Skimming
Просеивание
Price Holds Depth 4 · Mine
very many very small transactions
Moats it grows into
- #24 Efficient scale
At a take this thin the flow supports one filter, perhaps two; a newcomer paying full fixed cost for a slice of the volume never reaches the point where its arithmetic closes.
- #15 Economies of scale
Each added transaction costs the built network almost nothing, so unit cost falls with width and the widest filter can price below what a narrower one can afford.
Summary
A take so small per transaction that nobody argues about it, multiplied across a flow so wide that the sum is large. The apparatus that filters the flow — the rails, the network, the baleen — costs a great deal to run and almost nothing per unit, so whoever already has the width covers the cost and a second filter in the same water starves.
How it works
The edge comes from an asymmetry between the fixed cost of the filter and the size of each catch. Building the apparatus — a payment network, an ad auction, a settlement layer — costs years and capital. Running one more transaction through it costs nearly nothing. So the take per unit can be set below the level anyone bothers to negotiate: a few basis points, a fraction of a cent, a small percentage the merchant never sees as a line item. Nobody fights over it, and the flow stays.
Width is what makes the arithmetic work. A filter that touches a small share of the flow does not cover its fixed cost at that take, which is why a rival cannot enter by copying the price. It has to copy the width first, and width is the thing that took years. The strategy is quiet by design: no single customer pays enough to switch, and the ones who might are the only ones who ever see the sum.
Why it holds — or only takes
Skimming holds because the price cannot be copied without the width behind it. Economies of scale is the direct form: the cost of one more transaction through a built network is close to zero, so the widest filter has the lowest unit cost and can set a take nobody else survives on. Efficient scale is the stronger form: at that take the flow feeds one apparatus, maybe two, and a third built at full cost never reaches the volume where its arithmetic closes. Both moats are in the fixed cost, which is why the strategy belongs to whoever paid it first and kept the flow.
Do you have it
- Your take is quoted in basis points or fractions of a cent, and no customer has ever negotiated it.
- Unit economics closed only after you crossed a volume threshold, and you crossed it years ago.
- The fee never appears as a line the end customer sees; they meet it only as a slightly smaller number.
- Revenue tracks the width of the flow, not the number of deals your sales team closed.
- The only accounts that ever ask about the sum are the largest ones, and even they stay.
How to dig it
- Attach to a flow that already exists — payments, searches, trades — instead of asking anyone to start a new one.
- Price the take below the cost of thinking about it, so switching to save it is never worth a meeting.
- Automate the whole path from intake to settlement; one manual step per transaction kills the arithmetic.
- Go for width before margin: endpoints, merchants, integrations, every place the flow can pass through you.
- Make yourself the default on both sides of the flow, so leaving means both sides have to move at once.
How it erodes
- A regulator caps the take, as interchange caps did for cards, and the sum shrinks while the apparatus costs the same.
- A cheaper rail appears — instant bank transfers, a state-run payment scheme — and the flow leaves for it without anyone deciding.
- The largest accounts grow big enough to build their own filter, and take the widest part of the flow with them.
- Cheaper apparatus lowers the fixed cost, more filters fit in the same water, and the take is bid down toward zero.
- The flow itself moves to a channel you are not on, and width you built somewhere else counts for nothing.
Solo, without capital
A solo builder can put a thin take on a flow this quarter — a per-call price on an API, a cut of a niche marketplace, an affiliate slice — and it runs without staff. What is missing is the width. Economies of scale is closed to one person; it needs volume only distribution and years produce. Efficient scale is partly open: pick a flow so small that at your take it feeds one filter, be that filter, and keep the fixed cost near zero so nobody bothers to build a second.
What the matrix says of its moats
- #24 Efficient scale Solo ~ partly Capital ◐ medium AI → eroding Rent ✗ not for sale
- #15 Economies of scale Solo ✗ no Capital ● high AI ↑ AI-resistant Rent ✗ not for sale
In nature, in business
- Nature
- baleen whales, planktivores
- Business
- credit cards
Cases
- Visa — Runs the rails between banks and merchants and takes a small fee on each card payment, a sum no cardholder sees and few merchants can route around.
- Google Ads — Sells attention one auction at a time, cents per click, across so many searches that the pennies add up to most of the parent company's revenue.
- Vanguard — Charges a few basis points a year on index funds, a fee too small to shop around for, spread over one of the largest asset pools in the world.
Combines with
- Omnipresence — Omnipresence is the width itself: a filter accepted by default everywhere touches the whole flow, and the whole flow is what makes the take add up.
- Low-friction — Low-friction keeps the take invisible; a fee that costs a click to pay is one the flow will route around.
- Aggregation — Aggregation brings many small participants onto one surface, and many small participants is exactly the flow a thin take needs.
- Standardization — Standardization turns one format into the way everyone moves, so the flow passes through a single filter instead of many.
Fights with
- Specialization — Specialization narrows the flow to a niche; a thin take on a narrow flow never reaches the volume where its fixed cost closes.
- Curation — Curation turns participants away by design, and every participant turned away is flow the filter no longer touches.