FIELD NOTE |
February 25, 2026
Process Is The Surface
When a process keeps breaking, the problem may not be the steps. It may be the agreements, authority, or relationships underneath them.
When work is slow or inconsistent, the first instinct is often to fix the process.
Add a step. Remove a step. Document the workflow. Automate the handoff.
Sometimes that works.
But I've spent enough time designing and repairing organizational systems to become suspicious when something is described as just a process problem.
Process is often where another problem becomes visible.
Two teams can have a perfectly documented handoff and still fail if neither has the authority to resolve an exception. A workflow can be automated and still produce bad outcomes if the underlying definitions are contested. A checklist cannot repair incentives that reward one function for creating work another function has to absorb.
And no amount of process documentation will make a decision happen if nobody has actually been given the authority to make it.
The steps may be clear.
The agreements underneath them may not be.
A process is a series of commitments
I tend to think about processes a little differently than they're usually represented.
A process diagram shows activities, decisions, and handoffs. It tells us what happens next, who performs an action, and where the work goes.
But each of those handoffs represents a commitment.
Someone is agreeing to provide information another person needs. A team is agreeing to make a decision within a particular boundary. One function is accepting responsibility for something another function has initiated.
The steps describe how those commitments are supposed to move.
They don't necessarily tell us whether the people involved have the information, authority, capacity, or shared understanding required to fulfill them.
Consider a handoff between Sales and Finance.
Sales has negotiated an agreement with a customer. Finance needs to establish billing and revenue recognition. The workflow says the contract moves from one team to the other after signature.
Simple enough.
But what happens when the contract includes an unusual commercial term?
Does Finance have the authority to reject it? Was Finance involved before the customer agreed? Who decides whether the exception is acceptable? Who is responsible for explaining the implications to the customer?
A process can specify exactly when the handoff occurs without answering any of those questions.
And if the questions haven't been resolved, the handoff becomes the place where the disagreement surfaces.
The organization may respond by adding an approval step.
That might be appropriate.
But if the new approval doesn't clarify who can decide what, we've added another step to a disagreement rather than resolving it.
The process becomes longer without necessarily becoming more reliable.
This is why I think the agreement underneath the diagram matters more than the diagram itself.
A process designed collaboratively isn't automatically a better process. People can agree on something that doesn't work.
But the act of designing it together can reveal disagreements that would otherwise remain hidden until execution.
What does each team believe it's responsible for? What does it need from the others? What happens when the normal conditions don't apply?
Those conversations are part of the design work.
The diagram is a representation of the agreement, not a substitute for having one.
Automation doesn't resolve disagreement
Technology makes this particularly visible.
If two teams disagree about what a status means, automating the status change doesn't create alignment.
It creates faster disagreement.
If a customer can be considered active by one function and inactive by another, moving that information between systems more efficiently doesn't resolve the definition.
If ownership is ambiguous, assigning a task automatically may simply make the ambiguity harder to challenge.
I've seen organizations invest significant effort in automating workflows before deciding what the workflows are supposed to accomplish.
The result can be an impressively consistent execution of something nobody has fully agreed upon.
This doesn't mean every definition needs to be universal or every exception eliminated.
Different functions sometimes need different representations of the same underlying reality. Sales, Finance, and Customer Success may have legitimate reasons to classify a customer differently.
The question is whether those differences are understood and whether the relationships between them have been designed.
What does each definition mean? Where does it apply? What happens when information moves from one context to another?
Without those agreements, automation can turn local assumptions into organizational infrastructure.
And once those assumptions are embedded in technology, changing them becomes more expensive.
Sometimes the most important work before automation isn't technical at all.
It's getting people to make decisions they've been able to avoid while a human was managing the ambiguity.
Follow the symptom upstream
When someone tells me a process is broken, I want to understand what is actually failing.
Is information missing? Is the person responsible for the next step able to make the necessary decision? Are two teams operating with incompatible incentives? Is there enough capacity to do the work? Is the exception genuinely unusual, or has it become part of normal operations?
Sometimes I want to know what happened between the people involved.
A handoff that repeatedly fails can change how teams interpret one another's behavior. People start checking work they previously trusted. They build private tracking systems. They add protective steps because they've learned not to depend on the formal process.
Those responses may be reasonable from where they're standing.
But they also become part of how the organization operates.
What looks like an inefficient process may be an accumulation of adaptations to unresolved problems.
Remove the extra steps without understanding why they appeared, and the original problem may return.
Keep adding steps without addressing the cause, and the process becomes increasingly difficult to navigate.
This is why I don't assume that a workaround is evidence of someone refusing to follow the system.
I want to know what the workaround makes possible that the official process doesn't.
Sometimes it reveals missing information. Sometimes a decision-rights problem. Sometimes a relationship that has lost trust. Sometimes an exception the original design never anticipated.
And sometimes the answer really is that the process has unnecessary steps.
That's worth fixing too.
But I want to know which problem we're solving.
The process is where the organization becomes visible
One of the reasons I find process design interesting is that it reveals how an organization actually understands itself.
Not necessarily how it appears on an organizational chart or how responsibilities are described in job descriptions.
How work really moves.
Who people go to when they need an answer. Where decisions stall. Which teams routinely compensate for others. What information gets recreated because it isn't accessible. Which exceptions require intervention from someone several levels above the people doing the work.
A process can reveal the distance between formal authority and practical responsibility.
It can show us where a team is being held accountable for an outcome it doesn't control.
It can expose a commitment made in one part of the organization that another part doesn't have the capacity to fulfill.
And it can reveal where people have developed their own informal agreements because the formal structure doesn't adequately support the work.
That doesn't make process unimportant.
Quite the opposite.
I think process design is one of the ways we make organizational relationships explicit enough to examine and improve.
But the objective isn't simply to produce a clean workflow.
It's to create conditions in which people can reliably fulfill the commitments they're making to one another.
That may require changing the process.
It may require changing who has authority, what information is available, how incentives are structured, or what the organization has agreed to deliver.
Sometimes it requires acknowledging that the people involved have been solving a problem the organization never formally recognized.
I don't want to automate that problem away before I've understood it.
And I don't want to mistake a beautifully documented sequence of steps for evidence that the underlying agreements are sound.
Process is where the work becomes visible. It isn't necessarily where the problem begins.