A broken process, automated, is a faster mess

We once turned down exactly the work a client asked for, and it is still the best sales story we have.

The request was straightforward: automate our quote approval process. The process, once we mapped it, was a quote traveling through four people, two of whom added nothing but delay. One approval step existed because of a pricing mistake someone made in 2019. The person who made it was long gone. The step remained, a small monument to an old error, adding a day to every quote since.

The client wanted that process automated. Faithfully. All four stops.

We said no, and here is the reasoning, because it is the closest thing we have to a company motto: a broken process, automated, is the same broken process running faster. Automation is an amplifier. Point it at a good workflow and it multiplies the good. Point it at dysfunction and you get dysfunction at scale, now with fewer humans in the loop to catch it.

You can automate your way into a mess with surprising efficiency. Error rates do not fall when you automate a process with a flawed step. The flaw just executes more consistently than any human ever managed. And there is a nastier effect: automated processes calcify. A manual bad process gets quietly worked around by sensible people. An automated one becomes infrastructure. Changing it now requires a project, so nobody does, and the 2019 mistake becomes permanent at machine speed.

So the redesign comes first, always, and it is usually the highest-return hour of the whole engagement. In the quote case, the fix on paper took an afternoon: two approval steps deleted, one threshold rule added so small quotes skip review entirely. Cycle time dropped by half before we wrote a single line of automation. The client stared at the whiteboard and asked why nobody had done this years ago.

The answer is that nobody had ever drawn the process. That is the part I find genuinely strange after years of this work: companies run their revenue through workflows that exist nowhere except habit. Not documented, not designed, just accreted. The first complete drawing of a core process is routinely the first time anyone, including the owner, has seen the whole thing.

Then, on the redesigned process, we automated. And the automation was better for the redesign: simpler to build, cheaper to run, and pointed at steps that deserved to exist.

If you are lining up automation projects right now, steal the sequence. Draw the process as it actually runs, not as the manual claims. Delete before you optimize. Question every step that exists because of something that happened years ago. Then automate what survives.

The whiteboard is cheaper than the software. Use it first.