Certificates: a tool we chose, then measured.

1The problem

Certificates of insurance hold up a client’s job site, and every one handled by hand interrupts someone on the service team. Requests arrive in a shared mailbox all day, mixed in with everything else, so nobody could say how long a certificate took or how often one had to be redone.

2How it worked before

• Certificate requests answered from the shared service mailbox, with no clock on them.
• No record of which requests went through the certificate tool and which were handled by hand.
• Rework found only when a client or a certificate holder complained.

3What we built

• The agency chose an outside certificate tool. What we built is the measurement around it.
• 12,108 service-mailbox emails from the period rebuilt into 621 certificate requests.
• Requests matched to the certificate that answered them where both ends were clear, timed in business hours and split by path: the tool or a person.
• Requests without a clear start and finish reported as unpaired rather than guessed: 407 of the 621 were measurable.

4The verified result

This compares the two paths over the same months, not a before-and-after: nothing was measured before. The manual path includes the unusual requests the tool does not handle, so it is not a like-for-like speed test.

5What it means for the business

• The owner can see which kinds of certificate request still need a person, and why.
• A tool’s value is now a number the agency measured, not a figure from a sales page.
• The same measurement works for any agency with a shared service mailbox: it is how our audit starts.

Not measured yet

Hours returned to the service team. Turnaround tells you how fast; it does not tell you how much time a person got back, and we will not estimate it for you. When we measure it, it goes here with its method.

What did not work

We tried letting an automated job file incoming documents from the shared mailbox straight into the agency system. It flagged far more documents than it could file with confidence, so we changed its job to flag and prepare, with a person filing. It is paused while the operations machine is replaced. We also tested automation side by side against an outsourced back-office contract for several weeks; the side-by-side check did not pass, and that work went to a person on the team. Automate the machine-shaped tasks first.

Published October 2026. Sources and dates are kept on file. Names of staff and clients are left out.

All case studies · Every solution and its status