A moat is whatever makes a competitive advantage expensive to attack, and it is what makes a lead different from a position. An advantage says that the organisation can do something today that its rivals cannot do. A moat says how long that will still be true once those rivals decide to try. Warren Buffett made the term popular with investors in the 1990s, and Hamilton Helmer gave it a working taxonomy for operators in his 2016 book on the seven powers. A moat is built on the roadmap, or it is quietly skipped there, so the product manager is the person who decides. Most roadmaps are full of work that widens a lead. Nothing on them deepens a moat.
The sections below set out what a moat does. They then give the five kinds that exist and what erodes each one. Switching costs get a section of their own, since a product team builds those most directly. The page ends with the argument that the durable moat is changing shape.
What a moat actually does
A moat buys time, and the attacker is the one who pays for that time. It does not make an attack impossible. What it does is make the attack cost more than the attacker expects to get back. That is a sum any competitor can work out again, and because circumstances change, a competitor who decided against an attack two years ago can reach the opposite answer this year.
Putting a length of time on a moat keeps it honest. If a team writes down eighteen months of protection, ending when a competitor ships a migration tool, it has made a claim that anybody can watch and act on. If the same team calls the moat deep or strong, there is nothing to watch. The difference shows up in a review meeting. A moat measured in months starts an argument about what to build next quarter, and a moat called strong draws nods and nothing else.
The five kinds and what erodes each
| Moat | How it works | What erodes it |
|---|---|---|
| Network effects | Each user makes the product more valuable to the next one | Users joining a second network too, a rival serving one side better |
| Switching costs | Leaving costs the customer money, time or risk | A competitor funding the migration, a regulator mandating portability |
| Scale economies | Unit cost falls as volume rises, so price can fall with it | A rival reaching efficient scale, a technology shift resetting the curve |
| Proprietary data | Accumulated use improves the product for everyone on it | The data becoming common, a model on public data reaching parity |
| Brand in a category | Buyers pay more for the name they already trust | One public failure, or the generation that trusted it moving on |
Helmer adds a sixth kind worth knowing by name. Counter positioning is a position an incumbent cannot copy without damaging a business it already has, such as a free tier that would cannibalise its own licence revenue. That protection is rented and never owned. It lasts for as long as the incumbent judges the damage too high to accept. It ends on the day somebody there does the sum again and decides that losing the segment outright would cost more than the licence revenue the refusal protects.
Switching costs in detail
Switching costs are the one moat a product team can build on purpose. They come out of ordinary roadmap decisions, so it is worth knowing the three forms they take.
Accumulated state. Years of history, configuration, integrations and saved work that live inside the product. This is the form customers defend, because the thing they would lose is theirs and they know how long it took to build.
Learned process. The team knows how to do the work in this tool, has written its runbooks around it and trained its new starters on it. Retraining forty people is a real cost, and it rarely appears in a competitor's business case.
Contract and integration. Notice periods, committed terms and the six other systems now wired into this one. This form is the most brittle of the three. A motivated buyer treats it as a one off expense and pays it.
How a customer reacts to a switching cost depends on who created it. If a vendor makes a customer's own data hard to export, the vendor has imposed the cost. Customers resent a cost like that. It invites a portability regulation, and in the end it brings in a competitor whose whole pitch is the migration. Accumulated state is the opposite case, because the customer built that cost themselves. The customer is then the one explaining to a challenger why leaving would be painful.
The moat that is changing shape
The generative tooling that arrived through the 2020s made a competent feature much cheaper to build. Because building got cheap, an advantage that rests on a feature now lasts a shorter time than it used to. A feature that once bought a year now buys a quarter. The argument now common in product strategy writing is that protection has moved towards the two things that stayed expensive.
The first of those two is distribution. Building got cheaper and attention did not get cheaper with it. An organisation that already has a route to a segment can copy a rival's feature and put it in front of more buyers by the end of the week.
The second is the speed at which an organisation notices something and responds to it. If features are cheap to build, the advantage moves to the way the organisation decides which feature to build next. That way of deciding is part of how the place is run, so it resists copying for the same reason a culture does.
Most products have no moat
For most products the honest finding is that none of this applies yet, and writing that down is more useful than claiming a moat the product does not have. A strategy that names no moat and commits to building one is a plan. A strategy that claims a moat and cannot put a length of time against it is a document that will be quietly abandoned.
If the honest answer is none, two starting points are still within reach. The first is accumulated state, and it comes from being the place the work happens, so that a year of a customer's history and settings ends up sitting inside the product. The second is distribution, which comes from owning a route to the segment before a competitor does.
The choice of where to compete is now settled, along with whether a win there could last. The second half of a strategy answers a different question. It asks what this product will actually offer those buyers and how that offer turns into a business.