How to audit a services business for automation

Ask where work waits. A quote sits two days because only the founder can price it. An RFP stalls while someone digs through last year's answers for a paragraph they know exists. The strong candidates are the recurring delays the team already gripes about by name, the ones that show up in the same Monday standup week after week.

A good audit pins down the workflow, its owner, the trigger that starts it, the current turnaround time, and the dollar cost of the delay. Skip those facts and the project is a wish dressed up as a roadmap.

Most services workflows carry both. Judgment is the part a senior operator should keep, like a pricing exception or the call on whether a client is worth the risk. Labor is the repeatable work that surrounds the call, the collecting and reformatting and first-draft writing and routing.

AI earns its keep when it clears the labor and keeps the human decision in plain sight. A workflow that can't tell you where the person decides isn't ready, and forcing it through anyway buries the one judgment call that mattered.

We score each candidate on six things: how often it runs, what it's worth in dollars, whether the data is clean enough to use, who owns the workflow, the risk if it goes wrong, and how fast we can ship a first useful version.

The flashy idea usually loses. The one that wins has the volume to move a number and a clear owner who will keep using it the Tuesday after launch, once the novelty is gone.

The first project should prove a new operating loop. A rep quotes from the truck instead of driving back to the office for a number. A BD lead opens an RFP draft that's already 80 percent written rather than a blank page at 9 a.m. the day it's due.

Here's the bar we hold it to. A real workflow changes, and the team can measure the change in weeks, not quarters.