Halfway through a major festival, the campaign dashboard says you are winning. Revenue is ahead of plan, your ROI and return on ad spend (ROAS) look strong, and someone drafts the optimistic note to headquarters. The note is wrong — and it is wrong in a predictable direction. Mid-festival ROI is always too high, never too low.
The reason is timing, not optimism:
- A sale is recognised the moment the order is placed.
- The return arrives seven to ten days later.
- Mid-festival, you have booked nearly all of the revenue and almost none of the returns.
- ROI calculated on complete revenue and incomplete returns looks its best exactly when it is least true.
By the time the return window closes — after the festival, after the recap deck, after the budget decision for next quarter — the real number has settled several points lower. You acted on the version that flattered you.
The lag is worse than a simple delay, because a large share of festival returns is deliberate. Festival demand is pulled in by threshold coupons: customers over-buy to unlock a discount, then return the excess once it is secured. Those returns were never disappointment. They were planned the moment the order was placed. During a coupon-driven event, the mid-festival number is not just incomplete — part of it was never going to stay.
In luxury and premium fashion, where return rates on Tmall and Douyin run high, this is not a rounding error. A campaign showing a comfortable return on ad spend mid-festival — strong ROAS on paid traffic, healthy ROI — can land below breakeven once returns are recognised. Teams that re-up spend on the strength of that number are funding a result that has not happened yet — and partly never will.
The lag also corrupts comparison. Early-festival days look more profitable than late-festival days for one reason only: their returns have not caught up. Read at face value, that pattern says "lean into the early window, the ROI is better there." It is not better. It is younger.
So what is the discipline? It is simple to state and rare to find:
- Judge performance on a returns-adjusted basis, never on gross sales.
- Build an expected return rate — by category, price point, and promotion type — into the live dashboard, so the number you act on already carries the refunds it has not yet seen.
- Hold final budget decisions until the return window has actually closed.
This is the difference between measuring what happened and measuring what settled. The festival is built to make you decide fast. The returns decide whether you were right, and they arrive after the decision. A brand that confuses the two will keep funding its least profitable days and calling them its best.
In a mature market, the edge is no longer in spending more during the festival. It is in knowing, before anyone else, what the festival actually earned.
Foresight Performance Partners helps luxury and prestige brands measure China e-commerce by what settles on the P&L, not by what flashes on the dashboard. To start a conversation, please get in touch.