The best automation we didn't build.

A client asked us to build a system that would work out when his customers were due for service. We started designing it — then found his software had been doing it for two years.

He runs a water-treatment company. The ask was reasonable: build something that runs through the CRM, figures out that a customer's filter needs changing every year, and texts them to schedule.

WORKFLOW AUDITSLOOK BEFORE YOU BUILDthe fact youneedcrawl every jobnormalize productsbuild our own ledgerWEEKSalready in their recurrence rulesONE QUERYPull the data before you design.The answer was already there.

So we designed it. An equipment ledger — crawl every job's line items, normalize the product names, infer what each customer owns and when it was last serviced, then store all of it in our own database, because his software had nowhere to put it. We wrote the design doc. We drafted a list of questions to send him about service intervals per product.

Then, before sending any of it, we pulled his actual data.

Of his roughly 4,100 jobs, about 2,900 already belonged to a recurring series. Five hundred and forty-three series across four hundred and two customers, each carrying a real recurrence rule. Yearly. Every three months. His recurrence rules WERE the service intervals we were about to ask him for.

His software was already generating the jobs. Most of them already had a technician assigned. Most had an arrival window.

The only thing missing was anyone telling the customer.

The entire ledger was unnecessary. The build we were a day away from proposing would have taken weeks, cost him real money, and duplicated something he was already paying his existing software to do.

The answer was already in his CRM.

— We just hadn't looked.

The rule we took from it

Take the client's framing of the problem as a symptom, not a spec.

Before designing anything, pull the actual data and go looking for the fact they're asking you to derive. The question we now ask on every scoping call is: where would this already be written down, if they'd accidentally solved half of it themselves?

A business owner describing a gap is describing the gap they NOTICE. That's rarely the same as the gap that exists. He noticed that customers weren't getting reminded. He assumed that meant the system didn't know who to remind. It knew. It just never spoke.

What this means for you

If someone is about to sell you automation, make them survey your account first. Counts, statuses, timestamps, recurring rules, whatever automation your existing tools already run. Ask them to come back and lead with numbers from your own system.

Two things happen when they do. Sometimes you find out half of what you were about to buy already exists, which saves you the money. And sometimes you find out it genuinely doesn't, and now you're buying with evidence instead of a hunch.

Either way you're better off. It also changes the nature of the conversation: a vendor showing you numbers from inside your own business isn't pitching anymore, they're reporting. That's a much more useful person to have in the room.

That's most of what our free workflow audit actually is. We go look first. Sometimes the most valuable thing we can tell you is that you don't need us for this part.

Want us to go look first? Grab a free workflow audit.

← Back to all posts Free Workflow Audit →