The stock is in the warehouse. The listing still isn’t live.
Between “in the warehouse” and “on sale everywhere” sit missing attributes, hand-written copy, approval queues and channel rejections. Where each step stalls, and why.
On this page
Ask a launch manager where the time went. Often you won’t hear “manufacturing ran late” or “freight was delayed.” You’ll hear this instead. “Content wasn’t ready.” “Translations weren’t done.” “Amazon rejected the listing, attributes were missing.” “Zalando wouldn’t take it, the compliance fields weren’t validated.” “The approval sat with the brand team for two weeks.”
The product sits in the warehouse and the listing isn’t live, because the data still has to be collected, structured, enriched, checked, approved and published for every channel. The revenue clock runs the whole time. And often it isn’t manufacturing or freight that holds the launch. It’s the catalog.
This isn’t a copywriting problem. It’s a coordination problem. The values live in fragments across systems. The enrichment needs several teams. The checks need to know what each marketplace asks for. The approvals need a queue that nobody built. And the whole thing repeats, by hand, for every product, every channel, every market, every season.
Product data is the critical path.
Supply chains have had decades of investment. Manufacturing, logistics and inventory management all have mature tooling behind them. Getting a product from “it’s in the warehouse” to “it’s live and purchasable on every channel” has had far less.
So for many brands, the longest stretch of a launch isn’t production to warehouse. It’s warehouse to live and purchasable on every channel. That stretch is product data work, start to finish.
Note
Do the arithmetic on your last launch.
Take your last seasonal launch. Count the days between “inventory available” and “live on all target channels.” Multiply that gap by the number of SKUs, the average selling price and a conservative daily conversion rate. What comes out is the revenue you didn’t earn because the catalog wasn’t ready. Compare that figure with what fixing the process would cost. The number is an estimate, not a measurement. It still tends to settle the argument.
The cost lands in several places at once.
Scroll sideways to see the whole table.
| Where it lands | What the delay costs | How it compounds |
|---|---|---|
| Position on the marketplace | Whoever lists first takes the early demand and the first reviews. | A late listing starts with no history, against rivals that already have reviews and a sales rank. |
| Seasonal windows | Fashion, home and seasonal ranges sell inside a fixed window. Miss it by a couple of weeks and you lose a real share of the season. | The weeks lost at the start of the window don’t come back. |
| Campaigns | The campaign moves, or it runs against listings with holes in them. | Ad spend pointed at an incomplete listing converts poorly. |
| Channels out of step | Live on DTC but not on the marketplace, or the other way around. | Brand trust erodes when a shopper can’t buy the product in the channel they use. |
| Inventory turns | Stock that isn’t on sale anywhere is dead capital until the listing exists, and it costs you something to hold in the meantime. | Catalog delays pull inventory turns down, and finance notices. |
Where the friction actually sits.
A launch rarely misses for one reason. It misses because small blocks stack up across a process that is split between teams and systems: in the inputs, in the writing, in review and at the channels.
- Supplier onboarding one at a time. Products arrive from a supplier or from the ERP in a different shape each time, with only part of the required fields filled in. Mapping the file is a manual project per supplier, and brand, compliance, merchandising and localization type in the rest.
- Enrichment done by hand. Descriptions, keywords, bullet points, the mapping into each channel’s categories: written product by product, channel by channel.
- Translation waiting. Launching in 5 markets means 5 locale versions. Agencies take weeks, review cycles add more, and localization often ends up the longest serial dependency in the chain.
- Approval queues. Brand, legal, compliance and merchandising each review in turn. One person’s vacation is a 2-week block on the whole catalog.
- Compliance sitting in a queue. GPSR and marketplace compliance fields, and DPP data as it phases in, need legal or regulatory review. Those teams are reading hundreds of products, so the queue is measured in weeks.
- Channel rejections. Every channel wants a different shape: bullet points for Amazon, structured feeds for Google Shopping, metafields for Shopify, reformatted by hand each time. A listing comes back for a missing required attribute, a category mapping or an image that misses the spec. Each one is an investigation, a correction and a resubmission.
Each block adds days. Together they add weeks or months. Then they repeat for the next product family, the next season, the next market expansion. The problem isn’t one slow launch. It’s operations that are slow by construction.
Storage isn’t the hard part. Coordination is.
The first reaction to a slow launch is usually “we need to centralize our data.” That’s necessary, and it isn’t enough. Centralizing answers where the data lives. It doesn’t answer how a raw supplier file becomes a listing that passes a channel’s checks, or who touches it on the way.
The steps in between each need somewhere to happen.
Scroll sideways to see the whole table.
| Step | What it does | Without it |
|---|---|---|
| Bringing the data in | Supplier files, ERP exports and typed-in values all arrive through saved mappings that line up the fields, check them and settle what wins when two sources disagree. | Data arrives as email attachments. Someone re-keys the values into the PIM. The mistakes travel with them, and no two imports look alike. |
| Getting the content written | Writing moves through set stages: an AI draft, a team review, the locale versions, then the per-channel wording. Each stage has an owner and a date. | Writing happens wherever there is room for it. Nobody can say what is done and what is pending. The gaps turn up at publish time. |
| Checking before it goes out | A product can’t reach publish-ready until it passes the completeness checks, the compliance checks and the scoring against what each channel asks for. | Products go out with fields missing. The marketplace refuses the listing. Someone investigates, fixes and resubmits, and that is days for every rejection. |
| Moving it through approvals | Each product reaches the people who have to approve it, at the stage where their approval matters, side by side where that is possible, and it escalates when a review stops moving. | Approvals happen in Slack messages and email threads. Products sit in queues. Nobody can see the pipeline. |
| Publishing to the channels | Once a product reaches publish-ready it goes out to every channel it is configured for at the same time, in the shape each one wants. | Publishing is done by hand, one channel at a time. The Amazon listing goes live Tuesday. Shopify Thursday. Another marketplace next week. In between, the channels disagree. |
| Holding a regional launch together | Every locale has to reach readiness before any market goes live, so a market doesn’t open on its own unless you meant it to. | The product launches in some markets and not others. Shoppers in the markets left out find it and can’t buy it. Brand trust erodes. |
Read down that table and every row is a place where the work either moves or stops. Take those places away and the PIM is a database with a form on top: the data is in a system, and the product still isn’t live. Build them and launches run the same way each season instead of being improvised.
Before and after.
An illustration with round numbers:
Scroll sideways to see the whole table.
| Job | Steps split up | Steps wired together |
|---|---|---|
| Seasonal collection launch (200 SKUs, 4 markets, 3 channels) | Data comes over in Excel from the design team. Enrichment by hand: 3 weeks. Translation agency: 2 weeks. Compliance review: 1 week. Channel formatting: 1 week per channel. Call it 8-10 weeks before the whole collection is out. | The PLM export comes in as a structured import. AI drafts descriptions, bullets and attributes for review. Translation drafts follow for each locale. Agents flag the compliance gaps. Amazon and Shopify publish once the products are approved; other channels get a file feed. Each step starts from the one before it, without a handover by spreadsheet. |
| Marketplace expansion (a new marketplace) | “What does Zalando need?” is a 3-week research project. Gap analysis per product. Reformatting by hand. Category remapping. A 6-week project. | The Channel launch agent drafts the new feed and the missing channel fields per product for review. Its assumptions come from the AI model, so someone checks them against the retailer’s guide. |
| Regional product launch (launch existing products in new EU market) | Translations to commission. Compliance to check. Prices to adjust. Per-product work by hand. 4-6 weeks per market. | A translation agent fills in the new language for a person to review. Compliance gaps are flagged. |
| Bulk attribute update (new compliance field required across 10,000 products) | Export the spreadsheet. Research the values per product family. Re-import. Hand it to compliance to check. A project measured in months. | AI drafts the new field from the specifications you already hold, where they support it, and people review the rest. No spreadsheet round-trip. |
When the buyer is a machine.
The steps above already matter when people do the buying. They matter more when software does it. Agents find products through the feeds and store catalogs merchants expose, marketplaces check listings by program, and a shopper asks an assistant instead of reading a page.
- AI shopping agents work from the product data merchants expose through feeds and store catalogs. Missing attributes make a product harder for them to match.
- Marketplaces already check listings when they’re submitted, and some reject them automatically.
- A conversational answer is only as good as the structured fields behind it. Ask an assistant about a product with gaps in its data and the answer is unreliable.
- Enrichment agents can take a lot of the writing off the queue: drafting descriptions, pulling attributes out of specifications, scoring what is ready. They need a defined schema to write into, and a review step before anything reaches the catalog.
- Personalized descriptions, contextual recommendations and adaptive pricing all read from complete attribute sets. A gap in the data becomes a gap in the page, on its own.
- For a launch to run from import to enrichment to checks to publish with nobody in the middle, each of those steps has to fire on an event, not on someone remembering.
Slow catalog work doesn’t only push revenue back. It can keep the product out of the systems that do the finding and the buying. The gap between “the product exists” and “an agent can find it” is the one worth measuring now.
Some of this exists in CataZenta now: specialist agents, flows that chain them, an approval queue their suggestions wait in by default, and scheduled Amazon and Shopify runs that send changed, approved products. Saved import mappings, agent instructions and channel field mappings carry over from one launch to the next. CataZenta doesn’t have source-conflict rules on import, compliance findings that block publishing, routing with escalation, a pipeline view, direct publishing to other channels, or a hold that keeps a launch back until every locale is ready.