Choosing a localization platform on technical depth, not feature checkboxes.
- 01NCR needed content localized across global web properties, and the incumbent vendor wasn’t fitting the capability or the workflow. The replatforming had made translation a hard dependency.
- 02This wasn’t a plugin you drop into a CMS. How translated content was stored, reviewed, and published across international properties was something we had to design, not configure.
- 03I ran a formal RFP across Transperfect, Weglot, and Lokalise, selected Weglot on the depth of its scoping, and co-built a novel WYSIWYG architecture with IT — 49% under budget.
A dependency the whole migration was waiting on.
NCR operated in multiple countries and needed content localized across its global web properties. The existing vendor, translation.com, wasn’t cutting it — in capability or in workflow fit. And the broader Cardtronics migration and AEM-to-Webflow replatforming had created a hard dependency: the translation infrastructure had to be in place before the full migration could complete. Without a functioning localization layer, you couldn’t publish localized content reliably across NCR.com.
The technical challenge was real. This wasn’t a matter of finding a translation plugin and dropping it into the CMS. How translated content was stored, managed, reviewed, and published across international properties was something we’d have to design, not just configure.
There was a cost-discipline question underneath it. Localization infrastructure had a $155,000 approved budget line. That was a meaningful number — and I wasn’t interested in spending to the ceiling just because the ceiling was there.
Three vendors, and a deciding factor that wasn’t price.
I ran a formal RFP against three vendors: Transperfect, Weglot, and Lokalise. The evaluation covered translation quality, AI-assisted workflow, integration fit with Webflow, pricing structure, and the technical depth of each vendor’s approach to implementation. With Lokalise we went far enough in diligence to execute an NDA — not something you do unless you’re seriously weighing what a partnership would involve.
Transperfect was the largest player, with enterprise-grade positioning, but its pricing and integration model weren’t the right fit for what NCR needed. Lokalise was a strong contender — capable platform, good API coverage — but once we got into the specifics of the architecture, Weglot’s approach aligned more closely with how we needed to handle content rendering and editorial workflow.
The deciding factor wasn’t price or a feature checklist. Weglot brought genuine technical depth to the scoping conversations, and that mattered because what we were building wasn’t an off-the-shelf integration. The architecture would be co-developed with our IT team and the vendor — a novel setup designed to make machine-translated content publishable and editable at scale within NCR’s content workflow.
“The deciding factor wasn’t price or a feature checklist — it was which vendor brought real technical depth to the scoping.”
A purpose-built WYSIWYG localization layer.
We selected Weglot and built a localization platform that used machine translation at scale with a WYSIWYG editing workflow layered on top. The WYSIWYG capability mattered operationally: editors could see exactly how translated content would render before it went live, which cut QA overhead substantially against the previous process, where editors were effectively reviewing content blind.
The architecture — the system tying together translation management, editorial review, and publishing controls across NCR’s international properties — was purpose-built. It wasn’t a standard Weglot deployment; it took close collaboration between our IT team and the vendor to get the integration right inside NCR’s Webflow environment. That work became the foundation for localized publishing across the global site.
A cleared dependency and a platform that kept working.
This was infrastructure, not a campaign — the honest outcomes are the disciplined procurement, the architecture decision, and the dependency it cleared. The one hard number worth claiming is the budget: 49% under, on a single line.