The most costly thing in a small business is often the work you can’t afford to do.
Not the work you do badly. It’s the work that never happens at all, because doing it would mean buying expertise you don't have at a price you can’t afford. Historically, this has always been one of the costs of being a small business; you do what you can reach, and the rest stays stuck on the distant horizon.
But that’s started to change. This is the story of how a data export brought that home for me and reframed what’s possible.
I came to small business ownership through an acquisition, when I bought a small consumer goods company. Premium products, top shelf customer service, passionate customers. The company was about 40 years old, and while that’s great from a brand and product standpoint, a lot of the processes and tech underlying the business were long past their prime.
Our order history, inventory management, and payment processing all sat in a single, antiquated system run by an outside vendor. That vendor reveled in its role as gatekeeper. Credit card processing? “Use our [expensive] preferred vendor”. Want to offer customer financing? “Use this [clunky and obscure] preferred vendor, unless you want to pay for a custom integration.” How about getting a basic sales report? “Learn a computer programming-style query language from a thick, badly written manual, or pay by the hour to have the report run.” No help forums or YouTube tutorials. If you didn't know the magic words, you paid someone who did.
We knew we were leaving within just a few months of my taking over, as part of a broader website redesign. We’d launched the new site on November 4th, and were excited about the new stack, as it modernized the inventory management and payment processing legs of the stool.
Except for one thing. We had no way of getting our customer history out of the legacy system. All of our order records (serial numbers, purchase history, inventory sell-through info, everything) were locked in the old system with no obvious way of getting it out. Leaving the vendor meant losing all of it.
We had two bad options. Either pay the vendor a few thousand dollars to export the data into a mess of indecipherable database files, or renew for another year (about $10,000) just to keep reading the history we created and already owned. Ten grand to stand exactly still.
I was resigned to paying. Despite all my anger at the vendor, this felt like the cost of doing business with a software suite built so the little guy really couldn’t leave. The work of getting our own data out in some kind of a usable form felt way beyond what we could do in-house.
Then a friend changed everything when he asked: “Have you tried asking ChatGPT?”
This was late 2024. I’d treated AI as a novelty until then. “Write this in the voice of a pirate”-type stuff.
But after an hour of poking around our server, following ChatGPT’s step-by-step instructions, the whole situation changed. It turns out the “complex” legacy order history system was just a wrapper around a very standard SQL database, a widely used data architecture. It was never exotic, it was just covered in the veneer of proprietary complexity. Neither I nor anyone else on our team knew the difference. The vendor’s moat wasn’t technology. It was our lack of knowledge.
I didn’t know the first thing about SQL (still don’t). But after following the steps, asking questions when I got stuck, and running the export, I had our files.
Relief! I dragged the CSVs straight to Google Drive, happy that we at least owned our own data and had managed to achieve that without paying that few grand that the vendor wanted for the export. All for the price of a Saturday afternoon at the office (and $20 for a monthly ChatGPT subscription).
But that was only part of the solution, and my celebration was premature. Raw CSVs straight out of a SQL database are nearly impossible to use manually, or for a team of non-technical people. You can’t just look up a customer name or a serial number, or find out a product’s warranty status. I’d freed our data. Making it usable was a separate problem. And December 31st was still looming.
More to come next week.
Zain
P.S.: I’m curious to learn what’s sitting in your “can't afford to do it” column. Reply and let me know.