Product Truth · SEO and GEO · Catalogues at scale

AI product descriptions for ecommerce at scale

We create and localise product descriptions at scale using validated data, brand voice, market-specific SEO and GEO criteria, and quality controls that prevent invented attributes.

Specialists in the destination marketWorkflow matched to risk and volumeQA linguistic and editorial
MULTILINGUAL CONTENTSOURCE → MARKET
01 · DiagnoseGoal, audience, channel and marketbrief · context · terminology
02 · AdaptAI product contentlanguage · intent · brand voice
03 · ValidateReviewed and ready to publishQA · consistency · delivery
LANGUAGE EXPERTISEVALIDATED CONTENT ✓
01 / AI product content

AI can accelerate catalogue content when it knows the product truth

A useful product page combines accurate information, commercial hierarchy, natural language and channel fit. We convert dispersed data into Product Truth, define category-specific content contracts and generate content ready for review, approval and publishing in each market.

We structure attributes, benefits, restrictions and brand rules before generation. Missing or unvalidated product facts are never invented to complete a description.

Tell us about your project →
01

Product Truth

Verified attributes, benefits, proof, restrictions and authorised source data.

02

Category voice

Tone, structure, length, claims and vocabulary adapted to the product.

03

Product SEO and GEO

Local queries, entities, questions and comparison information.

04

Multichannel publishing

Variants for ecommerce, marketplaces, feeds, PIM systems and campaigns.

02 / How we work

From scattered product data to publishable market variants

Calibrated prompts, terminology, SEO/GEO criteria, human oversight and category-specific QA support consistent publication across catalogues and channels.

  1. 01

    Catalogue and channel diagnosis

    Audit source data and representative product samples.

    Catalogue mapped
  2. 02

    Product Truth validation

    Build Product Truth and identify missing or non-inferable fields.

    Product Truth validated
  3. 03

    Content contract by category

    Define content contracts, voice, claims, terminology and channel rules.

    Content contract approved
  4. 04

    Pilot generation and localisation

    Generate and localise a representative pilot for human validation.

    Pilot accepted
  5. 05

    Quality control and production

    Run factual, linguistic and format QA before publishing or export.

    Catalogue produced
  6. 06

    Publishing and change control

    Learn from corrections and update the system before scaling further.

    Publishing verified
03 / Scope

What the system must know before it writes a single product page

01

Product Truth and category rules

We can create product pages from structured or dispersed sources, localise existing catalogues, generate channel variants and connect approval to publishing.

02

Generation and localisation

The system may include a Product Truth schema, category templates, prompts, glossaries, automated controls, validated samples and risk-based review criteria for Shopify, PrestaShop, marketplaces, PIM systems and custom feeds.

04 / Evidence

AI product content · Ontranslation

238 product pages in two months—without giving the model permission to guess

For Chachifarma, OnCatalog supported the production of 238 product pages in two months using validated information and explicit content rules. The principle is consistent across categories and markets: govern data and voice first, then generate, localise, check and publish.

Discuss the project →
05 / Choosing the right scope

When to choose AI product content

Choose this service when the central problem is creating or localising many product pages from structured or dispersed data. Governed AI translation is designed for approved source copy; ecommerce localisation covers the full buying journey; OnCatalog becomes relevant when Product Truth, approvals and publishing must be governed across large catalogues, channels and markets.

06 / Delivery criteria

A catalogue system that protects product truth

Catalogue scale creates operational questions beyond writing. We define how variants inherit information, which fields may be reused, how discontinued claims are removed and what happens when source data changes after approval. Search optimisation is applied at category and market level, while factual controls remain tied to each product. The resulting system supports new launches and updates without treating every SKU as an isolated copywriting assignment.

01

Source-data audit

We identify the fields that can support a claim, the attributes that are missing and the conflicts between feeds, supplier sheets and existing product pages. Unresolved gaps remain visible instead of being completed by inference.

02

Category-specific contracts

A skincare product, a garment and a technical accessory need different evidence and hierarchy. Content contracts define mandatory fields, benefit logic, tone, length, restrictions and questions for each category and destination channel.

03

Representative pilots

Samples include easy, complex and edge-case products. Human reviewers validate factual accuracy, brand voice, search language and format before the workflow is approved for a larger catalogue.

04

Publishing and change control

Outputs can be mapped to Shopify, PrestaShop, PIM or marketplace fields. Versioning and validation rules help teams update descriptions when product data changes without recreating the entire catalogue manually.

07 / Frequently asked questions

Questions catalogue teams ask before generating product content

Does AI invent product information?

The workflow only permits attributes present in authorised or validated sources, with controls for deviations.

Can you preserve our brand voice?

Yes. We translate it into operational instructions and examples by category and market.

Does this work for Shopify and marketplaces?

Yes. We generate variants for the fields, limits and customer needs of each channel.

Is the content optimised for search?

Where relevant, we incorporate local demand, entities, headings, metadata and useful questions without sacrificing accuracy.

How much human review is required?

It depends on data quality, product risk and market exposure. We define proportionate oversight.

Can you work with incomplete product data?

Yes, but we first identify gaps and decide which fields can be published, which need enrichment and which must never be inferred.

Do different categories use different content rules?

Yes. Structure, benefits, vocabulary, claims and comparison needs can change by category, product risk and buying decision.

Can you publish the content directly?

Yes. Delivery can use files or connect with CMS, PIM, marketplaces and approval workflows, depending on the client technology.

Next step

Connect product content with the wider ecommerce system

Tell us about your project

What do you want your content to achieve?

Tell us about the scope, languages and timeline. We’ll respond with a clear proposal.

    Or send us your docs for a quote

    I've read and accept the Disclaimer and Privacy policy

    I consent to receive the information Ontranslation & Interpretation, SLU considers appropriate to send me by email (you can unsubscribe at any time).

    Privacy Overview
    Ontranslation

    This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.

    Strictly Necessary Cookies

    Strictly Necessary Cookie should be enabled at all times so that we can save your preferences for cookie settings.

    Analytics

    This website uses Google Analytics to collect anonymous information such as the number of visitors to the site, and the most popular pages.

    Keeping this cookie enabled helps us to improve our website.