Five simple tools, bundled, do not make a simple company.
The Four Debts, part 2 of 4.
What I never told the team, when I handed them a plane already in the air, was that the wiring was a mess of my own making.
I wrote last Friday about the letter where I gave them the stick. This is the part I left out.
Every company runs on software the way a plane runs on wiring and electronics. You never see it from the cabin. You only meet it when something stops responding.
This is the debt founders feel least, and it is dangerous for exactly that reason. Mine started with an assumption that felt obviously right. Buy the best tool for every job, a suite of best-in-class software, a quiver of perfect sails, and you will have the best way to run the business.
On the day you go live, you do. It works, and everyone is happy.
Then you sail some miles. Time passes, and the connections between all those perfect parts begin to fray. The cracks show. That is the first half of the mistake, a wrong assumption made at the very start.
The second half was more expensive. I kept the setup far too long, because I counted everything we had already sunk into it and underweighted what it was quietly costing us to stay. Sunk cost is loud. Opportunity cost is silent. For years I listened to the wrong one.
And I made it worse. I left the wiring in the hands of people for whom the complexity made perfect sense. Not for us. For them. A system that needs permanent maintenance is a permanent invoice. The Frankenstein kept growing, and I kept signing for it. My mistake.
Call it Technology Debt. It is the second of the four debts, and most people get it wrong before they start, because they assume it means old technology.
It rarely does. Ours was not old. Ours was excellent.
That is the trap.
Every tool we bought was genuinely good. The accounting package was good. The system we ran the business on was good. The file store, the chat tool, the video tool, each was best in its category and each solved the exact problem it was bought to solve.
Then you add them together, and something happens that no business case predicts.
Five simple tools, bundled, do not make a simple company. They make a permanently complex one. Each arrives with its own release cycle, its own opinion about how a process should run, its own way of naming the same customer. The complexity does not sit inside any of them. It sits in the seams between them, where nobody owns it and nobody is measured on it.
For a stretch of our growth, two of our systems told two different stories about the same month.
The accounting package counted invoices. It did not carry stock movement. The system we actually ran the business on carried stock, but only what had physically arrived, and it saw customer prepayments the accounting package did not.
Both were telling the truth. Neither was telling the whole truth. And the gap between them was not a rounding error. It was a different picture of the same weeks.
If you sail, you know the moment when a boat is working hard and you cannot yet tell whether it is about to lift and plane, or whether it is just pushing water and throwing a big wake. Same noise. Same effort. Completely different outcome.
That was us for longer than I would like. Plenty of speed, plenty of spray, and no quick way to answer the only question that matters: are we actually on course, or twenty degrees off?
We fixed it. I will come back to how. But be clear about what the cost actually was, because nothing ever crashed. The cost was decision latency. Every question worth asking had to be reconciled before it could be answered, and by the time you have finished reconciling, the moment you were deciding about has moved on without you.
Then there is the half you do not control.
We learned to treat upgrades properly. Test before you touch anything, because you are rewiring an aircraft that is already flying with people on board. So we planned, we tested, we scheduled the work for the quiet weeks.
At least half the time, it made no difference.
One of the tools in our carefully chosen suite would announce a fundamental change of its own. A new data model. A rebuilt way of connecting to everything else. Live on a date we did not choose. And because that one tool touched all the others, everything else had to move with it.
The clearest case was our accounting package rewriting how it handled data and how other systems talked to it. It was wired into everything we did. That single upgrade landed like a shot of entropy straight into the veins of the company.
Here is the line I would put on the wall. When you run your company on a stack of separate tools, you do not own your own upgrade calendar. You have rented it out to every vendor in that stack, and none of them coordinate with each other, or with you. The bill never arrives as an invoice. It arrives as a weekend.
You can feel this one before you can prove it.
Look for the human patch. Somewhere there is a person quietly re-keying the same data from one system into another, or rebuilding the same report by hand every week. That person is not inefficient. That person is the integration layer you never bought.
Ask who owns the source of truth. If three people give you three answers, you already have this debt.
Count how many of your last upgrades you actually chose. If it is fewer than half, your timing belongs to somebody else.
And time how long it takes to answer a simple question about your own business. Not how long to find the number. How long to trust it.
The way out is not a better tool.
This is where most advice fails. The answer to technology debt is not a better tool. Adding a better tool to a stack of good tools is precisely how you got here.
The answer is to start from the experience you want to deliver and work backwards to the technology, instead of buying technology and then bending the experience to fit whatever it happens to do. Most software is a solution looking for a problem, and it is very good at finding one inside your company.
So we stopped buying pieces and stitching them together. We moved onto a single system, one flexible and modular enough to bend around the way we actually wanted to serve partners, and we made everything live inside it. Not something we built ourselves, and not the newest or the cleverest tool on the market. One backbone we could shape to our own process, instead of five excellent tools that were never designed to live together.
That is the move. Decide what the experience is. Name one source of truth. Stop admitting anything that cannot feed it.
A warning from the far side, though. Once entropy is in, you do not fully undo it. You can rebuild, and we did, but you cannot put the universe back into a box. Every tool is easier to let in than to remove, which is why the discipline belongs at the front door, not in the cleanup.
Simplicity does not stand a chance against complexity when software is allowed in the fight. Unless somebody in the building is deliberately fighting for simplicity.
At first that somebody has to be you. Then, if you are wise, you stop leaving it to whoever happens to have time, and you build a team who owns your data and your systems, and you make simplicity the thing they are measured on. We did that years ago. They are still winning that fight, quietly, every week. That, in the end, was the real fix. Not a newer tool. A team who guards the front door and never stops.
That is the second debt. Not old technology. Too many good things, never designed to live together, quietly deciding your timing for you.
Next Friday: Decision and Speed Debt. The most expensive of the four, and the one I paid for in 2007.
w. 🌊
Make My Mess My Message is a book written in public, one Friday at a time. This issue first appeared on LinkedIn. Every issue lives here.
Leave a Reply