It all comes out in the wash.
My first manager at Microsoft was Ken Bullock—Kenny Bullock, No. 19 in Georgia Southern’s 1987 football media guide. The university’s archives place him on the 1986 and 1987 Eagles teams, including the 1986 I-AA national-championship season under Erk Russell. The 1987 guide also identifies Ken as one of three quarterbacks Russell was evaluating for the starting job. As I remember it, Ken later gave me a piece of career advice that was considerably shorter than most leadership frameworks:
Do right.
That was it. But it was not a slogan he put on a slide and forgot. It was the filter he used.
Ken had my back. He supported me, trusted me, and invested in me as a person—not merely as a resource assigned to his team. He was a good leader and, more importantly, a good person. Perhaps sometimes to his own detriment, he put real energy into his people in ways that did not always convert neatly into a metric about his own performance.
I cannot run the counterfactual, but I honestly cannot imagine what my extended career at Microsoft would have looked like if he had not been my first manager here. He helped establish what leadership looked like before I had absorbed enough of the company to confuse leadership with organizational theater.
In my vocabulary, Ken was an OG. A true baller of a manager. Not because he had once played quarterback, although that certainly completes the metaphor. He was a baller because he protected his people, made the team better, and ran decisions through a standard more durable than personal advantage:
Do right first.
I do not want to reverse-engineer a tidy leadership origin story from a football roster. Playing quarterback does not automatically make someone a good manager, and a sports database cannot explain why a leader chose to invest in a new employee years later.
Still, the resonance is hard to miss. Georgia Southern describes the program culture Erk Russell built in those years as blue-collar work, doing the most with the least, and a legacy of integrity—doing the right thing regardless of the circumstance. The university’s GATA tradition is less a slogan than a call to get the team moving once the plan is in place and it is time to do the job correctly.
I cannot say where Ken’s “Do right” filter originated. That part of the story belongs to him. I can say that the posture traveled.
A quarterback has to see more than his own assignment. He has to understand how the pieces fit together, distribute the ball, make a decision before every variable is settled, absorb responsibility when the play breaks down, and help the people around him succeed. Those are football metaphors, but they also describe the manager I encountered at Microsoft.
Ken did not need to take every snap of my career. He gave me enough protection, context, confidence, and room to start seeing the field for myself. That was the precursor. The plays changed—from customer support, to cloud, to learning, to operations, leadership, and now AI—but the first filter remained useful through every formation:
Do right.
The Georgia Southern connection has become even more meaningful with time. This year, two of my four children will be Georgia Southern Eagles, and there is a real possibility that soon it will be three of four. Georgia Southern entered my Microsoft story through the manager who helped launch my career here. Now it is becoming part of my family’s story, too.
That feels like a circle I could not have seen from my first day.
And yes: Freedom, Freedom II, and GUS the Eagle are all awesome. They are part of what makes Georgia Southern feel less like a logo and more like a living tradition—one that now stretches from Ken’s era to my children’s.
Not manage the narrative. Not optimize the scorecard. Not make sure your pillar is visible in the executive review. Do right.
I have been thinking about that advice lately because this is an anxious time to work. We are post-COVID, mid-restructuring, and somewhere in what feels like the next wave—or maybe the wave happening right now—of AI. Roles are changing. Organizations are flattening. Work that once took a week can be produced before lunch. Everyone is trying to understand what remains valuable and how to make sure their value is seen.
That anxiety is real. So is the temptation it creates.
Call it visibility management. Call it perceived impact. Call it “token maxxing,” a phrase I recently and regrettably heard used by someone far too senior for it to feel like a joke. Whatever we call it, the posture is familiar: generate more artifacts, claim more surface area, attach yourself to more initiatives, and make sure the machine registers your motion.
Another dashboard. Another workstream. Another pillar. Another status report. Another visualization. Another carefully formatted HTML report proving that everything is all green.
Or perhaps Al Green: smooth, soulful, and extremely well produced.
The problem is that production is not the same thing as value. Motion is not the same thing as progress. A measurement can tell us something useful without becoming the thing we are actually trying to produce.
And eventually, reality gets a vote.
This is not a new lesson for me
One reason the “token maxxing” language bothered me is that I recognized the underlying trap. I have fallen into versions of it myself.
At the end of my first year at Microsoft, I wrote that earlier in the year I would have tried to contribute to everything. By then, I had started to understand that effectiveness was not measured by participation. It was measured by impact. Sometimes the most useful thing I could do was connect the right people and then get out of their way.
More than a decade later, near burnout, I recognized a related mistake. I had treated availability as commitment and responsiveness as impact. Neither was true. Being reachable at every hour created plenty of evidence that I was working. It did not prove that I was doing the right work, and it certainly did not make the work sustainable.
Those were different seasons of my career, but the error was the same: confusing an observable proxy with the thing the proxy was supposed to represent.
Participation can be evidence of contribution. It is not contribution by itself.
Responsiveness can be evidence of commitment. It is not commitment by itself.
Output can be evidence of value. It is not value by itself.
The proxy becomes dangerous when it is easier to maximize than the real outcome—and especially when the people doing the measuring begin to forget there was ever a difference.
The self-licking ice cream cone
There is an old phrase for a system whose main purpose becomes sustaining itself: a self-licking ice cream cone.
You can spot one when the work exists to feed the reporting, the reporting exists to justify the work, and both generate demand for the next reporting cycle. The initiative is successful because it produced the artifacts the initiative said it would produce. The dashboard is green because the dashboard measures completion of the dashboard’s inputs. Everyone can point to activity. It is harder to point to what became meaningfully better.
These systems can survive for a surprisingly long time. They can even look impressive. But they eventually self-own because people downstream experience the difference between the representation of the work and the work itself.
The customer knows whether the problem was solved. The operator knows whether the process works. The executive knows whether the briefing answered the actual question. The teammate knows whether collaborating with you made the outcome better or merely made the meeting longer.
The line often attributed to Abraham Lincoln says that you can fool some of the people some of the time, but not all of the people all of the time. There is no known contemporary record establishing that Lincoln actually said it, which is an almost comically appropriate provenance problem for this particular essay. Whoever said it, the idea survived. There are less presidential versions of the same point:
Real recognizes real.
Game recognizes game.
Ball don’t lie.
Or the one I grew up hearing: it all comes out in the wash.
Different idioms, same underlying truth. Reality is an undefeated auditor.
Artifacts are not operating systems
I am not arguing against reports, dashboards, KPIs, OKRs, workstreams, or any of the other machinery organizations use to make work visible. I use those things. I build those things. They can create inventory, focus attention, expose dependencies, and help people coordinate.
But the artifact is not the operating system.
A collection of workstreams is not an operating model unless the pieces fit back together. A list of owners is not accountability unless somebody owns the whole answer. A sprint full of completed tasks is not proof that the underlying system improved. A weekly status email is not executive readiness if the reader still has to reconstruct reality from five different truth fragments.
This has been a recurring theme in my recent work. We are often good at creating the pieces and less good at designing how they recombine.
In one recent executive-preparation process, we had most of the things a responsible process is supposed to have: a central tracker, a recurring review, identified owners, briefing materials, milestones, and a weekly message calling out gaps and actions.
Then the updates started arriving.
One person owned the operational context. Another had the customer history. Someone else understood the technical repair. Another person was building the briefing. Each contribution was legitimate. Each person could accurately report progress on their part. Yet the executive trying to understand the engagement still had to ask the most basic question: who owns the whole answer?
That question cut through a mountain of nominal ownership. We had owners for the fragments, but the system was still expecting the consumer to perform the synthesis.
The fix was not merely a better subject line, although that helped. The deeper requirement was one accountable person who did not need to perform every task but did own the coherent answer: what is happening, why it matters, what remains open, who is doing what, and what decision or help is needed next.
That is the difference between assigning work and designing an operating system.
I have seen the same pattern at the portfolio level. A first pass at organizing work into pillars and workstreams created useful inventory, visibility, and momentum. But a collection of independent work items did not automatically become a coherent operating model. The next questions were about governance and dependency mapping: how do the pieces fit together, where do decisions get made, what happens when priorities collide, and how does bottom-up learning reach top-down prioritization?
Without those connections, every pillar can look successful inside its own boundary while the organization accumulates seams between them.
Top-down priorities without bottom-up integration become slogans. Bottom-up activity without governance becomes a pile. Many contributors without one accountable synthesizer leave the most senior person in the thread doing the integration work themselves.
The answer is rarely another artifact floating beside all the others. It is usually a clearer system: an authoritative record, visible dependencies, explicit decision rights, one person accountable for synthesis, and a communication path designed around what the consumer actually needs to understand or do.
The report can support that system. It cannot substitute for it.
A beautifully organized pile is still a pile
I have also been experimenting with using AI to organize my own work across separate projects: operating-model design, tooling and data, engagement models, knowledge strategy, evidence and impact, people leadership, stakeholder context, career development, and the personal writing that may eventually emerge from those experiences.
The separation matters. Technical machinery is not the same thing as the operating model that consumes it. A working idea is not authoritative guidance. A person’s self-report is not independently verified evidence. An internal lesson is not automatically a public story. A career hypothesis is not an established conclusion about where I should go next.
Giving those distinctions a home reduces the chance that a confident summary silently turns one category into another. It preserves useful boundaries and makes provenance visible.
But there is an obvious danger: I can create a beautifully labeled museum of my own context.
Twenty projects, perfect templates, meticulously dated summaries, and an elegant taxonomy would still be another self-licking ice cream cone if the system did not help me make better decisions, reduce repeated work, transfer context, develop people, improve an outcome, or remember something that would otherwise be lost.
Structure is valuable when it reduces entropy. Structure for the purpose of admiring the structure is bureaucracy with excellent metadata.
The test is whether the system can eventually operate without requiring me to remain its permanent human integration layer. If every connection, reconciliation, and translation still routes through me, I have documented the bottleneck without removing it.
Context is part of the truth
There is another failure mode in anxious organizations: presenting the technically correct fragment while omitting the context that would change its meaning.
A status can be green inside the narrow boundary of one team’s commitment while the broader outcome is red. A metric can improve while the customer experience deteriorates. A person can complete every assigned task while making the people around them absorb the integration cost. An answer can be factually defensible and still be wrong for the conversation it entered.
Context first.
What happened? What outcome were we trying to create? What did we know at the time? Who owns the next decision? What is documented fact, what is someone’s report, what is interpretation, and what remains an inference that needs validation?
I have started using a simple chain to discipline impact claims:
activity → output → adoption → outcome → impact
Those stages are related, but they are not interchangeable.
We held the workshop. That is activity.
We produced the model, report, dashboard, or tool. That is output.
People began using it. That is adoption.
Their behavior, decisions, speed, quality, or reliability changed. That is an outcome.
The changed outcome created durable value for customers, employees, or the business. That is impact.
Most overselling happens by jumping across one or more arrows. We built it, therefore it was adopted. People tried it, therefore it improved the outcome. The metric moved, therefore our work caused it. A team contributed, therefore one person “drove” it.
Maybe. But each arrow carries an evidentiary burden.
Honest impact reporting preserves the chain. It names contributors and dependencies. It keeps contradictory evidence instead of editing it out. It distinguishes led, created, enabled, contributed to, resulted in, and caused because those words make different claims about reality.
This is not modesty theater. It is how we learn which mechanisms actually work.
AI makes those distinctions more important, not less. It can turn scattered fragments into a confident narrative faster than any of us could do manually. That is powerful when the inputs are trustworthy and the destination is clear. When they are not, AI can manufacture polished coherence around unresolved ambiguity.
That may be the most sophisticated form of token maxxing: not merely producing more, but producing so smoothly that nobody notices we have not decided what is true, who owns the outcome, or why the thing exists.
What survives the reporting period
So what should we optimize for in a brave new world of AI, restructuring, shifting roles, and uncertain measures of contribution?
I keep coming back to four things.
-
Create real value. Make something better for a customer, teammate, operator, or organization. If the artifact disappears tomorrow, something useful should remain.
-
Be worth collaborating with. Bring context. Reduce ambiguity. Share credit. Own the answer when the pieces cross boundaries. Do not make everyone else pay an integration tax for your participation.
-
Do the right thing. Tell the truth when the status is not green. Preserve the difference between evidence and inference. Resist claiming more than the work supports. Choose the durable outcome over the flattering snapshot.
-
Think beyond the measurement window. The KPI, OKR, semester, sprint, performance cycle, or success measure du jour is a partial view of reality. Use it, but do not shrink the work to fit inside it.
This is not a case against visibility. Good work that nobody understands can be difficult to sustain, fund, repeat, or recognize. Communicating impact is part of the job.
Nor is it a case for becoming quietly indispensable. I have written before that heroes do not scale. During uncertain times, being the only person who can reconcile the trackers, explain the exceptions, repair the report, or connect the right people can feel like job security. It may even be rewarded for a while.
But “we cannot do this without you” is not always a compliment. Sometimes it is a diagnosis.
The more durable contribution is to build the warp pipe: the shared vocabulary, ownership model, trustworthy information, repeatable workflow, and transfer of judgment that allows the work to keep moving when the hero is not in the room.
But there is a difference between making value legible and manufacturing the appearance of value. One begins with the work and helps other people see it. The other begins with the desired perception and works backward toward whatever evidence will sustain it.
That distinction may blur for a quarter. Maybe even for a year. It does not stay blurred forever.
The dashboard can be green. The deck can be flawless. The HTML report can have six different visualizations and a beautiful executive summary. The model can generate another ten thousand tokens explaining why the initiative is strategic.
Eventually, somebody still asks the only question that matters:
Did this make anything meaningfully better?
It all comes out in the wash.
Ball don’t lie.
Do right.
These reflections are personal and do not represent Microsoft.
Down the rabbit hole
- Goodhart’s Law and its variants — a deeper treatment of what happens when a proxy measure becomes an optimization target. The useful warning is not “never measure.” It is that optimizing the measurement can break its relationship to the outcome we actually care about.
- Donald T. Campbell, “Assessing the Impact of Planned Social Change” — the source of the closely related observation now called Campbell’s Law: the more a quantitative indicator is used for consequential decisions, the more pressure there is to corrupt the indicator and distort the process it was intended to monitor.
- NASA’s oral history with S. Pete Worden — Worden discusses his “self-licking ice cream cone” criticism of institutional activity that becomes oriented toward sustaining its own support. The metaphor is ridiculous enough to remain useful.
- The Abraham Lincoln Presidential Library and Museum on the recurring problem of things Lincoln did not actually say — a good reminder that a durable idea and a reliable attribution are separate questions.
- The W. Edwards Deming Institute, “A Bad System Will Beat a Good Person Every Time” — the systems-level companion to “Do right.” Character matters, but leaders also have to build conditions in which doing right is possible and repeatable.
- Mustafa Suleyman and Michael Bhaskar, The Coming Wave — the book behind much of the “wave” language around AI, power, proliferation, and containment.
- Microsoft, “How Frontier Firms Are Rebuilding the Operating Model for the Age of AI” — useful context for why AI transformation is becoming an operating-model question rather than simply a personal-productivity contest.
- Georgia Southern’s 1987 football media guide, archived guide index, and 1986 season record — the university’s own historical record of No. 19 Kenny Bullock, the quarterback competition, and the Eagles program in which he played. The point here is context, not a statistical résumé.
- Georgia Southern, “A Look at Georgia Southern Football Traditions” — on Erk Russell, GATA, blue-collar work, doing the most with the least, and a culture of integrity that makes Ken’s later “Do right” leadership filter feel resonant without proving its origin.
- Ken Bullock’s public professional profile — the other side of the bridge: from the Eagles years into the long Microsoft career through which he became my first manager and changed the trajectory of mine.
- My First Year at Microsoft — where I first wrote that effectiveness is measured by impact rather than participation.
- The Boundaries That Gave Me My Life Back — the later realization that availability is not commitment and responsiveness is not impact.
- Heroes Don’t Scale (But Warp Pipes Do) — on the difference between saving the day and building a system that does not need to be saved every day.
- The Ground Moves First — on building clarity and durable systems while roles, priorities, and organizational terrain keep shifting.
- AI Doesn’t Have an Information Problem. We Do. — on AI’s ability to accelerate either trusted knowledge or beautifully synthesized ambiguity.
- Email Always Wins — a small operational example of why communication still needs context, ownership, and a path back to the source of truth.
