Late deliveries, parcels stuck in transit, failed delivery attempts, customs holds — a copy-paste prompt that finds them in your order export. Works properly for a small store shipping a few dozen parcels a week. Here's the method, and the exact point it falls apart.
First, the thing nobody says: ChatGPT can't see your parcels. Paste a tracking number and ask where it is, and you'll get an invented answer.
What it's good at is reading a list and telling you which rows matter. You bring the data. It does the sorting.
Shopify admin → Orders → filter Fulfilled → last 2 weeks → Export → CSV.
Last status and Status date. Shopify has the tracking number but not the tracking events — those live with the carrier. So you check each one and paste the latest event in.
No events at all? Write no scans. Those rows matter most.
Paste it into ChatGPT, Claude or Gemini, attach your sheet, and read the table it gives back.
Triage these Shopify shipments for me. Each row has: order number, fulfilment date, carrier, destination country, last tracking event, date of that event. Today is [DATE]. Sort every row into one of four groups: ACT TODAY — something is wrong and waiting makes it worse. Failed delivery attempt, refused, returning to sender, customs waiting on someone, reported lost or damaged. WATCH — nothing confirmed wrong, but quieter than expected for the route. NO SCANS — label exists, carrier never physically scanned it. Say how many days. FINE — moving normally, or delivered. For anything in the first three, give me: order number, one line on what the tracking actually says, what to do about it, and how urgent versus the others. Rules: - Use only what's in the data. If a row is ambiguous, say so instead of guessing. - Sundays and public holidays aren't working days. Most carriers don't scan then. - Never call a parcel lost or delayed unless the tracking says so. - International parcels move slower. Four quiet days crossing a border is not four quiet days within one country. - Table, ordered by urgency, worst first. - End with one line: how many need action today, how many to watch.
The rules at the bottom do the heavy lifting. Without them you'll get told a parcel is lost because it hasn't moved since Friday — and you'll ring the carrier about a parcel sitting in a depot over a normal weekend.
It isn't the volume. It's the timing.
A parcel that starts heading back to sender on Tuesday can still be intercepted on Tuesday. By Monday's export it's already at your warehouse and you're refunding an order that could have been delivered.
Same with a failed delivery attempt — there's a window where the customer can redirect it, and it closes in a couple of days.
A weekly check does find your delivery exceptions. It finds them after the window to do something about them has closed.
A few dozen a week: use the prompt, set a reminder, keep your money.
Past that, the thing you need isn't smarter analysis — you already have that for free. It's for step 2 to stop existing.
It reads the carrier scans for every parcel your store ships, on its own, and surfaces every delivery exception — parcels stuck in transit, late deliveries past what the route allows, failed delivery attempts, customs holds, refusals and returns — with the reason and a message ready to send.
No export. No pasting statuses. And it never contacts your customers by itself.
Free for the first 50 shipments a month — around the point where the manual method still works fine anyway.
See it on the Shopify App StoreWe'd like to measure how people find this site. We use Google's advertising cookies to see which searches and ads bring merchants here. Nothing runs until you choose, and declining changes nothing about how the site works. Read the privacy policy.