Product catalogue management: complete, accurate product pages published on time across every channel, with no re-keying.
The catalogue manager holds the product data on which the website, the marketplaces, the stores and the retailers depend: attributes, descriptions, visuals, variants, translations, regulatory information. Data arrives from the PLM, the ERP, suppliers and the photo studio, each in its own format, and is still copied into the PIM by hand a great deal. Automation collects, checks and writes; the team keeps accuracy and tone.
The role
Between product, the photo studio, e-commerce, marketplaces and suppliers.
The catalogue manager turns a reference created in the ERP or the PLM into a sellable product page: completing attributes, linking visuals, having copy written and translated, checking mandatory statements and publishing on each channel. They set the rules of the reference data (taxonomy, mandatory attributes by family, formats) and enforce their quality. In a mid-sized company, it is often a small team that absorbs the arrival of each collection or each supplier catalogue.
Day to day
A typical day.
- MorningCollect new references from the ERP or the PLM and product files received from suppliers.
- MorningComplete missing attributes: material, composition, dimensions, colour, care, use.
- MiddayLink the packshots and on-model visuals received from the studio, and check their order and format.
- AfternoonHave descriptions written, reviewed and translated, then validate the product pages ready to publish.
- AfternoonDeal with marketplace rejections and corrections requested by e-commerce or customer service.
- End of weekTrack catalogue completeness by family and by channel, and chase contributors who are behind.
The objectives
What the role is measured on.
- Catalogue completeness rate
- Time to put new products online
- Product pages published without errors
- Marketplace rejections
- Translation coverage
- Compliance of mandatory statements
Indicators are named, never quantified: levels depend on each company, and the diagnostic measures them.
Day to day
The problems that come up most.
Data arriving from everywhere
PLM, ERP, supplier spreadsheets, e-mails from product: each source has its own format, and everything is copied into the PIM.
Incomplete product pages published
A missing attribute keeps the product out of the website’s filters or gets the offer rejected by a marketplace.
Visuals to link one by one
Packshots arrive with free-form file names; they have to be found, renamed, ordered and attached to each variant.
Copy to write in bulk
With each collection or supplier catalogue, hundreds of descriptions to write and translate, at short notice and in the brand’s tone of voice.
Mandatory statements to keep up to date
Composition, care, origin, INCI list, environmental information: an error exposes the brand, and the rules change.
A catalogue that clutters up
Discontinued products still online, duplicates, orphan variants: the catalogue ages without anyone cleaning it.
By sector
One role, several realities.
- Fashion and luxury
Editorial product pages with carefully crafted copy, materials and craftsmanship described precisely, many polished visuals, permanent products and collection pieces, translations into many languages.
- Retail
Collections renewed every season, size and colour variants, size guides, mandatory composition and care, volumes of product pages to produce in little time.
- Cosmetics and beauty
INCI list, precautions for use and claims to check, sizes and shade variants, gift sets made up of several products, product pages to adapt by country.
- B2B distribution and wholesale
Supplier catalogues to integrate and standardise, tens of thousands of references, heterogeneous technical attributes, duplicates and EAN codes to make reliable.
- Industry
Technical catalogue with specifications, drawings and data sheets, spare parts linked to equipment, compliance documents and bills of materials to publish for customers.
What changes
Automate the task, keep the judgement.
What gets automated
- Creation of product records in the PIM as soon as a reference arrives in the ERP or the PLM, with the attributes already known carried over automatically.
- Reading of supplier files and data sheets by an AI agent, which extracts and standardises the attributes.
- Writing and translation of descriptions by an AI agent, from the attributes and according to the brand’s tone of voice guide.
- Linking of DAM visuals to products and variants, with checks on format, background and order of views.
- Completeness checks by family and by channel, with publication of incomplete product pages blocked.
What stays human
- The rules of the reference data: taxonomy, mandatory attributes, tone of voice guide.
- Review of sensitive copy, claims and regulatory statements.
- The decision to publish, withdraw or merge products.
Use cases
How to automate, case by case.
Each case starts from a question leadership teams ask. The starting point, what gets automated, the tools involved, and what stays in the teams’ hands.
How do you automate putting products online?
- Today
- A reference created in the ERP or the PLM is copied by hand into the PIM, completed attribute by attribute, then published on the website once the visuals and copy have arrived. Nobody knows exactly what each product is missing, and going online slips.
- What gets automated
- Each new reference automatically creates its record in the PIM, with the data already known in the ERP or the PLM: name, family, price, composition, dimensions, variants. The record follows a workflow of steps (attributes, visuals, copy, translations, validation) and each contributor sees what is left to do. When it is complete for a channel, it is published on the planned date.
- What stays human
- The catalogue manager validates product pages and decides on early or delayed publication.
- ERP
- PLM
- PIM
- E-commerce platform
How do you automate product pages with AI?
- Today
- With each collection, the team or an agency writes hundreds of descriptions from data sheets and samples. The tone varies from one writer to another, and product information is missing or wrong in the copy.
- What gets automated
- An AI agent writes each description from the PIM attributes and a tone of voice guide written by the brand: hook, description, key points, care or usage advice, adapted to the length each channel expects. It only uses the attributes provided and flags those that are missing. The copy lands in the PIM awaiting review.
- What stays human
- The team reviews the copy, corrects the tone and validates claims before publication.
- PIM
- AI writing agent
- DAM
How do you automate catalogue completeness checks?
- Today
- The completeness rate is calculated now and then through an export and a spreadsheet. Incomplete product pages go online, missing from the website’s filters, or are rejected by marketplaces without anyone knowing how many.
- What gets automated
- For each family and each channel, the list of mandatory attributes is written once. Completeness is recalculated continuously, by product, family, channel and contributor, and publication of a product page incomplete for a channel is blocked. A dashboard shows the attributes most often missing and chases the contributors concerned.
- What stays human
- The catalogue manager sets the mandatory attributes and grants publication exceptions.
- PIM
- Reporting tool (BI)
- Messaging
How do you automate the translation of product pages?
- Today
- Product pages are sent to a translator as a spreadsheet export, language by language, then re-imported into the PIM. Each change to a source text means sending everything again, and some languages lag behind on the website.
- What gets automated
- As soon as a product page is validated in the source language, its copy is translated into each language by an AI agent, with a brand glossary (materials, line names, terms not to translate). A change to the source text triggers an update of only the translations concerned. Sensitive languages and families go through human review before publication.
- What stays human
- Native reviewers validate the copy for key markets and enrich the glossary.
- PIM
- Translation tool
- AI writing agent
How do you automate linking packshots to products in the DAM?
- Today
- The studio delivers packshots and on-model visuals with free-form file names. The team finds them, renames them, sorts them by product and colour, chooses the order of views and checks the format, before attaching them to each product page.
- What gets automated
- Visuals uploaded to the DAM are linked automatically to the product and the variant, from the file name, a code read on the image or product recognition by an AI agent. Their format, resolution and background are checked, and the order of views follows the family rule (front, back, detail, on model). Products without a complete set of visuals are listed.
- What stays human
- The team validates uncertain links and chooses the main visual for hero products.
- DAM
- PIM
- AI image recognition agent
How do you automate enriching product attributes from supplier data sheets?
- Today
- Suppliers send their data in spreadsheets or PDF data sheets, each with its own columns and units. The team copies the useful attributes into the PIM and converts values by hand.
- What gets automated
- An AI agent reads each file or data sheet received, extracts the attributes, converts units and matches values to the PIM lists (colours, materials, standards). Validated mappings are remembered by supplier, and doubtful values are flagged rather than guessed.
- What stays human
- The team validates flagged values and asks the supplier for missing information.
- PIM
- AI document reading agent
- Supplier portal
How do you automate the management of variants and EAN codes?
- Today
- Sizes, colours, volumes and shades are created in the ERP then recreated in the PIM. Duplicate EAN codes, variants linked to the wrong product or with no visual appear on the website and block feeds to marketplaces.
- What gets automated
- Variants are taken from the ERP with their parent product, and each EAN code is checked: format, check digit, uniqueness in the catalogue. Orphan variants, duplicates and variants with no visual or no specific attribute are flagged before publication.
- What stays human
- The catalogue manager corrects the links and settles duplicates with product and the supply chain.
- ERP
- PIM
- DAM
How do you automate updating mandatory statements on product pages?
- Today
- Composition, care, country of origin, INCI list, environmental information: statements are entered product page by product page from product or supplier documents. When a rule or a composition changes, all the product pages concerned have to be found.
- What gets automated
- Mandatory statements are linked to the source that holds them (PLM, data sheet, regulatory database) and carried over automatically into the PIM. Each family has its list of required statements by country; a product page missing one is not published. A change of composition or rule flags every product page to review.
- What stays human
- Quality or regulatory affairs validate the statements and their interpretation, and the catalogue team applies the corrections.
- PLM
- PIM
- AI document reading agent
How do you automate catalogue migration when changing website platform or PIM?
- Today
- When changing e-commerce platform or PIM, thousands of product pages have to be migrated. Duplicates, badly filled attributes and old copy are carried over as they are, for lack of time to clean them.
- What gets automated
- Before migration, an automatic assessment measures the quality of the existing catalogue: duplicates, empty or inconsistent attributes, values outside the lists, missing visuals, discontinued products. Repetitive corrections are made by rules, an AI agent standardises free-text values, and the migration only loads product pages that pass the checks of the new model.
- What stays human
- The team decides on the target data model and the products not to migrate.
- PIM
- E-commerce platform
- DAM
- Spreadsheet
How do you automate the end of life of products on the website?
- Today
- Discontinued products stay online, sold out, or disappear overnight leaving pages in error and broken links in search engines. Nobody clearly decides when to unpublish.
- What gets automated
- Each product carries a life-cycle status, taken from the ERP or the PLM. When a product is discontinued and out of stock, its page is unpublished according to a set rule, with a redirect to the category or the replacement product, and the record is archived in the PIM with its visuals.
- What stays human
- The team chooses the replacement product and keeps online the pages that must remain accessible.
- ERP
- PIM
- E-commerce platform
Questions from the teams
What teams look to improve.
- How do you automate product catalogue management?
- By first fixing the data model: families, mandatory attributes by family and by channel, formats and lists of values. The sources (ERP, PLM, suppliers, DAM) then feed the PIM automatically, checks block incomplete product pages and AI takes on writing and translation, under review.
- What is a catalogue completeness rate?
- The share of expected attributes that are actually filled in, calculated by product, by family and by channel, since each channel has its own requirements. It is the indicator that says whether a product page is ready to publish.
- Can AI write product descriptions?
- Yes, from reliable attributes and a written tone of voice guide: it produces consistent descriptions, adapted to each channel and translated. It must not invent features; everything it writes comes from the attributes, and the team reviews claims and sensitive products.
- What is the difference between a PIM and a DAM?
- The PIM holds product data (attributes, copy, variants, translations), the DAM the files (packshots, on-model visuals, videos, documents). Automation links them: each DAM visual is attached to the right product and the right variant in the PIM.
- How do you integrate supplier catalogues automatically?
- By reading each file or data sheet received, whatever its format, and matching its columns and values to the PIM data model. An AI agent suggests the mappings and standardises the values; newly validated mappings are reused for later sends.
- Do you need a PIM to automate the catalogue?
- A single reference base is essential; it is not necessarily a PIM at first. A PIM becomes useful when channels, languages and families multiply, and automation can prepare its introduction by cleaning the existing data.
- How do you make EAN codes and variants reliable?
- By automatically checking the validity and uniqueness of each code, and verifying that each variant (size, colour, volume) is linked to a parent product with its own attributes. Duplicates and orphan variants are flagged before publication.
- Where do you start when automating the catalogue?
- With creating product records from the ERP or the PLM and with completeness checks: these are what delay going online the most. AI writing comes next, once the attributes are reliable.
02 — Getting started
Describe what’s holding your teams back.
A first conversation to qualify your need. Then a scoped diagnostic: the scope, the solution, the estimate.
Discuss your projectReply within 48 business hours.