Being good at something buys you about a year. After that it’s been copied, and whatever keeps customers paying has to be something other than quality.
That something is a moat, the word Buffett used for any reason a competitor who understands your business perfectly still can’t take your customers at a profit. Most writing on the subject comes from investors sizing up companies with billions in revenue, so the answers are network effects, scale economics, regulatory capture, and brands built over fifty years. Every one of them is real, and not one is available to you if you run a software product with four people.
Below is the full list as I understand it, grouped by where the moat actually sits rather than by what it looks like from the outside. The last three sections are the ones I’d read first if I were short on time.
Moats That Live in the Customer
| Moat | What it is | Where you see it |
|---|---|---|
| Switching cost | Leaving costs money, downtime, or the risk of getting it wrong | Payroll, accounting, ERP, anything the finance team signs off on |
| Data gravity | The customer’s own history sits inside your system and won’t move cleanly | CRMs, photo libraries, analytics with five years in them |
| Workflow lock | Your product’s shape has become the customer’s process | Jira, Salesforce, design files treated as the source of truth |
| Habit | Used often enough that choosing it stopped being a decision | Keyboard shortcuts, Excel, the browser someone has had since 2011 |
| Trust | The buyer is risking something and picks whoever they’d least regret | Auditors, security vendors, enterprise procurement |
| Identity | Owning it says something the buyer wants said | Patagonia, mechanical watches, open-source purism |
| Relationship | There’s a human the customer can call, and they will | Local agencies, private banks, small consultancies |
| Institutional knowledge | Your staff know the customer’s setup better than the customer does | Long-running maintenance contracts |
Moats That Live in the Market
| Moat | What it is | Where you see it |
|---|---|---|
| Network effect | Each additional user makes the thing more valuable to the rest | Messaging, protocols, social graphs |
| Two-sided liquidity | Buyers show up because sellers are already there, and the reverse | eBay, job boards, freelance marketplaces |
| Distribution ownership | You control the surface where demand arrives | App stores, search, the WordPress.org repository, shelf space |
| Default placement | You’re preselected before the buyer has an opinion | Preinstalled apps, search deals, hosting bundles |
| Standard | Everyone else builds against your shape | HTTP, the spreadsheet file, WordPress hooks |
| Ecosystem | Other people’s businesses depend on yours continuing to exist | Shopify apps, the plugin economy, CUDA |
| Regulation | The rules make entry expensive on purpose | Banking, insurance, medical devices |
| Granted scarcity | The state issues a limited number and you hold one | Taxi medallions, spectrum, liquor licenses, water rights |
| Category ownership | Your name is what people call the category | Google, Hoover, Photoshop |
Moats That Live in the Economics
| Moat | What it is | Where you see it |
|---|---|---|
| Scale cost | Volume drops your unit cost below what a smaller rival can reach | Logistics networks, chip fabs, cloud infrastructure |
| Capital wall | Entry needs more money than a competitor can raise | Mines, satellites, semiconductor plants |
| Negative working capital | Customers pay you before you pay your suppliers | Annual prepaid software, restaurants, insurance float |
| Self-funded compounding | Profitable early, so you can outlast anyone who needs another round | Bootstrapped software, family businesses |
| Owned attention | An audience you already have, so acquisition costs almost nothing | Newsletters, channels, a mailing list built over a decade |
| Bundling | Given away alongside something the buyer already wanted | Office suites, Prime video, hosting control panels |
| Price floor | Cheap enough that competing isn’t worth anyone’s time | Free tiers, commodity tooling, loss leaders |
| Low burn | Your costs are small enough that a bad year is survivable | Solo operators, no office, no headcount |
Moats That Live in an Asset
| Moat | What it is | Where you see it |
|---|---|---|
| Proprietary data | Information nobody else can gather at that quality or depth | Financial terminals, transaction archives, industry benchmarks |
| Catalog | A back library that keeps earning without new work | Music rights, stock archives, evergreen articles |
| Brand | The name does work the product would otherwise have to do | Rolex, Michelin, Coca-Cola |
| Enforceable IP | Patents, trademarks, copyright, in a jurisdiction where courts act | Pharmaceuticals, semiconductors, publishing |
| Physical position | Land, location, or a right of way that can’t be duplicated | Ports, ski lifts, the corner shop |
| Supply access | Exclusive or preferential claim on an input everyone needs | Long-term contracts, mineral rights, allocation deals |
| Reputation record | A public, checkable history of results | Verified track records, review counts, case archives |
| Certification | An accreditation that took years and can be revoked | Medical licenses, security audits, notary appointments |
| Archive of failure | Documented knowledge of everything that doesn’t work | Old engineering firms, insurers, anyone with claims history |
Moats That Live in the Operator
| Moat | What it is | Where you see it |
|---|---|---|
| Taste | Knowing which version is the right one before the market says so | Product design, editorial, curation |
| Scar tissue | Years of watching one specific market fail in specific ways | Anyone who has shipped in the same niche for a decade |
| Iteration speed | You ship while the competitor is still scheduling the meeting | Small teams, single decision-makers |
| Focus | Willingness to serve a market too small for anyone larger to bother with | Niche software, trade-specific tools |
| Personal audience | People follow you, and the product is downstream of that | Independent writers, creators who ship products |
| Access | The people who take your call | Deal flow, hiring, partnerships |
| Unpleasant work | You’ll do the migrations, the compliance, the phone support | Enterprise onboarding, legacy conversions |
| Endurance | You’re still here after everyone else’s funding ran out | Any category after a hype cycle |
Moats That Are Really Just a Head Start
These get mistaken for moats constantly, and every one of them decays on its own if nothing else is built underneath it.
| Looks like a moat | Why it isn’t |
|---|---|
| Feature set | Features are a public specification of what to copy |
| Engineering effort | Difficulty of construction says nothing about difficulty of replacement |
| First-mover | Being early is an advantage only if you convert it into one of the moats above |
| Lower price | Anyone can match a price, and the one who can afford to lose money longer wins |
| A good idea | Ideas travel faster than the businesses built on them |
| Content volume | Volume was expensive when producing it was expensive |
| Design polish | Copyable in an afternoon, though taste that keeps producing it isn’t |
What AI Has Thinned Out
Anything whose defense was “this took a long time to build” is worth less than it was three years ago. In the plugin market I work in, feature sets that would have held a competitor off for a year or two are now a few weeks of work for someone who knows what they’re doing, and most of that time goes on the parts that were never the hard bit. Content moats built on publishing volume have gone the same way, since the marginal cost of the thing they were made of collapsed. Expertise sold by the hour is under the same pressure at the routine end.
What survived is everything that was never about capability in the first place. Distribution still has to be earned or bought. Trust still takes years and can’t be generated. Switching costs still sit in the customer’s own accumulated data. Regulatory position, physical assets, relationships, an audience, a catalog that already exists: none of those got easier to acquire, and several got harder as attention got more expensive.
The practical read is that the moat and the product have come apart. It used to be reasonable to assume that a good product would accumulate a moat as a side effect, because building it was hard enough to keep others out for a few years. That assumption has stopped paying, and the moat now has to be a deliberate line item, chosen and worked on separately from the thing you ship.
What You Can Build From Nothing
Most of the catalog above needs capital, scale, or a license from the state. Take those out and what’s left is short, and all of it is available to a small team with no funding.
Focus is the first one: a segment too small for anyone with a sales team to bother with, served properly. Then taste, which takes years to develop and can’t be hired in quickly. Speed, which isn’t a virtue so much as a structural property of having one decision-maker. An audience you own, built by publishing long before you need it, and the trust that comes from the same place. Switching cost, if you design your product so the customer’s data accumulates inside it. A position other people’s work depends on. And endurance, which nobody plans for and which decides more outcomes than the rest of the list put together, because staying solvent removes competitors at no cost to you.
Every one of them is slow, which is why they’re worth anything at all. A moat you could build in a quarter is one your competitors already have.
What I’d Bet On, and What I’m Not Sure About
I’ve spent most of my working life selling software inside an ecosystem I don’t own, which makes me a poor advertisement for ecosystem position as a moat. It works well right up until the platform’s own trajectory turns, and then you find out the moat was a lease.
What I’d put money on now is the dull end of the list. An audience built before there was anything to sell to it. A niche too small to be worth a competitor’s quarter. Enough margin to still be here in five years without needing anything to go right.
The one I keep turning over is switching cost, which has been the most dependable software moat there is. What made it work was that migrations are painful: the export is incomplete, the field names don’t match, and somebody has to sit there for three weeks reconciling it. That pain was labor, and labor just got cheap. If an agent can move a decade of records between two systems over a weekend, a lot of businesses are going to discover their retention was never loyalty.
I don’t know how fast that plays out, or whether trust ends up absorbing the difference. If you run a product whose customers stay mostly because leaving is a hassle, I’d be interested to hear whether you’re seeing any of it yet.
Related
Leave a Reply