Switching costs are everything a customer gives up to move to an alternative, over and above the price of the alternative itself.

They are the reason a customer stays with something they would not choose again today, and they are most of what makes revenue predictable.

The four kinds

Financial. Contract termination fees, unused prepayment, the cost of the new thing. The easiest to see and usually the smallest.

Procedural. Migration, setup, retraining. The work of moving. For anything that has been in place for years this is large and badly estimated, which is itself part of why people do not move.

Relational. A supplier who knows your business does not have to be told. Starting again means explaining everything to somebody new, and the value of not having to explain is invisible until it is gone.

Risk. The current thing works. The new one might not. A known mediocre outcome beats an unknown distribution for almost everyone, particularly for whoever would be blamed.

The last two are usually the largest and are the ones nobody puts in a business case.

Why this determines what a business is worth

A buyer of a business is buying future revenue, and they discount it by how likely it is to leave.

Revenue with high switching costs is worth a multiple of the same revenue without them. That is most of the gap between a project business and a retainer business. Recurring revenue is revenue the customer has to take an action to stop, so inertia works for you, and predictability is a consequence of that rather than the whole of it.

The same logic underlies client concentration being scored the way it is. One large client with low switching costs is a much worse asset than several smaller ones who cannot easily move.

Earned and imposed

The distinction that matters, and the one most often ignored.

Imposed switching costs are barriers to leaving: long notice periods, data held in a format nobody else reads, contracts that auto-renew, processes only you can operate. They work, in the sense that they retain customers. They also produce a customer who is looking for an exit, and who will take it the moment a plausible one appears — usually at the worst moment, usually loudly, and often taking a reference with them.

Earned switching costs are accumulated value that would be lost by leaving: history, tailored configuration, a supplier who knows the business, results that compound with tenure. The customer could go and would be giving something up, and they know what.

Both look identical in a retention number. They behave completely differently under stress, and the difference only shows up when a competitor calls.

Building the earned kind

Accumulate something for the customer. History, data, tuning, institutional knowledge. Anything where the version of your service they get in year three is better than the one in year one specifically because it is year three.

Learn their business deliberately and show it. The relational cost is real and it has to be visible to count. A supplier who reads a situation without being briefed is demonstrating the switching cost every time it happens.

Make onboarding excellent, and do not make it a weapon. A hard setup is a switching cost in both directions — it keeps your customers and it keeps everybody else's. Competing against a rival with a painful migration is easier than you think, because the customer is already unhappy.

Never rely on the data lock. Holding somebody's data hostage is the purest imposed cost and it is the one that generates the worst story about you. It is also increasingly illegal.