked as a vet tech there for six years.

Mapped onto PICK: position is the same, vet techs can't churn from their job over a flawed note-taking tool the way a pet owner could churn from a clinic. Impact: an owner's silent dissatisfaction shows up in a lost repeat booking; a tech's silent workaround shows up as private shorthand scrawled on a paper chart, invisible to the practice's own systems. Cost asymmetry: after the tool missed a real allergy flag in one note, techs didn't file a complaint, they just started double-writing everything by hand as a backup, which quietly erased most of the time the tool was meant to save. Kill criteria: a rising rate of techs keeping paper backups alongside the digital tool would mean the captive-user assumption had already failed, regardless of what the login logs showed.

A different shape of picture than Section 2 used: a timeline instead of a step-by-step flow. The pattern underneath is identical.

Swap the trigger and it still runs.
Speed: an interviewer caps you at sixty seconds. Say "internal users are higher-risk because their silent reaction is a workaround, not a cancellation, so watch for shadow tools, not complaint volume," and stop.
Cost: if tracking shadow-tool usage directly isn't possible, ask techs informally once a quarter whether they keep a personal backup system, and treat a rising yes as the real signal.
The model gets better, for real: even if FieldMind's accuracy improves company-wide, a crew that already built a private notebook after one bad call may keep using it out of habit, so trust lost once doesn't automatically come back when the tool does.

Where people run it wrong.
They treat low complaint volume as proof an internal tool is trusted, without ever asking whether people quietly stopped relying on it instead.
They count an app staying open as usage, without checking whether anyone's still acting on what it says.
They make reporting a problem harder than working around it, then wonder why nobody reports anything.

How to use it live. When someone asks what's different about internal users, ask yourself one question: if this tool quietly stopped working, would I find out from a complaint, or would I have to go looking for the workaround myself? If it's the second one, that's where the real risk always lives.

Flashcards (tap any card to flip it)

Check yourself Score: 0 / 0