Economics & Markets
Lower Fees, Weaker Moat
A fee cut does not just lower your price — it selects which sellers you get, and they leave first.
2026-08-291 min read
Take-rate is not only revenue; it is the budget for everything that makes a marketplace worth using. Curation, fraud losses, dispute handling, search that surfaces the right item, delivery that arrives when promised, all of it is funded out of the margin on each transaction. Cutting the fee to win sellers spends that budget on volume instead. The catalog grows while the machinery that made the catalog navigable stops improving, and buyers experience the change as a site that has more things and is harder to buy from.
The deeper problem is who the discount recruits. A seller who joins because the fee dropped is, by definition, a seller for whom the fee was the deciding factor: price-sensitive at the margin, running thin, competing on being cheapest. That is the population now selected into the catalog, and it behaves differently from the sellers already there. Consider a niche marketplace for handmade goods that halves its commission to accelerate seller growth. It gets the growth. It also gets resellers of mass-produced items who survive on volume, buyers who came for handmade begin finding factory output in the same search results, and the curation that was the entire reason to shop there quietly stops being true.
The trap is that the move is hard to reverse. Raising the rate back hits the price-sensitive cohort hardest, so the sellers who leave are precisely the ones the cut was made to attract, and the growth number goes backwards in public view. A fee cut is also the single easiest thing a competitor can copy: same afternoon, no capability required. What cannot be copied cheaply is the curation and reliability the fee was funding, which is why the lever that feels fastest usually trades a durable advantage for a temporary one.
Key insights
Take-rate is the funding line for curation, trust, and delivery — the things buyers choose you for.
A discount selects sellers for whom price was decisive, which is the cohort most likely to leave next.
A competitor can match a fee cut in an afternoon; the capabilities it was funding take years to match.
Why it matters
The seller mix a discount attracts is selected for price sensitivity, which makes the decision expensive to reverse and trivial for a rival to match.
Margin funds the differentiation buyers actually notice, so spending it on volume erodes the reason anyone chose the platform.
Use this tomorrow
1Pull the sellers who joined in the three months after your most recent fee change and count what share list products that already had five or more sellers on your platform.
2Add up last quarter's spending on curation, fraud, and delivery reliability, then divide it by total transaction margin to see what share of that budget a one-point fee cut would remove.
Go deeper
Two-sided platforms face a coordination problem: each side's willingness to participate depends on the other side's quality, not merely its size. A fee cut moves one lever, seller count, while quietly defunding the quality signals that keep buyers arriving, so the two sides can drift out of balance even as the headline growth metric improves. Economists describe this pattern as adverse selection: the price you set determines who self-selects in, and a lower price systematically recruits the participants least attached to everything else you offer. The number that looks like success is measuring the side of the market that is cheapest to grow.
There are conditions where the cut is correct, and they are worth naming precisely. If a platform's value genuinely comes from raw selection rather than curation, and buyers are close to indifferent between sellers, then volume is the moat and a lower rate compounds it. The diagnostic is whether a buyer would notice the difference between your best seller and your median one: where they would, curation is load-bearing and the fee funds it; where they would not, you are running a commodity exchange and should price like one. Getting that diagnosis wrong in either direction costs more than the fee level itself ever will.
