Transformation tends to conjure images of addition, a new system, a new team, a new layer of process. We describe it as building something, launching something, deploying something. The language of change is too often the language of construction.
But some of the most meaningful transformation I’ve witnessed has come from removing something.
A few years ago, I was in a room with a group of city staff in Austin, Texas. We were mapping a permitting process. The participants spanned the full reach of that process: residents who had navigated it, and city staff from planning, fire, food safety, and more. These were department leads and residents who rarely sat in the same room.
My team was doing what service designers often do, tracing a single transaction from end to end, following it through every hand it passed through on its way to resolution. We also documented the policies and systems in play.
At one point, someone described a step in their process.
“I receive this document,” they said, “and I fax it to the next team.”
The person from that next team nodded.
“And then I take it,” they said, “and I shred it. We don’t actually need it. That requirement was removed from the code years ago.”
The fax had continued because no one had ever traced the whole system together. The sender didn’t know the receiver was discarding it. The receiver didn’t know the sender thought it was essential. And residents were still being asked to complete the form that set this entire step in motion. Both teams were doing their jobs faithfully, with no reason to question a handoff that had existed for years.
Removing that step wasn’t administrative cleanup. It was transformation. And the opportunity only became visible when the process was seen as a whole.
This kind of issue happens because most processes are not designed end to end. They are designed in parts, by the people responsible for each part, without visibility into what comes before or after. Over time, systems grow around communication structures. Melvin Conway, of Conway's Law, identified this dynamic in 1967: organizations design systems that mirror their own communication structures.
A process that moves through five departments tends to accumulate five layers of handoffs. Each is locally optimized. None are examined together. The seams between departments become the places where friction hides.
In Good Services, Lou Downe defines a good service as one that requires the least number of steps necessary to achieve its goal. The emphasis is on least. Every additional step is a point of friction, a place where something can break, or where a person must compensate for a lack of coordination. Sometimes that person is a resident, completing a form that no longer serves any purpose, simply because the system has not caught up with itself.
The goal is not to add steps until everything is covered. The goal is to remove steps until only what is necessary remains.
I think about this constantly in transit technology.
Data moves through transit systems the way people move through permitting processes. It originates somewhere and then passes through a series of teams, each making adjustments. One team validates it. Another reformats it. A third adapts it for a downstream use case. A fourth, without full visibility into those changes, adjusts it again, sometimes undoing or duplicating what came before.
No single team is doing anything wrong. Each is solving a real problem within its scope. But the data has a job to do across the entire pipeline. When that pipeline is never mapped as a whole, the adjustments compound. The result is a system where data is continually reshaped, not always in ways that serve its ultimate purpose. And just like the resident completing an unnecessary form, the burden shows up somewhere, often with the person at the edge of the system, who has to navigate complexity that should have been resolved upstream.
This is the same problem as the fax.
It persists not because of negligence, but because the conditions for seeing it clearly do not exist until someone creates them.
Service design, at its best, is the discipline that creates those conditions.
A service blueprint traces a transaction from initiation to completion, across every internal team that touches it. It surfaces the gaps between what each team believes is happening and what is actually happening. A jobs map for a data pipeline does the same thing, tracing data from origin through every transformation it undergoes, making visible where teams are working at cross purposes without realizing it.
The moment of transformation is not the map itself. It is what happens in the room.
When people see the system end to end, they recognize that they have been operating on assumptions about other parts of the system, assumptions that are not entirely accurate. The fax sender assumed the document was needed. The receiving team assumed it was obsolete. The data team assumes their transformation is additive. Another team assumes the opposite.
No one is wrong. They are just working without the full picture.
When the full picture appears, the most important question is not “what should we add?” but “what can we remove?”
This is the practice I find consistently undervalued in systems work. Not the design of new components, but the audit of existing ones. Not the launch, but the retrospective that asks whether what we built is still doing what we think it is, and whether any part of it has quietly become the fax.
This also matters when designing new systems.
If a team is defining how data should flow, for example in a bus detour, they need to understand that flow end to end. What format does the data arrive in? What transformations are required? What must it do downstream? Every change must work across the full chain, not just within a single system. Otherwise, we recreate the same pattern: local optimization and global friction.
The most important work often looks invisible.
A step removed.
A handoff eliminated.
A team no longer carrying something they never needed to do.
And a resident no longer asked to complete a form that serves no purpose.
Nothing new was added.
Something unnecessary was finally seen.
And once it was seen, it could be removed.