Your spreadsheets are a confession

No one sets out to build a monster. The monster starts as a fix. Sales needed to track something the CRM couldn't hold, so someone exported everything and added a column. Finance needed to reconcile two systems that don't talk, so someone started pasting both into one sheet every Friday. Operations needed a status view no tool provided, so a tab appeared, then a second tab, then conditional formatting, then a formula nobody dares touch.

Each of those moments was a feature request that never got filed. The business needed software to do something, the software couldn't, and a human bridged the gap by hand. Then the bridge became permanent. Give it a few years and you get what most companies actually have: a shadow layer of hand-run software that does real operational work and shows up in no org chart and no budget line.

That shadow layer isn't a mess to be embarrassed about. It's a map. Every workaround in it marks a spot where your business already decided the work mattered enough to do manually, week after week, forever. You don't have to guess what to automate. Your team already voted with their evenings.

Open your worst spreadsheet and read the columns like an archaeologist. Each kind of column names the automation that's missing.

A column filled by copy-pasting from another system is a missing integration; two tools should be talking, and instead a person carries data between them by hand. A status column someone updates manually is a missing workflow — the source system already knows the invoice was paid or the shipment landed, but nobody wired that knowledge to the place people actually look. A notes column full of shorthand is a judgment call happening off the record, which is exactly the kind of decision an agent can draft and a human can approve. And the weekly refresh ritual, that standing Monday hour where the whole thing gets rebuilt, is a scheduled job that happens to have a pulse.

Then run the math with your own numbers, because they're the only ones that count. Pick one sheet. Count the people who touch it in a week and the minutes each of them spends. Add the meetings where someone reads it aloud. Multiply by fifty and put an hourly cost on it. When I first did this with our own sheets, the result was embarrassing, and we build automation for a living. Whatever your number turns out to be, that's the salary you're paying a spreadsheet. And it doesn't include the Monday the sheet didn't get updated and someone quoted stale numbers to a customer.

We ran this exercise on ourselves before we ever ran it for anyone else. We found a hand-maintained sheet tracking our content pipeline, and numbers being reconciled by hand between two tools that both had perfectly good APIs. The worst was a weekly reporting ritual that ate a chunk of someone's Friday, every Friday.

Some of what we found became agents that still run today: reports that assemble themselves and wait for a human sign-off, records that move between systems without a courier. One item became a cheap off-the-shelf tool and ten minutes of setup, because not everything deserves an agent. And a few sheets we left exactly as they were, which brings me to the honest part.

None of this makes the spreadsheet the villain. Spreadsheets are superb at the thing they were built for: thinking. One-off analysis, scenario modeling, a quick sort of messy data, the scratchpad for a decision you'll make once. If a sheet gets built, used, and abandoned, it did its job.

The confession is a spreadsheet doing recurring process work. The tell is the calendar, not the complexity. Does this file need a human at the same time every week to stay true? Then it stopped being analysis a long time ago. It's an unpaid job that nobody ever posted, held by whoever was standing closest when the music stopped.