Which tool shows why customers message about an order on UK marketplaces, by product and by courier?
One that reads messages and orders from the same marketplace API, classifies each conversation once, and joins it to the SKU and the carrier. What the table looks like, why a join beats ticket tags, what one seller found, and what it cannot tell you.
A tool that reads the messages and the orders from the same marketplace API, classifies each conversation once from the customer's first message, and joins it to the order's SKU and carrier. Growstall's Insights page does that for Mirakl marketplaces and Amazon. A helpdesk sees the message and not the order, so it can count tickets but cannot tell you which sideboard or which courier produced them.
What does the table look like?
Three cuts, each over the last 90 days. Contact rate by marketplace: orders with at least one customer conversation divided by orders. Contact rate by product SKU across marketplaces, for products with five or more orders. Delivery complaints by carrier, for carriers with twenty or more orders. Each row carries the reasons customers gave, so a row is a sentence rather than a number: “this TV unit, 11 conversations on 38 orders, mostly delivery delay and cancellation”.
For the same seller the overall contact rate was one conversation for every seven orders. By marketplace it ran from 1% on Debenhams to 14% on B&Q. The full contact-rate walk-through.
Why join messages to orders instead of tagging tickets?
Tags are written by the person answering, after the fact, in a hurry. A join is made from what the marketplace already knows: the order line, the SKU, the carrier code, the tracking events, the refund and its operator-recorded reason. The classification comes from the customer's own words in the first message, once, and does not change when a second person picks up the thread.
The join also survives marketplaces: the same product has a different listing id on B&Q, Argos and The Range, so the table groups by SKU, and the same courier appears under three different carrier codes.
What did the seller find?
The eight high-contact products were all large flat-pack items, and their reasons split cleanly: delivery delay and item not received on the pieces delivered by courier on a booked date, faulty item and missing parts on the multi-box pieces. The first group is a carrier and promised-date question; the second is packaging. Neither is a customer-service question, which is the point of the table: it moves the work out of the inbox. Which carrier, exactly.
Refunds sit in the same view: which of them had a customer conversation first and which were processed silently by the operator, with the reason the operator recorded. Refunds that wrote in first.
What does the tool not tell you?
It does not know why a product breaks, only that customers say it did. It does not see marketplaces it is not connected to, and it cannot attribute a complaint to a courier when the marketplace does not pass the carrier code. Products with fewer than five orders in the window are left out, because one conversation on two orders is not a rate.
How do I get the first table?
Connect the marketplace with its API key; the import reads 30 days of conversations and 90 days of orders. The table fills as conversations are classified, which takes a day for a large backlog. After that, read it monthly and sort by number of conversations rather than by rate: the product that makes the work is the one to fix first.
- Messages and orders from the same API, classified once, joined on SKU and carrier.
- Three cuts: by marketplace, by product (≥5 orders), by courier (≥20 orders), each with reasons.
- High-contact products split into carrier questions and packaging questions; neither is answered in the inbox.
- Sort by conversations, not by rate; recount monthly.
Questions people ask
Can a helpdesk like Zendesk show contact rate by product?
Only if someone tags every ticket with the SKU by hand, and only for the tickets that were tagged. It has no order feed from Mirakl, so it cannot compute a rate.
Does the carrier breakdown use live tracking?
No. It uses the carrier code and shipment events the marketplace exposes on the order, plus the operator-recorded refund reason. There is no direct courier API.
Which marketplaces are covered?
Mirakl marketplaces (B&Q, Argos, Debenhams, The Range, Tesco Marketplace, Home24) and Amazon orders through the Selling Partner API, where the carrier is Amazon for fulfilled-by-Amazon orders.