Product Data That Sells: Building Listing Content That Works on Every Channel
Marketplaces read your listing as a database record, not a page. A field-by-field look at what channels actually parse, and why packed dimensions cause the most downstream pain.
Most teams treat a product listing as a page to be written. Marketplaces treat it as a database record to be parsed — and when the two views disagree, the marketplace wins. Listings get suppressed, variants split apart, search filters skip your item, and nobody can point to a single broken thing, because nothing is broken. The data is just incomplete.
Getting product data right is unglamorous work that pays off every time you add a channel, a language or a warehouse.
Your listing is a record, not a paragraph
A marketplace does at least four things with a listing before a shopper ever sees it: it categorises the item, matches it against existing catalogue entries, indexes it for search and filters, and passes attributes to shipping and tax logic. All four read structured fields. The prose description is nearly the last thing consulted.
That reframes what "good product content" means. A beautifully written description attached to a record missing its material attribute will lose to a plain one that filled the field in.
The fields that do most of the work
| Field | Why the channel cares | Common failure |
|---|---|---|
| Product identifier (GTIN/UPC/EAN) | Matches your item to the channel's catalogue record | Reused across variants, or invented |
| Category assignment | Decides which attribute set and filters apply | Placed one level too shallow, so filters never apply |
| Attributes | Powers faceted search and comparison | Free text where a controlled value was expected |
| Variant relationships | Groups sizes and colours under one parent | Variants uploaded as unrelated standalone items |
| Images | First and often only evaluation the shopper makes | Inconsistent framing across a variant family |
| Title | Search matching plus human scanning | Keyword stuffing that trips channel style rules |
| Dimensions and weight | Shipping cost, packaging, sometimes eligibility | Copied from the product, not the packed carton |
That last row causes more downstream pain than the rest combined.
Ship dimensions belong in the product record
The dimensions that matter operationally are the packed carton's, not the bare product's — and they need to live in the same record everything else reads from. When they are wrong or missing, storage planning, packaging selection and shipping quotes all inherit the error, and the correction has to happen at the packing bench under time pressure.
The practical rule: measure and weigh the packed unit once, on receipt, and write it back to the master record. Do it during onboarding of a new SKU, when nothing is urgent yet. This is one of the checks worth completing before you go live anywhere new, alongside the rest of the pre-launch fulfillment checklist for a new sales channel.
Images are structured data too
Channels increasingly parse images rather than merely display them: background rules, minimum resolution, fill ratio, whether the main image may contain props or text overlays. A set that passes on one channel can be rejected on another for a reason that never appears in your dashboard.
Shooting to the strictest specification you sell under, and keeping the raw files, is what makes the next channel a re-crop instead of a re-shoot. Keep image roles explicit — main, alternate angle, scale reference, in-use — because most channels map roles to slots, and an unmapped set arrives in whatever order it uploaded.
One source of truth, many channel views
The failure pattern is predictable: the master data sits in a spreadsheet, each channel gets a hand-edited export, and within two quarters no two channels agree on the same SKU. The fix is not a bigger spreadsheet. It is deciding that one system holds the canonical record and every channel receives a view of it, transformed to that channel's rules.
Which system holds it matters less than the discipline that only one does. What has to be true is that a correction is made once and propagates, rather than being made in the place someone happened to notice the problem.
That same principle applies to the quantity field, where the consequences are immediate rather than gradual — the mechanics of which we cover in splitting stock across channels without overselling.
Decide who owns each field
Product data decays because ownership is ambiguous. Merchandising edits titles, operations edits weights, finance edits tax codes, and each assumes the others will notice a conflict. A one-page owner map — field, owner, where it is edited, what depends on it — resolves most of it, and it takes an afternoon.
Add two habits: a required-field check that runs before anything publishes, and a quarterly audit against what the channels actually hold rather than what you believe you sent them. The gap between the two is usually the interesting part.
If this is work you would rather hand over, our e-commerce management team takes on catalogue build and channel maintenance as an ongoing service. And because packed dimensions are the field that crosses from catalogue into the warehouse, keeping them accurate in AIDWMS is what stops a listing error from becoming a packing-bench problem.
