MY ROLE
THE TEAM
THE TIMELINE
THE PRODUCT
THE PROBLEM
THE OUTCOMES
77%
CSAT from students
130+
Teachers choosing the feature
2500+
Students using the feature
Transforming Student Feedback in Real-Time
This conversational coach gives real-time, AI-powered feedback to students on their peer reviews. It provides suggestions for improvement tailored to student language use and content of the review. Although having AI as a co-pilot can be daunting, students love it! This is demonstrated by the remarkable 77% customer satisfaction score.
This feature has the highest CSAT recorded in the product!
Where it lives
Feedbackfruits is a dynamic feedback tools provider nestled within university Learning Management Systems. With different configurable activities, it empowers teachers to craft impactful learning experiences, helping them save time in and outside the classroom.

Cracking the code of constructive feedback
Feedback giving is one of the most important skills for the workforce but teachers don’t have time to provide guided, timely feedback to students.
this results in students giving too short, overly positive or negative feedback that is not actionable.*
*insight from interviews with instructors from top level universities teaching a class of 50+ students.
Modernizing feedback coaching
Within our company, there was a previous effort to have an “Automated feedback coach” but unfortunately, it was outdated both from a design and technical perspective.
Essentially, the homegrown model provided eight possible suggestions to students' reviews, which didn’t scale and had an erratic performance.

The team approach
I was part of the Analytics and Insights team and together with the PM we analyzed the existing feature and came up with requirements for the redesign. For a two month period I closely collaborated with the developers and machine learning engineer to decide on the best solution to the problem. While they investigated on how we could leverage generative AI for the coach and iterated on the prompt, I focused on the experience that the students would have with it.
To know more about the technical side, click here

Upholding transparency, empowering choice
This involved several back and forth with our legal team to ensure that the design:
Reflect that AI is being used
Give the chance to turn the feature off
Include a link to an article about transparency with AI
Crafting a seamless user experience
My main job was making sure that the redesign would seamlessly integrate with the existing product while complying with the legal AI-regulations for all regions and keeping the focus on the prime action for the user: giving feedback to their peers.
Decision #1
I decided to use an existing pattern in the product that would convey the idea of a discussion between peers. I gave our coach a profile picture next to their suggestions, achieving a conversational effect.
Decision #2
With minor adjustments to the alignment and sizing of the fonts, I decided to keep the ‘Automated Feedback Coach’ title and add a discreet help icon to disclose information about what the feature does and how to disable it. It was important to ensure that the suggestions are coming from AI and that the student is in control at all times.
Decision #3
Since the transparency note prepared with the legal team didn't need to be displayed constantly, I chose to present it as a link only at the initial stage
Designing for every state
Part of the requirements was to design for all the possible states of the coach, this included and initial state with a presentation, a "thinking" state while suggestions were being processed, the actual feedback from the coach and an error state.
Designing for scalability
I ensured the design would fit various placements, such as sidebars, making it a flexible element in the product. Additionally, I considered language options, allowing students to receive feedback in their preferred language.


Making designs inclusive for all students
Last but not least, I needed to make sure the designs were accessible to our students. With the iterative efforts to be compliant with the WACG level 2.1, I annotated the designs with the relevant labels to make the elements tabbable and available to screen readers.
Driving success, one feedback at a time
These achievements highlight the true impact of our efforts!

Instructor
@ one of the world's top learning institutions
The coach was really straightforward to set up. And none of the students had any trouble using it, which is pretty big. […] it was helping them give more detailed feedback and gave them good tips on what to improve on and assisting giving detailed responses about specific things. I would rate the quality with: 8/10 (missing 2 points due to low sample size), uniqueness: 10/10 and would recommend to colleague: Oh, 10. For sure. 10
Final reflection
LOREM IPSUM
LOREM.
IPSUM
On top of that, we saw amazing adoption of the feature from the instructors’ side, with 346 activities performed instead of the 15 we planned for.








