Checkout Signals That Predict Bracketing Before the Order Ships
Most bracketing programs react after delivery, when the return label is already the cheapest remaining outcome. But bracketing is rarely a surprise. The order itself announces it: two sizes of the same item in one cart is not a prediction, it is the thing itself, visible in the warehouse queue hours before it ships.
subtler signals sit one step earlier, in browsing behavior and account history. Used well, they let a store spend its fit help where the uncertainty actually is, instead of blasting every customer with size popups that confident shoppers learn to ignore.
Two sizes in the cart is a fact, not a forecast
Start with the obvious signal because it needs no model. A cart holding the same product in two sizes will, in the overwhelming majority of cases, return one of them. Everything about that order can be planned around the near certainty: pack the return label and instructions where they will be found, and make the keep one path the easiest thing in the box.
The pre shipment window is also the last cheap moment for a fit check. A short message between order and dispatch, offering to confirm the more likely size based on the details you have, saves a small share of these orders outright. Even when it changes nothing, it primes the shopper to use the exchange flow rather than a refund when the parcel arrives.
Size toggling tells you where your pages fail
Shoppers who switch sizes repeatedly on a product page, open the size guide twice, or move between two product pages in the same style are showing you uncertainty in real time. Aggregated, those sessions point at the exact styles and size runs where your information breaks down, which is a merchandising to do list ranked by shopper confusion.
At the individual level, a light touch works better than a hard sell. A fit note that appears after repeated toggling, with model measurements or a comparison the shopper can check at home, answers the question they are visibly asking. The same popup shown to every visitor becomes wallpaper within a week.
Account history separates guessing from gaming
A two size order from a new account and the same order from an account with five kept purchases mean different things. The first is a stranger guessing. The second may be a loyal shopper whose size genuinely sits between yours, or a pattern worth a closer look if nothing is ever kept. History turns a raw signal into a decision about how much help, or scrutiny, an order deserves.
Keep the scrutiny proportional and mostly invisible. High cost responses like holding shipments or demanding confirmation belong on the small set of accounts with repeated keep nothing behavior. Fit help, by contrast, scales to everyone the signals flag, because its downside is a ignored email rather than an insulted customer.
Close the loop after delivery
Signals are only worth collecting if outcomes flow back into them. When a flagged order keeps the size your fit check suggested, that is evidence the intervention works. When flagged orders keep the other size, your size guidance for that style is probably wrong, which is more valuable to know than the bracketing itself.
Review the signal list monthly against kept outcomes. Retire signals that flag everyone, sharpen the ones that predict well, and feed the product page fixes back to the styles generating the most toggling. The program matures from prediction into prevention, which is where the real savings live.
Related reading
These earlier guides go deeper on connected questions: Holiday Gift Orders and Bracketing: Separating Gift Uncertainty From Fit Bracketing, Returns Data by Size: Spotting the SKUs That Invite Bracketing, and Bracket Contagion: Why One Bracketed Order Predicts the Next.
Frequently asked questions
What is the strongest signal an order will bracket?
The order itself: the same item in two sizes in one cart. That is not really a prediction, it is bracketing already happening, and it lets you plan packaging, inserts, and a fit check before the parcel ships instead of reacting when the return arrives.
Can browsing behavior predict bracketing?
Partly. Repeated size switching on a product page, multiple size guide opens, and comparing the same style across pages all correlate with size uncertainty. Aggregated, they are most useful as a ranked list of product pages whose fit information needs fixing.
Should flagged orders be held or challenged before shipping?
Almost never for fit signals alone. Bracketing is usually honest uncertainty, and blocking it insults good customers. Save hard responses for accounts with repeated keep nothing patterns. For everyone else, use the signal to offer fit help and to package the likely exchange well.
How do we know if acting on signals works?
Compare flagged orders that received a fit check with flagged orders that did not, on kept revenue and exchange rate. Then check whether flagged styles keep generating signals after you fix their pages. Prediction that never reduces future bracketing is just surveillance with extra steps.