The handoff that shouldn't exist
Most process exists to manage handoffs that a smaller, more blended team would never have created in the first place.
Most teams treat the handoff as a fact of nature. Design finishes the mockups, writes a ticket, throws it to engineering. Engineering builds it, finds the edge cases design never considered, files them back. Product mediates. Everyone is busy, everyone is professional, and the thing that ships is a worse version of what any one of them could have made alone. We call the overhead "process" and the lost quality "the cost of doing business." It is neither. It is the cost of a structure we chose.
Here is the claim: most process exists to manage handoffs that a smaller, more blended team would never have created in the first place. The handoff is not a step in the work. It is a seam introduced by the org chart, and every seam needs ceremony to survive — specs, sign-offs, status meetings, a Jira column named "Ready for Dev." Remove the seam and most of the ceremony has nothing left to do.
The handoff manufactures the work it claims to coordinate
A spec is not free information. It is a translation, and every translation loses something. When a designer hands a screen to an engineer, they are compressing a hundred small decisions — what happens when the name is forty characters, what the empty state feels like, why the button is disabled here but hidden there — into a document that can only hold the ones they thought to write down. The engineer decompresses it, guesses at the rest, and guesses differently than the designer would have. The gap between those guesses is the bug backlog. It is also most of the review cycles, the "can we hop on a call," the second and third rounds of redlines.
I have watched a single feature accumulate four artifacts — a Figma file, a PRD, a tech spec, a QA checklist — that existed only so that four people who never sat together could pretend they had. Each artifact had an owner, a review, a place it could go stale. None of them moved the product forward. They moved information across a boundary that did not need to exist.
You cannot specify your way out of a problem that the specifying created.
When the person who designs the interaction is the person who builds it, the spec collapses into the work. The decision about the forty-character name gets made once, by someone holding both the visual intent and the actual constraint, at the moment it comes up. There is nothing to write down because there is no one to write it down for.
Most process is scar tissue around a seam
Walk into a team's process and you can usually reverse-engineer the org chart that produced it. Three planning meetings a week means three groups that don't trust each other's defaults. A heavyweight design-review gate means design and engineering have been burned by shipping something neither fully owned. Each ritual is a scar — a reasonable response to a real wound — and the wound was almost always a handoff that went badly.
The trap is that we treat the scar as the cure. We add a step, the step prevents one class of error, and we never ask whether the error existed because of a boundary we could have removed instead. Process accretes. It rarely gets garbage-collected, because removing a step feels like removing a safeguard, even when the thing it safeguards against is an artifact of the structure itself.
- →A status meeting is the price of work split across people who can't see each other's progress directly.
- →A formal handoff doc is the price of a decision made by someone who won't live with the consequences.
- →A long approval chain is the price of accountability that was diffused the moment the work was sliced into stages.
None of these are wrong, given the structure. They are correct local responses. They are just expensive, and the expense is invisible because it shows up as everyone's calendar rather than as a line item.
Blend the roles before you optimize the pipeline
The instinct, when handoffs hurt, is to make them smoother — better templates, tighter Figma-to-code tooling, a crisper definition of "done." This is optimizing the wrong thing. A faster handoff is still a handoff. You are paving the road between two buildings that should have been one building.
The alternative is fewer, more capable people who hold more of the work end to end. One person who designs and builds the feature does not hand off to themselves. A pair who sit together and own a surface together do not need a ticket to remember what they agreed; they were both there. This does not scale infinitely — at some size you genuinely need coordination, and real handoffs to real specialists earn their cost. But most teams reach for process long before they reach that size. They install the machinery of a hundred-person org at fifteen people, and then wonder why fifteen people move like five.
The test is simple. Look at any recurring meeting, doc, or sign-off and ask what seam it exists to cross. Then ask whether the seam is load-bearing or just inherited. If you can dissolve the seam — by giving one person both halves of the decision, by seating two people on the same surface, by letting the builder make the call — the ceremony around it dies quietly and nobody misses it.
The best process is the process you deleted by not needing it. Before you design a better handoff, check whether you can design one out of existence.
