A strong answer. “There are two quite different failures hiding behind ‘underperformed VWAP’ and I'd separate them before saying anything, because the answers are opposite.

One: we tracked VWAP badly. Our execution schedule deviated from the volume profile — we were too slow early and had to catch up, or a volume forecast was wrong because of an index rebalance or an expiry we didn't account for. That is our fault, it is measurable, and I'd own it and say what we'd change.

Two: we tracked VWAP well and the client still lost money. This is the more common case and it is the one worth explaining properly. If the order is large relative to the day's volume, our own trading moved VWAP. We are part of the benchmark we are measured against, which makes VWAP self-referential above a few percent of volume. A client who was 20% of the day's volume and matched VWAP exactly still paid substantial impact, and VWAP simply cannot show it.

So I'd reframe onto the benchmark that actually measures cost: implementation shortfall — the difference between the price when they decided and the price they got, including the cost of what didn't get filled. That captures impact and timing risk, which is what they actually paid, and it is the honest number even when it's less flattering to us.

Then the decomposition, so it's diagnosis rather than defence: spread cost, market impact (temporary and permanent), timing/delay cost, and opportunity cost on the unfilled portion. Once a client can see that 70% of their cost was impact from an order sized at 25% of ADV, the conversation changes from ‘your algo is bad’ to ‘how do we schedule this differently’.

And the honest recommendation is usually not a better algo. It's to trade over more days, or use a liquidity-seeking strategy rather than a schedule-following one, or accept more timing risk in exchange for less impact. Telling a client their order is too big for one day is the most useful thing you can say to them and the thing desks avoid saying.”