I went looking for ways OnPay Solutions could run leaner because the manual work was hiding everywhere.
Interviews and watching people work turned it up across the business, from financial and sales reporting to customer billing to setting up payment files for new accounts.
My CTO asked me to find efficiencies, but I don't think he knew how deep it went.
Nobody was doing anything wrong. Everybody was doing it by hand.
Vision
Outsource tedious tasks to robots
I measured everything in hours, because time is the only universal. Every task takes more than zero seconds, and hours convert to money. The worst variability showed up in high-touch work like copying and pasting between systems, and I could see why.
People drag their feet on the tasks they hate, and quality falls as rework rises.
Engineering was capacity-locked on our core product. Rather than wait for headcount, I went after the capability nobody was using: our IT support interns.
At the start I did every part myself. I began with the work people hated most: the high-touch copy-and-paste, like setting up a payment file for a new customer by hand. I built the automations, ran the forecasts, and led the ceremonies, all in the open so the team could see how the work fit together.
Then I let them choose the parts they wanted to own. Adam took Product Owner and Avery took Scrum Master, and within a few sprints they were running them without me.
Me
Product Manager
Adam
Product Owner
Avery
Scrum Master
I measured everything in hours.
Time is the only universal. Every task takes more than zero seconds, and time converts to money. I tried counting touches and keystrokes; neither worked across a C-suite report and a payment file.
What the data said
Performance is as individual as the person. Two people on the same team, doing the same task, took wildly different amounts of time.
The worst variability showed up in high-touch work like copying and pasting between systems, and that is not a coincidence. People drag their feet on tasks they hate.
Quality fell as rework rose. The more times you touch it, the worse it gets.
The people doing the work wanted to do a task once and never again while getting time back.
Results
The automations gave people back 5 to 90% of the time each task used to cost, and the high-touch, high-variability work I'd flagged is where the 90s came from. It ran across eight functions, from sales to production support. The cost-benefit case I built is what earned it: the COO and CTO approved a $20,000 Automation Anywhere license when we outgrew Power Automate, and cleared us to start exploring customer-facing products.
Two interns became engineers.
A team that didn't exist paid for itself.