Embedded Files
Publicis Groupe in-house app | Language & translation services - AI - Digital E-com content management
Role: UX Designer
Assigned to project: since 10.2024 - until now
Summary
Trancreation Team at Publicis needed a tool which will automate their work on preparing translations for digital e-com content. The solution to this issue was platform for desktop - ECCA.
Tool that we offer speeds up the workflow as it extracts texts from the source videos or images automatically and offers 3 types AI-powered translations.
In this project I created another variant of the existing functionality & designed from scratch the AI-powered translations option.
Even though some of the UX ideas were rejected as of the AI limitations, still, the final design was positively received by user representatives & the business side.
Achievements
The vaule I delivered to Publicis Groupe translated into higher quality of the B2B services, growth income & global recognition of my department within the organization.
I comprehend a complex system on a very short notice and deliver a high-quality service to my new client.
My strengths for this project
Previous hands-on experience with complex ecosystems full of dependencies on business, tech & process side.
Ability to adapt quickly to the new project conditions.
Details of the project
I joined the team while the product was in the mvp deployment finishing phase. The product itself was a complex eco-system dedicated to various Publicis teams and clients used for creating & managing digital content for e-coms.
One of the areas I developed & optimised was a tool for Transcreation Team at Publicis. This team is preparing translations for online ads, and so far texts from the videos or static images were copied manually into the excel document. Later on, the excel docs were shared with translators who were creating content in a relevant language.
Our product automates the whole process as texts are being extracted from the uploaded source files automatically. Additionally, there are variety of AI-powered translations offered, to speed up translator's work.
My tasks
Create a new variant of the functionality where user can manage translation process for static images.
Design an AI-powered translation functionality.
Challenges
Only 4 days possible for onboarding then being introduced to a Jira ticket with high priority.
The language & translations topic was a novelty to me, as well as designing for AI-powered solutions.
Work-in-progress
After revising the current state of the project I defined which options are not applicable to the new variant of the functionality and listed areas of the UI needed to be slightly re-designed to address a new variant.
In this task, my focus was on keeping a consistent experience with existing solutions as users were already using our product to work on videos, however it appeared that one area needed bigger changes.
The main change required a section where user can upload source files, as it was not prepared for a new case - several files upload. After presenting this case to my team we got a green light from PO to proceed with a bigger changes in the existing screens.
The most difficult part becomes designing AI-powered translation functionality for videos as at some point it appeared that AI technology doesn't have a capacity to provide a desired experience to a user.
The problematic case was about a situation where in one text box appear parts of two sentences. My initial idea was to match translated texts to the source text precisely, and show in the side-panel translation suggestions for both sentences.
Final designs
New variant of the functionality: static images + general optimization of the tool
In the final design users got a fully functional & optimised solution to manage translation process of static images. On top to that, general experience of the tool is improved thanks to the small tweaks to the layout, hierarchy, copy or functions.
UX Highlights
Main change change to the old UX - file upload & management
New design is ready to handle a multiple files upload & provides consistent experience over the tool. There is improved general layout of the section, also visual & functional side of the uploaded file has been optimised. Thanks to those changes I achieved a decluttered layout and transparent rows where user can easily recognize and preview the source files.
A tricky part becomes designing a loading state of the files - my 1st proposal was to use a skeleton loader for the list with files. Although, due to the technical constraints the final design contains a simple loader.
Another optimisation to the experience is at the entry points to the upload files function.
I simplified & improved the visibility of the file upload function in the page.
Modals where user can upload files got a new microcopy which better explains all elements in the UI & are prepared for a new requirement - multiple-files upload.
Old modal - for video only
New modals - for video & static images
Contextual information to guide user through the functionality
English Master is a section where is available a transcription from the source file. The section below it - Localizations - contains translations of text from a source file.
The client's requirement was to block those functionalities at some point in order to get more control over a user behaviour.
In the old concept there were no cues which would help a user to understand logic of the elements. In the new concept I improved the user experience by adding a contextual information to the elements.
New functionality: AI-powered translations
The new functionality allows the users to translate texts extracted from videos & static files automatically, what is more it offers three types of AI-powered translation that differ in the database used by AI. Even though the solution is not ideal due to AI limitations still it improves workflow of Transcreation team at Publicis.
UX Highlights
Designing in technical limitations & providing temporary solution
One of the AI-powered translations option uses a database which contains copy created for previous e-com campaigns. The idea is that AI matches 3 best copy suggestions to the current translation and translator working on this project can select one of those or edit text manually if needed.
The final design works well for a variant with static files, but when it comes to the video translation - here the experience is not optimal due to the AI limitations at the moment. Video translation is more complicated case as text sentences can start and end in two text-containers.
Currently, the only solution the AI technology could provide was displaying the whole translated sentence in the one text-container, what required from a user manual adaptation of the text. Positive information was that system can detect such cases, thus I proposed to display an information while text needs a further edition.
Contact
For business inquiries, please contact me on Linkedin
Google Sites
Report abuse