Switching Costs — The Price of Leaving Is the One Nobody Quotes
That second number exists anyway. It is paid in an afternoon of moving records, in settings rebuilt from memory, in the small failures that follow a change nobody planned carefully enough. It does not appear on any statement, which is exactly why it does so much work.
The price of leaving is the quiet term in most long relationships between a buyer and a seller — and it explains more about who keeps whose business than loyalty, satisfaction, or the contract ever will.
What Actually Holds a Customer
Ask why a customer stays and the usual answers are about the product. It is better, or cheaper, or they are used to it.
Sometimes. But a product being good explains why someone arrived. It rarely explains why they are still there five years later, paying more than a newer option would charge, aware that better exists.
What holds them is the accumulated cost of undoing the arrangement:
Records and history. Years of data sitting in one place, in one format, useful only there.
Configuration. Settings, permissions, rules and exceptions built up one decision at a time, most of them undocumented and some of them load-bearing.
Connections. Everything else that was wired into it — other services, other processes, other people's habits.
Learned behaviour. The accumulated knowledge of how this particular thing behaves, including its faults. Competence is not transferable; it has to be rebuilt.
Risk during the change. The window where the old arrangement is being dismantled and the new one is not yet trustworthy. Nothing about the change is free while it is happening.
None of these is the product. All of them are things that grew on top of it.
Notice what that implies about timing. A customer evaluating an option before committing is comparing two products, and the comparison is roughly fair. A customer evaluating the same two options three years later is comparing one product against another product plus the cost of dismantling three years of accumulation. The second comparison is not a harder version of the first. It is a different question, and the incumbent is answering a much easier one.
Why the Number Stays Hidden
A joining price is quoted because it has to be. Somebody is asking for money, so a figure gets stated.
A leaving price is never quoted because nobody is collecting it. It is paid in hours, attention and disruption, distributed across a period long enough that it never presents itself as a single bill. And the parts a person could estimate — the afternoon of paperwork, the fee for closing early — are usually the smallest parts. The expensive part is everything that turns out to have been depending on the arrangement quietly.
Which produces a predictable error in both directions. Before switching, the cost is underestimated, because the invisible components are invisible. During switching, it is discovered. Afterward, it is remembered — often too vividly, which is its own distortion, and part of why people stay put longer after a bad move than the arithmetic justifies.
This is the same correction as the one about what a thing is called versus what actually sets its terms. The contract is the name. The switching cost is the term.
The Real Lock Is Never the Product
Here is where it gets structurally interesting, and it applies well beyond any one industry.
When a supplier wants a relationship to last, competing on product quality is the weakest available strategy, because product advantages are temporary and visible. Somebody else will match it, and the customer will notice.
The durable approach is to become the layer that everything else is built on top of — the place where the records live, where the permissions are defined, where the connections terminate. Not the thing being used, but the ground it stands on.
The advantage of that position is that it survives losing. If the ground is owned, whichever product wins on top of it still has to sit there. Being replaceable at the visible layer stops mattering very much when the layer underneath is not the one being shopped.
This is a different claim from profit collecting wherever the work is currently difficult. Difficulty is about capability, and capability erodes as things standardise. Entanglement is about accumulation, and accumulation does the opposite — it grows with every ordinary day the arrangement continues.
Which leads to the uncomfortable half: the strength of the hold is a measure of how expensive leaving became, not of how good the thing is. Those two get described with the same word — loyalty — and they are not the same thing at all. One is chosen. The other is the difference between being used and being preferred, which is a distinction any relationship built on accumulated cost has to keep answering.
The Household Version
This is not only a story about large organisations, and the household version is the one that pays for itself immediately.
Think about the account, subscription or provider you have been meaning to change for more than a year.
There is a reasonable chance it is not the best available option, that you know this, and that the gap has been known long enough to stop being news. The reason it persists is almost never that the alternative is unclear. It is that leaving has a price, the price is unquoted, and an unquoted price is very easy to keep postponing.
The exercise that helps is boring and quick: write down what leaving would actually involve. Not whether to leave — just the list. Records to move, things to reconnect, the disruption window, the fee if there is one.
Two things usually happen. Some items turn out to be smaller than the dread suggested, because dread does not itemise. And the genuinely large ones become visible as what they are — a real cost, worth weighing against a real saving, rather than a vague heaviness that ends the thought before it starts.
The list also separates two things that get treated as one. Knowing a better option exists is not the same as being able to move to it — finding a substitute and switching to one are different problems, and the second is where the expense lives. Most people who have not switched are not confused about the alternative. They have run into the second problem and filed it as indecision.
An unmeasured cost cannot be compared to anything. That is its whole power.
Costs That Grow While Nobody Is Looking
One more property, and it is the one that decides outcomes over long periods.
Switching costs compound. Every month of ordinary use adds records, settings, connections and habits. Nobody decides to increase them; they increase as a byproduct of the arrangement working normally.
So the cheapest moment to leave any arrangement is always the earliest one, and the moment a person is most likely to consider leaving is much later — after the accumulation has done its work, when dissatisfaction has finally grown larger than inertia.
The practical consequence is that the decision to stay is being made continuously, by default, without ever being framed as a decision. A choice nobody makes explicitly still gets made. It just gets made in favour of whatever is already in place, the way a stated intention slowly becomes a settled fact without anything being signed.
There is a policy footnote worth knowing here, because it shows the effect from the outside. In several markets, including Korea, regulators have deliberately attacked switching costs rather than prices — making it procedurally simple to move a phone number between carriers, or to redirect the automatic payments attached to a bank account. The stated goal was competition, and the mechanism was not lowering anyone's price. It was lowering the cost of leaving, and letting the prices follow.
That is the clearest available evidence that the leaving cost, not the joining price, was the thing holding the arrangement together.
Lines Worth Watching
Six ways to read the strength of a hold rather than the quality of a product:
- What would have to be rebuilt — records, settings, connections. The longer the list, the less the product is doing the work.
- How long the relationship has run — entanglement is a function of time, not satisfaction.
- Whether the difficult part is the visible part — if the hard bit of leaving is invisible, the hold is stronger than it appears.
- Whether anyone can state the leaving cost — an arrangement nobody has priced is one nobody has genuinely re-decided.
- Renewal behaviour without renewal effort — continuation that requires no persuasion is continuation held by cost, not preference.
- What happens when leaving gets easier — the moment a procedure gets simplified is the moment you find out what was preference and what was friction.
A Closing Note
Every arrangement has two prices. One is quoted at the start, discussed, compared, and negotiated. The other is never quoted, grows quietly, and eventually decides the outcome.
The unquoted one deserves the attention precisely because it is unquoted. A cost that is never stated is never compared, and a cost that is never compared is paid indefinitely, by default, by people who would have chosen otherwise if anyone had written the number down.
Nobody will write it down for you. The party best placed to calculate it has no reason to.
Data as of 2026-08-24. Sources: general descriptions of switching costs in competition and consumer-protection literature, and publicly documented regulatory measures in several markets — including mobile number portability and automatic-payment redirection for bank accounts — that were introduced to reduce the procedural cost of changing providers. This post describes a structural relationship in general terms. It refers to no specific company, service, security or transaction, and makes no forecast. This post is observation, not investment advice.
Related Posts:
Customer Concentration — When a Company's Future Sits in Someone Else's Filing
Finding a Substitute Is Not the Same as Switching to One
A Layer Only Earns While It Is Hard
Where Your Gold Sits Is Part of What It Is
A Backlog Is a Promise, Not a Payment
Visuals on this post are AI-generated. The author works with AI as a research and drafting assistant; topics, judgments, and final edits are the author's own. This post is observation, not investment advice. See full Disclaimer for details.
