# Your swap screen tells you a price, not what routing saved you > Every quote answers what your size costs. The harder question is what size fits a cost you name — and how much of it exists only because of splitting. - Source: https://defiloops.com/blog/what-a-price-does-not-tell-you - Published: 2026-09-16 - Category: Architecture - Tags: swaps, routing, liquidity, execution - Author: DeFiLoops --- Ask any swap interface for a quote and it answers the same question: *what does my size cost?* You type 200 USDC, it says 0.00% price impact, and you press the button. That is a fine answer to a question almost nobody actually has. If you are moving 200 dollars, nothing you do matters. If you are moving two hundred thousand, the question you have is the other way round: **what size fits inside a cost I am willing to pay?** Here is a real read of USDC→WETH on Base, at block 50,574,016, across 11 pools. ## Read the gap, not the bars The dark bar is the best single pool. The pale bar is every pool that clears the same cost, added together. **The gap between them is what the routing is worth**, in dollars, at that cost. At five basis points, one pool takes $103.1K and seven pools take $148.5K. Routing is worth **44% more size at the same price.** At fifty basis points it is $1.05M against $1.72M — the same shape, larger numbers. The instinct is that the deepest pool is the answer and the rest are rounding. It is not. At a fixed cost ceiling, the right quantity is the **sum** of what each pool can absorb inside that ceiling, not the maximum of them — and the tail is not small. Seven pools clear five basis points here; the seventh takes $90 of it, which sounds pointless until you notice the six above it are doing the work the single-pool answer throws away. ## What no other quote shows you Quotes are counterfactual-free. You are told a price and, if the interface is generous, a route — but never what you would have got *without* it. The number that would let you judge whether the routing did anything is the one number nobody prints. I went looking for a swap interface that does. Depth analysis exists as standalone tooling, computing slippage at fixed sizes like $1K, $10K, $100K and $1M. Aggregators [show splits at quote time](https://apify.com/gochujang/dex-liquidity-depth-analyzer) for the size you asked about. But those answer cost-for-a-size, and they answer it for one size, after you have already decided. Not that nobody else could build this, and not that no interface anywhere has something like it. I could not find one, which is a weaker statement than "it does not exist" and the only one I am entitled to make. What I can say precisely is what it inverts: **size-at-a-cost rather than cost-at-a-size**, and the counterfactual printed beside it. ## What it is actually for Three things, and none of them is the 200-dollar trade at the top of the page. - **Sizing.** You can see where your size stops being free. Somewhere between $148.5K and $308.2K on this pair, five basis points becomes ten. - **Deciding whether to split at all.** If the gap is narrow, routing is ceremony and one pool is fine. Splitting a trade [usually does nothing](/blog/splitting-a-big-trade-usually-does-nothing), and this is how you tell the usual case from the exception before you trade rather than after. - **Reading the venue, not the token.** A pair with one deep pool and a long thin tail behaves nothing like one with four comparable pools, and the total alone cannot tell them apart. So any row opens. Here is the five-basis-point row from the same read, opened up: That is one dominant pool and a six-pool tail, which is a particular shape with particular consequences. It also shows something the totals hide: **the cheapest pool in the tier is not the biggest one.** Two of the small pools fill under one basis point while the pool doing most of the work fills at 2.4 — so a router optimising purely for headline price and a router optimising for size at a bound will not pick the same thing. ## The honest limits The ladder is measured at one block. Liquidity moves, and a reading taken now is a statement about now — the block number is printed for exactly that reason. It also describes *capacity*, not *outcome*. It says a certain size fits inside a certain cost at that moment. It does not promise you fill there: what protects you at execution is the minimum output you signed, and that is still the thing doing the work. For the trade in the example — 200 USDC, 0.00% impact — none of this matters at all. The ladder earns its place on the screen for the person who is about to move a hundred times that, and it is worth saying plainly that most people looking at it will not need it.