How to Auto-Tag High-Risk Orders on Shopify

Yes — Shopify runs its own fraud analysis on most orders and computes a risk recommendation, typically shown as Low, Medium or High risk, sometimes with an explicit Cancel or Investigate suggestion, based on signals like a mismatch between billing and shipping details or unusual patterns for that customer or product. It's genuinely useful information, and it's real, computed data, not a manual guess. What it doesn't do is turn that recommendation into a tag on the order. The recommendation lives on the individual order's own page, visible the moment you open it, but nothing filters your order list, your fulfillment queue or a saved search by it, because none of those tools read Shopify's risk field directly — they read tags. Without a tag, the only way to find every risky order is to open them one at a time and check, which doesn't scale past a handful of orders a day.

It's visible the moment you open that specific order — the recommendation sits right there on the page — but that assumes a person opens every order before it ships, which stops being realistic the moment volume grows past a handful a day. On a store fulfilling automatically, or using a packing app, or routing orders through a saved search, an order can sail straight through to shipping without a human ever seeing the risk recommendation Shopify already computed for it. The information being correct and the information being ACTED ON are two different things, and the gap between them is exactly where a tag helps: it turns a fact only visible to someone who opens the order into a fact any downstream tool — including one that never shows a human the order page at all — can check and act on automatically.

Yes — you can build one rule per recommendation, or a single rule that matches both into one combined tag, depending on how your process actually treats them. Some stores handle a Cancel recommendation and an Investigate recommendation very differently: a Cancel-recommended order gets held entirely, pending a manual decision on whether to refund and cancel it outright, while an Investigate-recommended order ships as normal but gets flagged for a quick follow-up call or email to confirm the buyer's details. Two separate rules, each tagging its own outcome — 'hold-cancel' versus 'flag-investigate' — let a fulfillment queue treat them differently without anyone deciding case by case. Other stores treat both recommendations the same way operationally, anything flagged gets a second look before shipping regardless of which specific recommendation Shopify gave, in which case one rule matching both into a single 'needs-review' tag is simpler.

No — and it isn't meant to. The rule's job stops at flagging; a human still makes the actual call on whether to cancel, ship, or ask the customer for more information, because that decision depends on context (order value, customer history, how the specific risk signal was triggered) that isn't safe to automate blindly. What the tag changes is the shape of that review: instead of every order getting the same brief, unfiltered glance, review becomes a filtered queue of exactly the orders Shopify itself flagged, so the time a human spends actually goes toward the orders that need it. Stores that want a harder stop than a soft flag can pair the tag with a fulfillment hold, so a tagged order can't ship automatically until someone clears it — but that's a separate decision layered on top of the tag, not something the tag does by itself.