Real competitive advantage: what a moat is, which ones died in 2026, and which you can still build with no money

A moat is anything that makes copying you cost someone more than they gain by copying you. That's the whole definition, and notice what it does not say: it doesn't say you're better, that your product is more liked, or that you work longer hours. It says the arithmetic doesn't work for whoever wants your spot.

Being better and having a moat are different things, and confusing them is the expensive mistake. Being better is a position: you hold it today and it can be taken tomorrow, because any advantage bought with effort and money is available to anyone with effort and money. A moat is a structure: it exists even if your competitor knows exactly what you do, writes it down, and gets to work. The test is uncomfortable and useful: if tomorrow you published a complete manual of your business — prices, suppliers, processes, sales script — how many people could hurt you with it? If the answer is "plenty", what you have is an advantage, not a moat.

Because it stopped being something you own and became something you rent. For decades, having better technology meant having built it: years, a team, sunk money. Today the same model a thousand-person company uses is billed to you by usage, at the same public rate, with nobody asking for permission or references.

The numbers say it better than any argument: the top tier converged to nearly identical pricing across providers, and the cheap tier collapsed to cents per million tokens, with cuts of 50% and 80% within months. When the input that supposedly gives you your advantage has a list price, your advantage has one too. This doesn't mean technology doesn't matter: it means using it is the floor, not the ceiling. It's the difference between having electricity in 1900 — when it was an advantage — and having it now, when only the absence of it gets noticed.

Less than you'd think, for the same reason. Scale protected you when being big gave you costs the small player couldn't match: amortised factories, owned servers, purchasing power. Much of that is now rented by the hour. A three-person business can run on the same infrastructure as a three-thousand-person one and pay only for what it consumes, so the size advantage has lost the part that made it unreachable.

What still protects is the scale that produces something money can't buy quickly: proprietary data that only accumulates by operating, a network of users who make each other more valuable, a reputation with years behind it. There, size isn't the moat; it's the by-product of having been there earlier. The useful question isn't "am I big?" but "does my size give me something a funded competitor can't buy this week?". If the answer is no, your scale is an expense, not a defence.

The ones paid for in a currency that isn't money held. There are four, and it's worth separating them because they aren't built the same way.

**Distribution**: direct, repeated access to the people who buy — a list, a community, a channel, a shelf. Building got cheap; being found did not. That's why distribution is today the most underrated advantage among people who build and the most fought-over among people who sell.

**Switching cost**: what it costs your customer to leave, measured in work and risk, not in money. Their data inside your system, their team trained in your way of doing things, their processes wired into yours. Don't confuse it with locking them in by contract: that produces resentment, not a moat.

**Network effect**: each new user improves the product for everyone already there. It's the strongest moat and the rarest; almost no small business has one, and believing you do when you don't leads you to grow at a loss waiting for a tipping point that never comes.

**Brand**: not your logo, but someone choosing you without comparing. It's paid for in years of doing what you said you'd do, and it's one of the few assets AI hasn't made cheap, because it isn't generated: it accumulates.

Three questions, in this order, and all three have to give the same answer.

**One: can it be bought?** If a funded competitor can have it next month by writing a cheque — a tool, a campaign, one of your employees — it isn't a moat. It's matchable spending.

**Two: does it improve on its own over time?** A real moat gets deeper while you operate: more customers leave more data, more years build more trust, more integrations make leaving more expensive. If your advantage stays flat no matter what you do, it's a position you're defending, not a moat defending you.

**Three: does it survive being copied in plain sight?** Publish that manual again, mentally. A good moat survives total transparency: you can explain how you do it because the other party's problem isn't knowing, it's the time, the relationship or the position you already have and they don't.

If your advantage fails the first, stop there. If it passes all three, you have something to defend — and the work becomes not breaking it yourself.

The three paid for in time and attention, which is the only thing you have in surplus when you have no capital.

**The first is an owned audience**: people who read or listen to you on their own, whom you can reach without paying a toll. It's built by publishing for months before you have anything to sell — which is exactly why almost nobody does it, and exactly why it works.

**The second is a niche so narrow it bores the big players**: a small, specific, unglamorous problem where the market doesn't justify a large company's attention but does support a very good living. The advantage isn't that they can't do it; it's that it isn't worth their while.

**The third is knowledge you only get by operating**: the details written down nowhere, that show up only after a hundred conversations with customers in that sector. It's slow by definition, and that's why it can't be bought. All three share a shape: they aren't compressed money, they're compressed time. That's why they're within reach of someone starting out, and why there's no way to speed them up.

A moat almost never breaks because someone jumps it: it breaks because the ground moves and it stops protecting anything. The technology that made your speciality expensive changes, the channel people arrived through changes, the law changes, or what customers want from that category changes — and suddenly the wall that cost so much to raise encloses a place nobody walks past any more.

The two warnings that nearly always arrive first: one, your advantage starts being explained by what you buy rather than by what you've accumulated; two, newcomers appear doing something similar in a way that strikes you as obviously worse, and doesn't strike customers that way. That "obviously worse" is the most reliable signal there is that the dimension you were competing on stopped being the deciding one. So a moat has to be audited periodically with the three questions above, and it pays to always have something small growing outside it: the best time to start the next one is while the current one still works.