Comparison tables are honest about the wrong number.
They line up monthly prices, feature checkmarks, and storage limits, and they let you conclude that the difference between two products is eight dollars a month. The eight dollars is real. It is also, for anything you actually depend on, close to irrelevant.
The cost that matters is the one no table contains: what it takes to get from here to there.
Where the money actually goes
Migration cost is made of hours, and hours are easy to underestimate because they arrive in pieces.
There is the mechanical work — exporting, importing, fixing what the importer mangled. That part is visible and usually the smallest.
Then there is relearning. Not the interface, which takes an afternoon, but the hundred small accommodations you had built without noticing: the shortcut for the thing you do daily, the naming convention that made search work, the knowledge of which feature is unreliable and should be avoided. That knowledge does not transfer. It has to be rebuilt through use, and during the rebuild you are slower at your actual job.
Then there is the rebuild of everything downstream. The automations that pointed at the old thing. The shared links other people have. The exports someone else depends on. The document that explains the process to a colleague. Each of these is small. Collectively they are usually larger than the mechanical migration.
Set against that, eight dollars a month is not a rounding error — it is noise.
Lock-in rarely feels like a trap
This is the part worth sitting with, because it explains why so many people stay with tools they no longer like.
Lock-in is usually imagined as something coercive: proprietary formats, hostile contracts, data held hostage. Sometimes it is. But the ordinary version is much softer and far more effective.
The ordinary version feels like not worth the afternoon.
Nobody decides to stay. They decide, repeatedly, to deal with it later — and later is always a week when there is something more pressing. The switch never gets rejected on the merits. It gets postponed until it stops being considered.
This is why the strongest form of lock-in is not technical at all. It is the accumulated weight of small accommodations, each of which was reasonable, and none of which is worth a Saturday to unwind on its own.
Three layers, unequal difficulty
It helps to separate what actually has to move.
Data is the layer everyone checks, and usually the easiest. Most mature products can export something. The question is not whether an export exists but whether it is complete — comments, version history, attachments, metadata, relationships between items. Exports are frequently partial in ways that only become apparent after you have committed.
Workflow is harder. This is the shape your work has taken to fit the tool: folder structures, tagging systems, the sequence of steps that became routine. Some of it transfers, some has to be redesigned, and it is rarely obvious in advance which is which.
Knowledge is hardest and least discussed. This is what you know about how the tool behaves — where it is slow, what it does badly, which feature to avoid on a deadline. It represents real time invested, it is invisible on any comparison table, and it is lost entirely.
A tool with excellent data export can still be expensive to leave, because two of the three layers do not export at all.
Export is the feature to check first
Which suggests an inversion of the normal evaluation order.
Most people check export last, typically on the day they have already decided to go — which is the worst possible time, because the answer no longer changes the decision, only the pain.
Checking it first costs nothing and tells you a surprising amount. Not primarily about the file format, but about the company’s posture. A product confident in its own value tends to make leaving straightforward, because it does not need to rely on friction. A product that treats export as an afterthought has usually made a decision about what keeps customers, and it is worth knowing that decision before you become one.
Concretely, before committing to anything you expect to depend on:
- Export your data on day one, while there is almost nothing in it. See what the file actually contains. - Check whether the format is open or merely a file — a proprietary blob with a .json extension is not portability. - Find out whether attachments, history, and comments come along, or only the top-level records. - See whether export is available on your plan, or gated behind a tier you are not on.
Ambiguity in any of these is itself an answer.
The reasonable counterargument
There is a version of this thinking that becomes its own trap.
Optimizing purely for exit produces a particular kind of paralysis: choosing tools for how easily you could abandon them rather than how well they work, and ending up with a stack of portable mediocrity. Depth has value. Some of the best tools are deeply opinionated, and being opinionated is exactly what makes them hard to leave.
The point is not to avoid depth. It is to enter it knowingly. Committing to a tool that will be expensive to leave is a legitimate decision when the tool is genuinely better. It becomes a problem only when the cost was never counted — when you find out what leaving costs at the moment you most want to leave.
What this changes
Two practical adjustments.
Compare total cost of leaving, not monthly price. For anything load-bearing, the switching cost will exceed several years of the price difference you were comparing. That does not mean stay put. It means the comparison you were running was measuring the wrong thing.
Check the exit while you are still enthusiastic. Evaluate export during the trial, when you have no emotional stake in the answer and the data is small enough to inspect. This is the same reason a trial should happen during a busy week rather than a calm one — the informative test is the inconvenient one, which is covered in The tools that survive your workflow.
Dependency also has a pricing dimension. What a company is paid to care about shapes how easy it is to leave, and that mechanism is worth understanding directly: Subscription is a design decision.
The short version
The price you compare is rarely the price you pay. The real number is the hours between here and there, and most of those hours are invisible on the way in and obvious on the way out.
Check the exit first. It is free, it takes ten minutes, and it is the only part of the evaluation that gets more expensive the longer you wait.