The coffee had gone cold because I spent forty minutes trying to solve a permissions error for a user named jwilson, a name that appeared on every audit sheet and every license tally we produced for the board, only to find out that jwilson had not actually stepped foot in the building or logged into the remote environment since the previous administration.

It was a small, ordinary failure of data. The spreadsheet said he was active. The licensing server said he was a “seat” that we had paid for, accounted for, and defended during the last budgetary skirmish. But the human being attached to the name was a ghost. I looked at the list of thirty-four active users and realized that I could only name five people who had actually called the help desk in the last six months.

The Headcount Proxy

We are taught to love the headcount because the headcount is legible. It is a clean, round number that fits into a PowerPoint slide without requiring any messy explanation of human behavior. If we have fifty seats, we have fifty users. If we have fifty users, we are a department of a certain scale.

The license is a proxy for importance. The license is a proxy for growth. The license is a proxy for the very existence of the infrastructure we spend our lives maintaining, even when that infrastructure is mostly serving as a very expensive digital storage unit for people who have forgotten their passwords and have no intention of ever remembering them.

I recently found myself scrolling through old text messages from my predecessor, the guy who handed me this “thriving” remote desktop environment. He had sent me a series of warnings that I’d ignored at the time, cryptic notes about “the zombies in the rafters” and “the furniture that eats the budget.”

I thought he was being cynical. I thought he was just tired. But as I sat there with my cold coffee, looking at the “active” status of jwilson, I realized he was describing a fundamental law of remote work:

They are the ones who forget the MFA tokens, the ones who need the VPN reinstalled every single time they use it, and the ones who occupy a licensed seat that could have been used by someone-anyone-who actually needs to get work done. These are the spectators of the digital workspace.

The Concentrated Territory

We treat every unit in the license count as equal, but the territory is wildly concentrated. It is a power-law reality that we refuse to acknowledge because acknowledging it would mean admitting that our “scaling” is often just a slow accumulation of digital dust.

The vendor sees a stable block of revenue. They see thirty seats and they see a successful deployment. They see thirty seats and they see a reason to send a renewal notice. But the help desk lead, the person who actually fields the requests and listens to the frustration, knows that those thirty seats represent about five people who are doing 90% of the work.

“Ruby S.K., a virtual background designer I worked with during the peak of the remote-work pivot, once told me that she spent half her time creating hyper-realistic corporate environments for people who would only ever see them once.”

– Ruby S.K., Designer

She designed these beautiful, minimalist lofts and sun-drenched libraries for users who would log in, check their email, realize they didn’t want to be in the remote environment, and log out. She was designing for ghosts. She was painting the walls of a room that nobody lived in.

This concentration of usage is the great hidden truth of the modern server room. The map shows an even distribution of “User CALs” across the organization, but the territory is a handful of power users huddled around a single metaphorical campfire while the rest of the landscape is dark. We pay for the light to be on in every corner of the map, regardless of whether anyone is there to see it.

A Liability Called “Reach”

The pressure to maintain these numbers comes from a desire for legibility. Management wants to see a chart that goes up and to the right. They want to see that we are “expanding our reach.” But when you are in the trenches of technical support, you realize that reach is a liability if it isn’t backed by actual utility.

Every licensed seat that isn’t being used by a daily driver is just another point of failure, another credential that can be compromised, and another bill that has no ROI. We were paying for the privilege of managing jwilson’s non-existent problems.

The realization hit me that we had been buying packs of licenses like they were bags of flour-commodities to be stored in bulk-rather than treating them as the specific access tools they are. We had been over-provisioning for a phantom population.

When we finally decided to stop feeding the ghosts, we realized we needed a way to align our licensing with the five faces we actually saw every day. We shifted our focus toward right-sizing, toward understanding that a “seat” is only valuable if there is a person sitting in it.

This required a level of precision that our previous vendors hadn’t encouraged. They liked the bulk. They liked the “nominal” user. But we needed the actual user.

Surgical Precision in Licensing

In our search for a more surgical approach to our Windows Server environment, we looked for partners who understood that the license count should reflect the work, not the headcount.

We found that the RDS CAL Store allowed us to stop guessing and start matching our purchases to the concentrated reality of our help desk tickets.

Instead of buying for the 150 names on the HR roster, we started buying for the 40 people who actually knew how to log in. It was a quiet revolution in the way we handled our budget.

The problem with the “nominal seat” is that it erases the human effort required to maintain it. Every time we add a user to the environment just to “have them there in case of emergency,” we are adding a grain of sand to the gears of the help desk.

We are creating a record that has to be audited. We are creating a security profile that has to be monitored. We are creating a seat that has to be paid for. And for what? So that a spreadsheet looks “complete”?

Budget-Eating Furniture

I think back to those old text messages. “The furniture that eats the budget.” A licensed seat that isn’t used is exactly that-a piece of digital furniture. It sits in the corner. It gathers dust. It doesn’t contribute to the output of the company, but it still costs money to keep it there. It still has to be insured, so to speak. It still has to be accounted for during the quarterly “cleaning” of the server.

We had been so afraid of “under-licensing” that we had become a graveyard of “over-licensing.” We had prioritized the fear of a Microsoft audit over the reality of our actual needs. We had been sold on the idea that more is safer, that a larger buffer is a sign of a well-managed department. But a buffer is only a buffer if it’s protecting something. Our buffer was just a collection of names like jwilson.

Those were the people who mattered. The other twenty-five seats in our “block of thirty” weren’t assets. They were distractions. There is a certain kind of peace that comes from deleting a ghost.

When I finally removed jwilson from the active user list, the server didn’t crash. The company didn’t fold. The only thing that happened was that our “legible” count became slightly less impressive on a slide, and slightly more honest in the bank account. We were no longer paying for the silence of the ghosts.

The vendor will always want to sell you the seat. They will always want to sell you the idea that every employee is a potential user. But the practitioner knows better. The practitioner knows that usage is a jagged, lopsided thing. It is a few people doing a lot of things, and a lot of people doing nothing.

We eventually reached a point where our license count matched our face count. It was a smaller number. It was a humbler number. But it was a number that represented actual work being done in the actual world. We stopped buying for the “nominal” and started buying for the “essential.”

It turned out that the help desk was right all along. The same five faces were the only ones that ever mattered, and the other twenty-five were just a tax we had been paying for the comfort of a round number.