Making the call
Workflow automation vs. process redesign: which comes first?
Redesign first, automate second—because automating a broken process does not fix it, it just makes the mess faster. Software is very good at doing the same thing a thousand times a day. If the thing is wrong, you have industrialized the wrong thing.
The honest nuance: not every manual process is broken. When the steps are right and only the execution is slow, automate immediately. The skill is telling the two situations apart before money moves, and that is what this page is for.
The sequence that works
Whether you do this with a whiteboard or with a partner, the order does not change:
- 01
Map what actually happens
Not the documented process—the real one. Sit with the people who do the work and trace one order, quote, or claim from start to finish, including the side emails and the spreadsheet nobody mentions in meetings.
- 02
Mark the steps that earn their place
For each step ask: if we started this company today, would this exist? Approval layers, copied data, and status meetings usually fail the test. Decision points and quality checks usually pass.
- 03
Redesign the smaller process
Remove the steps that failed, reassign the decisions to the people with context, and agree on the result with everyone who touches it. This is a conversation measured in days, not a transformation program.
- 04
Automate what remains
Now software earns its keep: routing, validation, escalation, integration with the ERP. Every automated step is one you would defend on its merits.
- 05
Watch it run, then adjust
The first version teaches you what the map missed. Expect a round of refinement in the first weeks of real use—and treat that as the process working, not failing.
When automation alone is enough
Automate without a redesign phase when the process is sound and the pain is mechanical. The steps each have a clear purpose, the right people already make the decisions, and what kills the team is re-typing the same order into three systems, chasing signatures, and assembling the same report every Friday. Manual data entry and status-chasing are the classic cases: nobody defends the typing; it just has not been engineered away yet. These are the fastest wins in operations, and they build the organizational trust that funds the harder work.
When redesign has to come first
Redesign first when the process carries history instead of logic. Quotes that require three approvals because of one bad deal five years ago. Reports assembled for meetings that no longer happen. Work done twice because two departments never agreed on who owns it. Bottlenecks where everything routes through one person—not because they are the right decider, but because the process grew around them.
You can recognize these by a simple test: ask why a step exists, and the answer is a story instead of a reason. Automating on top of that just hardens the story into software. This is also why the first build should be chosen carefully—the workflow worth automating first is the one that is both expensive and structurally sound, or cheap to redesign.
Common mistakes on both sides
- Automating the process as documented. The documented version is a fiction from the last audit; the real one lives in people’s habits.
- Redesigning forever. Analysis that never reaches a build is just expensive sympathy. Timebox the redesign and commit.
- Letting the tool dictate the process. Buying software first and contorting the operation to fit it is redesign done by a vendor’s product team, for their average customer.
- Redesigning without the people who do the work. The ops manager and the person who actually keys the orders will describe different processes. Only one of them is true.
- Treating automation as the finish line. The system that runs the redesigned process still needs an owner and a habit of refinement.
Frequently asked questions
- Walk it with the people who do it. If every step exists for a reason and the pain is purely re-typing, waiting, and chasing, it is manual—automate. If steps exist because of a system limitation from years ago, approvals sit with people who no longer have context, or work gets done twice in different places, it is broken—redesign first.
- It forces consistency, which is not the same thing. Automation will make people follow the process as designed—so if the design routes quotes through a manager who adds no value, you have now made that detour mandatory and faster. The flaws become permanent features.
- For a single workflow, redesign is days of structured conversation, not months of consulting. Map how the work actually moves today, mark every step that would not exist if you started fresh, and agree on the smaller process. The discipline is in resisting the urge to skip ahead to tools.
- That is normal, and it is the finding. When three people describe three versions of the same approval flow, the process does not really exist yet—it is three habits sharing a name. Reconciling those versions is the redesign, and it has to happen before any software can encode the result.
- The person who owns the outcome, the people who do the work daily, and whoever approves the exceptions. Keep it small—three to five people. Leave out anyone whose only role is to be kept informed; inform them afterward. The common failure is a room full of stakeholders describing how the process should work, with nobody present who can say how it actually does.
- The redesign is built into how Praxyt works. Every engagement starts by mapping the workflow as it actually runs—including the steps nobody admits to—and the build proposal reflects the simplified process, not the accumulated mess.
Keep investigating
- How to Find the Workflow Worth Automating FirstRank workflows by friction, cost, and risk before committing to a build.
- Workflow BottlenecksThe queue-shaped problems that signal a process needing redesign.
- Workflow SystemsWhat automation looks like when it is built around a sound process.
- How It WorksPraxyt’s engagement model—mapping and redesign are built into every project.
- Internal Tools vs. SpreadsheetsWhy rebuilding the spreadsheet tab for tab fails the same way.
- How Long Does an Internal Tool Take to Build?Where redesign fits in the phases of a build.
Suspect the process, not just the tooling?
A Workflow Review maps how the work actually moves—and tells you what to fix before anything gets automated.