The Real Cost of Free Tools
I once spent a Saturday and most of a Sunday pulling three years of data out of a free tool that had decided free was going to mean something narrower going forward.
The export was a mess. Half the fields came out as a single unlabeled column. I rebuilt what I could and lost the rest.
At the end of it I did the math on what that weekend was worth against what the paid plan would have cost me over those three years. The paid plan lost by a wide margin, and I still made the wrong choice, because the paid plan asked for the money up front and the free plan waited until I was too far in to leave.
That is the whole thing, really. Free is a price. It is not a cost.
The invoice arrives in hours
The most common way a free tool bills you is time, in small pieces, so you never see the total.
It does eighty percent of what you need. The remaining twenty percent you handle by hand. Copy the numbers into a spreadsheet, rename the files, click through the thing that should have been one button.
Twenty minutes a week feels like nothing. It is seventeen hours a year. If you value your working hour at anything above ten dollars, you just paid more than almost any subscription would have cost you, and you paid it in the worst possible currency, which is Wednesday afternoons.
The reason this hides so well is that the hours never appear as a line item. There is no moment where you are asked to approve them. They dissolve into the ordinary friction of the week and you file them under “work” instead of under “what this tool costs.”
I now try to write the workaround down when I invent it. Not to fix it, just to see it. A list of eleven small manual steps you perform every week is a very persuasive document.
The exit tax
The second bill comes when you leave, and free tools are reliably worst at the part where you leave.
This is not a conspiracy, mostly. Building a good export is real engineering work that benefits the people who are on their way out the door. On a free tier, that work has the weakest business case in the entire product.
So the export is an afterthought, or a CSV with the interesting parts flattened, or a paid feature. And the cost of leaving quietly grows in the background the whole time you are happily using the thing.
My one hard rule now, and it is the closest thing I have to a policy: before I put anything real into a tool, I run the export. Not read about the export. Run it, open the file, see whether what comes out is usable without a rebuild.
If the answer is no, it is not a free tool. It is a mortgage with a nice onboarding flow.
You are not the customer
When you do not pay, you have no standing, and that is not a moral complaint, it is just how the incentives point.
You can file a bug and nobody has to answer. The limits can move and nobody has to warn you. The thing can shut down entirely on ninety days notice and you have no recourse, because your side of the arrangement was nothing.
Paying twelve dollars a month does not make a company love you. But it does change what you are to them. A paying user churning is a number somebody has to explain in a meeting. A free user leaving is a rounding error and sometimes a cost saving.
For anything a real workflow depends on, I want to be a rounding error in the other direction.
When free is exactly right
I want to be clear that I am not arguing for paying for everything. I would look ridiculous, since most of what I rely on daily is free or open source, and some of it is better than the paid alternatives.
Free is genuinely the right answer in four cases.
When you are testing. Obviously. Never pay to find out if you will use something.
When it is a commodity and switching costs nothing. If moving to a competitor is an afternoon and a copy-paste, the lock-in risk is not real and free is just free.
When it is open source and you can host it yourself. Then the escape hatch is structural rather than promised, which is a completely different kind of free.
When you are genuinely small. A one-person operation with eleven customers does not need the paid tier of anything. Outgrow the free plan first, then pay. Paying early is its own flavor of buying tools instead of doing work, which I get into in the piece about overcomplicating a first business.
The rule I actually use
Two questions, and they take about a minute.
Does it save me an hour a month? If yes, I will pay up to twenty dollars a month without thinking about it. That is two hundred and forty dollars a year for twelve hours back. I do not know where else I can buy an hour for twenty bucks.
Can I get my data out, today, in a form I could actually use? If no, then the price does not matter, and neither does the feature list. I am not putting anything I would miss into a room with no door.
Everything else is preference. Those two are the ones that have actually cost me weekends.
Free tools are not free. They are financed. You pay in hours, later, at a rate nobody quoted you.
