In the last piece, I argued that your employees stopped having ideas. Not because they ran out of creativity, but because the cost of turning an idea into something real got so high that having the idea stopped being worth the letdown. You can train a workforce into learned helplessness without ever meaning to. Most companies have.
I want to pick up exactly where that left off, because I skipped an obvious question.
If people stopped acting on their ideas, where did the ideas go?
They went into spreadsheets.
Which is the best thing that could have happened, because every spreadsheet quietly running your business is a precise map of what your company needs you to build, and in an AI-forward company it's also the last version of that workflow anyone should ever build by hand.
Your employees have stopped having ideas
Read the last article hereAn illusion of progress
An idea shows up. The historically honest path to making it work would run through IT, budgets, and bandwidth. But there's a blank spreadsheet one click away, and nobody has to approve it. So you open it. You start typing. You build a version of the thing yourself, by hand, in the one tool you're allowed to use without asking.
It feels like progress. For a while, it even is.
That instinct, to build the thing yourself rather than wait for permission, is one of the most important things happening in your company, and almost nobody reads it correctly. So let me show you what it looks like once it's been running for a few years. Three real examples.
The first is a multi-location operation that exports data from their ERP every morning. The export hits a distribution list. Someone catches it, parses it into a set of Google Sheets, one per region, and those sheets feed a KPI dashboard the executive team uses to make real investment decisions. Where to put capital. Which locations are struggling. Whether a region needs new leadership. The most consequential conversations in the company are running on a chain of manual exports and a few tabs of formulas held together by one person's morning routine.
The second came to us after they shipped a load of building materials to the wrong location. Real cost, real customer. The root cause was a broken formula in a spreadsheet that had worked for years and had quietly stopped working the week before.
The third business was sending customers their usage data two weeks late. Field operators logged it on paper, one office person keyed it into a master sheet and sent it out, and customers were getting information about their own equipment on a two-week delay. One person, one spreadsheet, one queue.
You likely have your own version of this. Most companies have several.
Built-in validation.
These companies are not wrong for operating this way. I mean it. They aren't lazy, cheap, or unsophisticated. They had a workflow that had to happen, and they made it happen using the only tool that was ever within reach. The fact that it is fragile and manual doesn't make the workflow invalid. It means the tooling was wrong, and it was the only tooling they had.
Because here's the thing about that spreadsheet. It was never built for collaboration. Despite the share button sitting right there, a spreadsheet was built for one person to model something on their own, not for a team to run a business process. When you take a single-player tool and ask it to carry multi-player work, it breaks.
So read what's actually in front of you.
Every spreadsheet in your organization that runs business-critical work is a workflow your team already validated. Somebody built it because the process had to get done. It is proof the process exists, proof it has a user base, proof it matters enough that someone maintains it by hand, week after week, because the alternative is worse.
Most enterprise software conversations get this backwards. They start with "what should we build?" or, worse, "what do users want?" Both sound reasonable. Both produce mediocre answers, because people are bad at telling you what they want in the abstract.
You don't have to ask them. You can read it. Open the shared drive. Look at what's been touched in the last thirty days. The handful of spreadsheets a whole department keeps open at the same time are validated workflows with a proven user base. That's your roadmap. It's been sitting there the entire time.
What the spreadsheet was actually doing
Here's where this piece connects back to the last one.
That spreadsheet is two things at once. It's proof of a validated workflow. And it's the residue of an idea that had nowhere better to go. It's the relief valve. When building the real thing was impossible, the spreadsheet was the consolation prize, the place the idea went to feel handled instead of getting built. Both readings are true, and together they tell you what the spreadsheet's real job has been all along.
Its job was never computation. Its job was to absorb the energy of ideas that couldn't ship.
Which means the moment ideas can ship, that job is over.
Read more of my thoughts on LinkedIn
Connect with me hereWhy an AI-forward company stops making spreadsheets.
This is the part most people aren't ready to say out loud, so I'll say it.
In a company that has actually lowered the cost of making things real, you should stop creating new spreadsheets. Not manage them better. Not govern them more tightly. Stop making them.
A spreadsheet was always four tools wearing one coat.
- A place to enter data.
- An engine to compute on it.
- A surface to display it.
- A canvas to collaborate on it all.
The collaboration was always broken, we covered that. But let’s look harder at the first one. The data entry. Manually placing values into a grid of cells was never something anyone actually wanted to do. It had no value of its own. It was the toll you paid to reach the computation, which led to the dashboard that fed the decisions. Formulas needed cells to point at, so you put numbers in cells.
So once you can hand a machine your raw numbers and your half-formed thinking in whatever shape they fall out of your head, and have it do the structuring, the math, and the display, the act of typing values into a grid is purely overhead. You're doing the machine's job for it. You keep the output, a clean table, a chart, a model you can poke at. You just stop authoring it by hand. You describe, and your AI tool structures.
So the trade isn't "we're taking away your tools." The trade is that you stop doing the part of your day you never liked, and you get back the part you did. You dump the raw material, you talk over it, you watch it take shape in front of you and change it in real time. That is a more alive way to do a job than feeding a grid one cell at a time. It is, genuinely, a freeing thing, and people feel it the first time it happens to them.
One warning, because this is easy to get exactly wrong.
You don't retire the spreadsheet by mandate. You retire it by obsolescence. Lowering the cost of having ideas is the precondition for all of this.
Read what your company has been writing down for years. Build the thing it was always trying to tell you it needed. Train your employees to build themselves. Give them a safe place to do this. And let the spreadsheet go back to what it was for in the first place, a scratchpad you open, glance at, and close, instead of the load-bearing wall it was never supposed to become.
