Two memories, two vocabularies.
Product data and web content are translated by different people with different memories. Product names and technical terms drift apart.
Use case · Illustrative scenario
01 · MULTILINGUAL CONTENT
One memory for product data and web content — in six languages.
A manufacturer runs its main website on Ibexa DXP and four campaign and product microsites on Drupal. PIM Gate delivers product data to the DXP today and brings product texts and web content into one translation workflow — with CMS connectors for Ibexa and Drupal coming soon.
CMS translation connectors COMING SOON — early access open.
“The product is called one thing in the data sheet, another on the website — and a third on the campaign microsite.”
02 · SITUATION
Two translation worlds that should be one.
The manufacturer's main website runs on Ibexa DXP: product pages, application stories, landing pages, news. Four microsites for campaigns and product lines run on Drupal, built by an agency. Around 15,000 articles in the PIM and roughly 2,500 content items across the sites, all in six languages.
Product data texts are translated by one agency from PIM exports. Web content is translated by another, from Word files that editors export and paste back into Ibexa and Drupal language by language. Each agency keeps its own translation memory. Neither sees the other's terminology.
Product pages on the website take product data from a nightly full import out of the PIM. When a specification changes, the page shows the old value until the next import succeeds — and the translation follows weeks later.
| TWO TRANSLATION WORLDS · TODAY | Role | Status |
|---|---|---|
| PRODUCT DATA | PIM · XLSX EXPORT → AGENCY 1 | AGENCY 1 TM |
| DXP CONTENT | IBEXA DXP · WORD / COPY-PASTE → AGENCY 2 | AGENCY 2 TM |
| MICROSITES | DRUPAL · WORD / COPY-PASTE → AGENCY 2 | AGENCY 2 TM |
| TERMINOLOGY | SPREADSHEET · — | NOT SHARED |
03 · CHALLENGES
Where multilingual content breaks.
Copy and paste into the CMS.
Editors paste translations field by field into Ibexa and Drupal. Formatting, links and embedded elements break; the effort grows with every language.
Whole pages for one paragraph.
When a landing page changes, the whole page goes out for translation again — unchanged sentences included.
Review without context.
Country reviewers check segments in a spreadsheet and only see the page after it is live.
Product pages lag behind the PIM.
Nightly full imports into the DXP are heavy, fail silently and deliver translations late.
Different content models.
Ibexa content types and Drupal nodes with paragraphs structure content differently; one process has to respect both.
04 · THE PIM GATE SETUP
Two lanes, one memory.
Product data and web content keep their own flows, but share one translation memory and one termbase. Product data reaches Ibexa DXP through the PIM Gate API as field-level deltas — available today. The CMS connectors, coming soon, are designed to pull content items from Ibexa and nodes from Drupal, protect markup and placeholders, send new and changed segments through the same memory and review, and write the result back as a draft language version. Editors publish in their CMS, as before.
Diagram with two lanes. Lane one translates product data from the PIM through delta detection, translation memory, terminology, an engine per language and a review queue. Lane two, coming soon, pulls content from Ibexa DXP and Drupal, segments it safely, translates it with the same memory and terminology, offers review with in-context preview and writes it back as a draft language version. Both lanes share one translation memory and termbase. Product data is delivered to Ibexa DXP via REST API and delta sync.
Product pages get only what changed
Shared translation memory and termbase
05 · WORKFLOW
From source change to published translation.
- STEP 01
Product pages from the gate.
Ibexa DXP reads product data through the REST API and receives field-level deltas or webhooks when a specification changes. No more nightly full import.
LIVE - STEP 02
Only the delta.
New and changed product texts are matched against the shared memory and termbase, pre-translated by the engine you choose per language, and reviewed by country reviewers.
2026 - STEP 03
Content items and nodes, field by field.
The connector will pull new and changed content from Ibexa content items and Drupal nodes and paragraphs — no Word export.
COMING SOON - STEP 04
Protect markup and placeholders.
Rich text is segmented without breaking HTML, links, embedded elements or tokens. Unchanged sentences come straight from memory.
COMING SOON - STEP 05
See the page, not a spreadsheet.
Reviewers approve segments next to a preview of the page, with terminology hits highlighted.
COMING SOON - STEP 06
Editors keep the last word.
Approved translations return as a draft language version in Ibexa or Drupal. Editors check layout and publish in their CMS as usual.
COMING SOON
06 · MODULES & CONNECTORS
What this setup uses — and what is live today.
- LIVE
PIM connector · canonical schema · multi-language with fallback
Product data source
- LIVE
Versioned REST API /v1 · delta endpoint · signed webhooks
Product data to Ibexa DXP product pages
- 2026
Product-data translation (TMX/TBX, delta, engine per language, review queue, XLIFF packages)
Product texts in six languages
LLM choice per language (OpenAI, Anthropic, DeepL, Azure OpenAI, EU/local)
Pre-translation proposals
available with translation
- COMING SOON
CMS content translation connector · Ibexa DXP
Pull, segment, review in context, write back as draft
- COMING SOON
CMS content translation connector · Drupal
Same for the four microsites
- LIVE
Customer Portal (optional)
Dealers and installers access the same translated data
07 · EXPECTED EFFECTS
What typically changes.
Indicative
Two memories, drifting terms
One memory and one termbase for product data and web content.
Indicative
Nightly full import into the DXP
Product data as field-level deltas — about 95% smaller payloads than full exports.
Indicative
Whole pages re-translated
Only new and changed segments go to translation (with the CMS connectors).
Indicative
Copy and paste into Ibexa and Drupal
Translations written back as drafts (with the CMS connectors).
Indicative
Review in a spreadsheet
Review next to a page preview (with the CMS connectors).
Indicative
Machine translation as a black box
Engine chosen per language, every language reviewed by people.
Qualitative expectations for this scenario, not measured results. Effects marked "with the CMS connectors" depend on capabilities that are coming soon.
08 · TYPICAL CONFIGURATION
Version M for six languages, a test tenant, connectors in early access.
Marketing, localization, web editors and product data add up to 11–25 seats. Six languages exceed Version S's three, so Version M is the fit; PIM, Ibexa DXP and Drupal are three of M's five connectors. A test tenant lets the web team try write-back against a staging CMS before production. Indicative list price €12,750, EUR, annual prepayment, excl. VAT.
| Item | € / year |
|---|---|
| Version M — 11–25 seats, ≤ 100k SKUs, 8 languages, 5 connectors, AI credits per month | €10,200 |
| Stage/test tenant · half capacity (25% of Version M) | €2,550 |
| Indicative list price | €12,750 |
| CMS translation connectors Ibexa DXP / Drupal | coming soon — pricing with general availability |
09 · ROLLOUT
Start with product data now. Add content when the connectors arrive.
- 01
WEEKS 1–2 · SCOPE
WEEKS 1–2Memories, terms, content model. Collect both agencies' TMX files and the terminology list, agree on one termbase per language, map Ibexa content types and Drupal paragraph types that need translation.
- 02
WEEKS 3–8 · PRODUCT DATA
WEEKS 3–8API and delta into Ibexa. Product pages switch from the nightly import to the PIM Gate API with delta sync; product-data translation with the merged memory and review per language.
- 03
EARLY ACCESS · IBEXA DXP
EARLY ACCESSContent translation on the main site. Connector on the staging site first, then production; one content type at a time, starting with landing pages.
- 04
EARLY ACCESS · DRUPAL
EARLY ACCESSMicrosites. The four Drupal microsites follow with the same memory, termbase and reviewers.
Indicative plan. Phases 3 and 4 depend on early access to the CMS connectors, which are coming soon.
10 · NEXT STEP
Recognize your two translation worlds? Book a scoping call.
Bring one landing page and its product data. We show how both would run through one memory — and what you can start with today.
COMING SOONCOMING SOON
Early access: Ibexa DXP and Drupal connectors
- Test the connector on your staging site first
- Help shape segmentation rules for your content types
- Direct line to the product team
// EARLY ACCESS
Join early access
11 · Keep reading
More on this scenario
2026
Translation & localization
Translate only what changed — in your terminology, reviewed by your people.
LIVE
API, feeds & delivery
One versioned API. Every delivery pattern. Only the changes.
Integrations
Every system PIM Gate connects to today — sources, consumers, translation engines and standards — with method and honest status.