
Product pages complete and consistent across website, marketplace and store
When a new collection arrives, the editorial team receives incomplete data from suppliers, each written in its own way, and has to turn it into pages ready for every channel and in several languages.
From the supplier's sheet to the product page
When a collection enters the catalogue, merchandising loads the codes into the ERP system and the editorial team receives spreadsheets from suppliers with composition, sizes and measurements, often without the fields the website and marketplaces require. The editors complete the attributes by hand, write the descriptions and send them for translation. Most of the time goes on chasing the missing data from suppliers and product teams, while items already photographed wait for their page before they can go online.
The agent puts the pages together, the editorial team chooses what to publish
The agent reads the supplier's data and the item's master data, maps them to the catalogue's attributes and writes descriptions and translations following the brand's tone-of-voice rules held in the Company Brain. When a piece of data is missing, such as the composition of a fabric, it leaves the field empty and prepares the request to pass on to the supplier, instead of inferring it from a similar item. The pages arrive in the queue of the editor, who corrects them and decides on publication, while cases outside the rules go to a person.
- The supplier's data and the item's master data arrive
- The agent completes the attributes and flags the missing data
- Writes descriptions and translations in the brand's tone
- The editorial team corrects and decides what to publish
Illustrative example
Data, modules and decisions behind a product page
| Item | In this case |
|---|---|
| Data | In this caseItem master data, suppliers' sheets, the attributes each channel requires, the brand's tone-of-voice rules, pages already published |
| Modules | In this caseData Foundation, Company Brain, Agents, Workspace |
| Who decides | In this caseThe editorial team on copy and publication; merchandising on product data; whoever is accountable for the brand on the tone-of-voice rules the agent applies |
| What we measure | In this caseThe time between the item's arrival and publication, pages corrected before going online and differences between one channel and another, against the starting point |
| AI Act risk class | In this caseThe case produces product copy and attributes and stays well away from the high-risk uses in Annex III most frequent in our industries, such as the creditworthiness of individuals. If the agent generated or altered images that could appear authentic, it would have to be disclosed that they are artificial; we set the classification with risk and compliance before development |
| Where to start | In this caseWith Deployment on a single category, when its master data is complete and the brand's rules are written down; otherwise the first job is on the data |
With incomplete master data the first project is on the data
If the same item has different codes in the ERP system, on the website and in store, or if suppliers send composition and measurements only on request, the agent would leave many fields empty and the editorial team would do the work twice. In that case we start from master data with Data Foundation, and the product pages come later. It is also worth waiting when the brand's rules exist only in a chief editor's head, because the agent would have no tone to follow.
The case comes from the Consumer page.
Let's start from a page corrected after publication
Taking a product page that was corrected when it was already online, we work out with your editorial team which source the wrong data came from.