Most workflow problems are not productivity problems. They are information problems: unclear ownership, scattered documentation, and the same questions answered for the fifth time. This post describes the approach I use to untangle them.
Where Workflows Break
- Handoffs: work stalls at the boundary between people, because “done” means different things to each side.
- Scattered truth: the process lives in someone’s head, a chat thread, and three documents that disagree.
- Repeated questions: every repeated question is a symptom of documentation that failed its reader.
- Invisible work: nobody can see status, so status gets asked about in meetings instead.
The Pattern That Fixes It
- One source of truth: a single canonical place for each piece of process knowledge. Everything else links to it, never copies it.
- Named owners: every step has a person, not a department. “The team” owns nothing.
- Templated recurrence: if a task repeats, it gets a template: checklist, structure, or boilerplate. Recurring work should not be re-invented.
- Automate the mechanical: anything that is copy-paste, renaming, syncing, or reformatting is a machine’s job.
A Working Example
This website runs on that pattern. Portfolio projects live as structured notes in an Obsidian vault; each note’s frontmatter declares its publication state. A small pipeline reads those notes, builds the WordPress posts, and keeps the vault’s dashboard view in sync with what is live, no duplicate tracking, no status drift between systems.
The same thinking scales from personal projects to team documentation systems: define the source of truth, name the owner, template the repeatable, automate the mechanical.