Zeeshan Khan

Where the Engineer Sits Decides What the Product Becomes

Essay

~5 min read · October 2026

Forward Deployment

In the first essay I said the thing that actually changed about field engineering is who pays for it. The vendor does now. Then I said the paperwork that used to fall out of handoffs is gone, and that the spec for a model-based system can only be written on site. Last time, that whatever gets built out there is a bug report somebody has to read.

All of it points at one decision. It's the one most companies make without noticing they've made it: where the field team reports.

Sounds like an admin question. Whose org chart, whose budget. It's not. It decides what the team is for. That decides what they do when a customer asks for something the product can't do. And that decides what the product turns into over the next three years.

Three places it can go

Under sales, it becomes pre-sales. No insult intended; pre-sales is a real job. But now the incentive is the deal. Whatever gets built in the field gets built to close, and once it closes there's another deal. Nothing comes back. The custom directory grows and nobody in product ever hears about it, because product wasn't in the room.

Under services, you've built a consulting firm inside the building. That's the warning from the first essay, and it's the outcome I've seen most. Utilization target, billable hours, a P&L that's pleased when the customer needs more help. The incentive runs backwards. A gap in the product is revenue. Nobody on a services P&L is going to argue hard for closing the gap that pays their team. Why would they.

Under product, it's expensive. This is the honest cost of the right answer. The team doesn't close deals and doesn't bill. It's pure spend on the product line and every quarter someone will ask why. There's only one reason that holds: it's the single part of the company that finds out what the product can't do, at the customer's site, with the customer's data, and comes back with the answer.

The field station

Agricultural research has field stations. People out in actual dirt, growing the actual crop, in whatever weather shows up. The station exists to send results back to the lab. Samples, measurements, what grew, what died.

Now suppose someone decides the station should pay for itself. Sell the harvest. Perfectly reasonable on a spreadsheet. Within a few seasons the station is a farm. Same land, same people, same sign on the gate. But it's growing what sells, not what the lab needs to see, and the samples stop coming home because there's a harvest to get in.

Nobody decided to stop the research. They decided the station should have a P&L. Everything else followed on its own.

A field engineering team under services is that farm. It's producing. It's often quite good at it. It isn't sending anything back.

What reporting to product actually means

Not a box on a chart. Four things, and I'd check for all four before believing anyone who told me their field team was product-led.

The team's manager is in product planning. Not invited to present once a quarter. In the room when the roadmap is fought over, with last month's bespoke-work list in hand.

The triage from the previous essay happens, and someone senior in product does it personally. If the head of product can't tell you what the field team built last month and what got decided about it, the reporting line is fiction.

The hard cases committed on site live in the product's test suite. Not in a customer folder. When the model gets swapped, they run. That's the spec coming home.

The team is measured on what it returns. Fraction of bespoke work decided. Number of customers using promoted work. Days from field finding to roadmap decision. Not utilization. Utilization is the farm's number.

The leader

I'll say the hands-on thing once, since the earlier series said it plenty.

Whoever runs this function should read the returns themselves. Not the summary. The list of what was built, the hard cases, the decline reasons. It's the only way to know whether the thing you're being told is working is actually working. A leader who reads the dashboard and not the returns has a field station on paper and a farm in fact, and won't find out for a couple of years.

An hour a week. Maybe two. The time isn't the hard part. Being the kind of leader who does it every week is, and that's rarer than the hour.

Where I'm probably wrong

Some companies are too small for any of this. Three engineers who all do everything, the reporting line is you, and the question is moot until suddenly it isn't. I've written this for companies past that point and should have said so at the top.

Some field work should be a services business. A customer wants a migration nobody else will ever need and will pay well for it? Take the money, put it under services, don't pretend it's research. The mistake isn't having a services arm. It's putting the research inside it.

And product has its own way to fail. Product can get precious. It declines everything the field brings back because it doesn't fit the vision, and the field team learns to stop bringing things back, same as under sales, just for a nobler-sounding reason. The reporting line is necessary. It isn't enough. I don't have a fix for that beyond the leader reading the returns and noticing when the decline rate hits a hundred percent.

Where this series ends

It started with money. Who pays for the person closing the gap between the product and one company that bought it. It ends somewhere harder to count: who reads what that person finds.

The engineer has to be in the customer's building. That was the argument for being on site and I still think it holds. The data won't travel and the expert won't. But being there is half the job, and it's the half everyone writes about. The other half happens when the engineer comes back, and it was never the engineer's call.

Where they sit decides it. Decide it on purpose.