I have watched several technology sectors go through a full cycle: the liberalisation of long-distance calling, the arrival of consumer broadband, the consolidation of operators, the shift of voice into software. In each case the same thing happened. A category grew very fast, attracted a great deal of capital, and then most of the businesses inside it disappeared — while a small number became almost impossible to remove.
The difference between the two groups was never the growth rate. It was almost always the same thing: how hard the business was to switch off.
Growth is a description; switching cost is a structure
When a market is expanding, everything in it grows. Revenue growth in a rising category tells you about the category, not about the company. It is only when growth stops that you learn which businesses were actually load-bearing.
The question I have learned to ask instead is simple to state and awkward to answer: if a customer decided tomorrow to stop using this, what would they have to do? If the answer is “change a setting,” the business has a customer list, not a franchise. If the answer involves a project plan, a budget, a migration window and a conversation with their own customers — that is a different kind of asset entirely.
This is why I keep coming back to infrastructure and to the layers immediately above it. A fibre route between two cities is not exciting. It is also not replaceable within eighteen months by a competitor with more capital, because the constraint is not capital. It is right of way, permits, civil works and time. The same logic applies to a number range that a company’s customers have been dialling for a decade, or to a provisioning system that three other systems have been built on top of.
The three questions
When we look at a business, three questions do most of the work.
What would it cost the customer to leave? Not in licence fees — in disruption. I want to hear a specific answer, ideally from a customer rather than from the founder.
What did this take to build, and could it be built again faster now? Some assets were expensive to create and are now cheap to replicate; that is the dangerous category, and it is where a lot of software sits. Others were expensive then and are more expensive now, because the ground has been built over. Physical networks are the obvious case, but deep integrations into a customer’s operating process behave the same way.
Who has to be convinced to buy it? A product bought by a department head is a product that a different department head can cancel. A system that the operations director depends on to run the week is not.
Where this gets you into trouble
The honest weakness of this approach is that it makes you slow to see genuinely new categories. Durability is a backward-looking property. Something that has never existed has no switching cost yet, and by this framework it looks like nothing. I have missed things this way and I expect to miss more.
I have made my peace with that, because our capital is structured for a specific job. We have no fund life and no exit date, which means our returns come from owning businesses for a long time rather than from selling them at the right moment. If you own something for fifteen years, the only thing that matters is whether it is still there in year fifteen. Growth in years two and three is pleasant. It is not the thesis.
What I would tell a founder
If you are building something in this space, the most useful thing you can do for your own valuation has nothing to do with your growth rate. It is to be able to describe, precisely and with evidence, what your customers would have to go through to replace you.
Most founders cannot do this. They describe their features. The ones who can describe the switching cost tend to be running the businesses that are still there a decade later — and they tend to be the ones we want to talk to.
Evgeny Vasilyev is Chief Executive Officer of Boxwood Group. He previously held senior roles at PeterStar, Petersburg Transit Telecom, MegaFon and MTT.