“Our systems are a mess” is the most common opening sentence in my inbox, and it’s always wrong in an encouraging way: the mess is never uniform. A handful of load-bearing problems generate most of the pain, and four checks find them fast.
Check one: walk a lead end to end
I take one real lead and trace it from first touch to money: where it entered, when each response happened, who touched it, where it waited. That single trace usually surfaces the biggest revenue leak in the business — a form that emails an unwatched inbox, a rep assignment that takes a day, a follow-up that stops after one attempt.
Check two: count the re-typing
I ask each person to show me anything they type into one system after reading it in another. Every instance is a tax paid in hours and errors, and each is an automation candidate with instantly calculable ROI. Finding a person who spends ninety minutes a day moving data between tabs is depressingly routine.
Check three: find the silent failure points
Which existing automations could fail without anyone noticing? I check for error alerting on every Zap, scenario, and workflow. The answer is usually “none of them,” which means the business discovers breakage through customer complaints. Adding alerting is cheap and pays immediately.
Check four: map the tribal knowledge
Whose head holds each system? If one person’s departure would orphan the billing automation, that’s a business risk masquerading as an IT detail. The audit output ranks all of it — leaks, re-typing, silent failures, key-person risk — by cost, so the fix order is an economic decision, not a vibes one.
A logistics client’s audit found their bottleneck wasn’t any broken tool — it was 300 people queuing behind a 4-person data team for answers. The fix was a copilot, not a cleanup. That’s why the audit comes before the shopping list.
Related service: AI Automation Consulting · Proof: Natural-language operations dashboard for a logistics company