Finding the words that slowed guests down
Reading a mobile usability study of the elgouna.com booking flow for the problems caused by language, then localizing the site’s 1,045 strings across English, German and Arabic.
Overview
elgouna.com is where guests book stays across El Gouna’s hotels. In April 2026, a mobile usability study tested its booking flow with real participants.
I went through the findings with one question in mind: which of these problems were caused by words?
Context
Hotel booking is a sequence of small decisions. Where to stay, which dates, which room, which rate. Each decision depends on a label, and on a phone there’s no room to explain a label that isn’t clear.
The site also runs in three languages, English, German and Arabic, so every fix has to work three times.
The problem
Two kinds of content issue came through clearly in the study.
Naming. The same place, Marina Beach, wasn’t named consistently across the flow. When a name changes between screens, guests reasonably wonder whether they’re still looking at the same place.
Labels. Some filter and room-rate labels weren’t clear to participants, which made it harder to compare options and choose.
If two screens use two names, guests assume two places. Naming is a usability problem before it’s a style one.
My role
- Analyzed the study findings and separated content issues from layout and interaction issues.
- Wrote a structured report: each issue, the participant evidence behind it, and a specific recommendation.
- Localized the website’s copy across English, German and Arabic.
The approach
Every recommendation is tied to something a participant said or did. That keeps the conversation with the team on evidence rather than taste, and makes it easier to prioritize.
For localization, the site’s 1,045 translation keys lived in JSON files that only developers could comfortably read. I converted them into a spreadsheet with English, German and Arabic side by side, so reviewers could check every string in context. The same view showed which Arabic strings had never been translated, and I prepared those for correction.
Where it stands
The findings and recommendations were delivered as a report grounded in participant evidence. The trilingual strings were reviewed, with untranslated Arabic copy identified and prepared for correction.