CASE STUDY
How Rotterdam Business School Piloted AI Formative Feedback Across 18 Final-Semester Project Teams
A full-cohort Eduface pilot across 18 final-semester project teams over two complete feedback rounds. 79% of students were positive or open to AI formative feedback, and the final evaluation concluded that doing nothing is not an option.
By Eduface · June 2026 · 10 min read
Rotterdam Business School is the business faculty of Hogeschool Rotterdam, one of the Netherlands' largest universities of applied sciences. In the final semester of their Business Administration programme, student teams write a complex, multi-disciplinary report as their capstone project. Providing rubric-aligned feedback to 18 project teams, each submitting reports of more than 20 pages, was beyond what available staff capacity could support at the required depth and speed. Eduface was piloted to fill that gap.
What did the Rotterdam Business School Eduface pilot find?
Rotterdam Business School ran a full-cohort pilot of Eduface across 18 final-semester project teams over two complete feedback rounds. 79% of students were positive or open to AI formative feedback. Students rated Round 2 feedback higher than Round 1, confirming that rubric refinement compounds in quality over time. The final evaluation concluded that doing nothing is not an option.
About Rotterdam Business School
Rotterdam Business School serves approximately 12,000 students across bachelor's and associate programmes in business and management. It is part of Hogeschool Rotterdam, which enrols more than 27,000 students across all programmes and reflects the international diversity of the city, with 180 nationalities represented on campus.
The school has established a deliberate strategy of hands-on engagement with AI in education. Faculty leadership has explicitly committed to active adoption rather than observation. The pilot evaluation itself concluded with the statement that doing nothing is not an option.
Key facts
• 27,000+ total students at Hogeschool Rotterdam
• 12,000+ students at Rotterdam Business School
• NVAO accredited
• LMS in use: Brightspace D2L
• Location: Rotterdam, Netherlands
The challenge
In the final semester of the Business Administration programme, student teams write a capstone project report. Interim feedback on drafts is critical to the quality of the final submission. The challenge was providing rubric-aligned feedback across 18 project teams simultaneously, on reports exceeding 20 pages each, with limited staff turnaround time.
Four specific pressures shaped the brief:
• Complex, multi-disciplinary documents: feedback had to span strategy, KPI frameworks, financial analysis, and argumentation coherence, not just writing quality.
• Feedback that maps to course criteria: students needed comments tied directly to rubric criteria so they could act on them. Generic observations are not enough at this level.
• 18 teams, limited turnaround time: manual interim feedback on draft reports was practically impossible within available staff capacity.
• Scalability without burnout: the solution had to scale across all teams without proportionally increasing lecturer workload.
How the pilot was structured
The pilot ran across a full final-semester cohort, covering two complete feedback rounds. Eduface was configured around the existing course rubric, integrated into the module workflow, and deployed across all 18 project teams.
1. Configure the rubric
Existing course rubric in Eduface
2. Submit and receive feedback
Per-criterion scores and inline annotations
3. Process and revise
Then a second full round
Two full feedback rounds completed. Round 2 was rated noticeably better than Round 1 by students.
A structured evaluation framework collected both individual questionnaire data (13 statements, full cohort) and structured group-level feedback from all 18 project teams after each round.
What the pilot delivered
79% of students were positive or open to AI formative feedback. This combines those who found the feedback directly useful with those who were neutral but open to its potential.
All 18 project teams completed structured group evaluations alongside individual questionnaires, providing a rich dataset across two full feedback rounds. Two full feedback cycles were completed, and students and lecturers both noted clear improvement in quality between Round 1 and Round 2, confirming that rubric refinement compounds in value over time.
Student sentiment breakdown (individual questionnaire, 13 statements):
Response category
Share of students
Found AI feedback useful
57%
Open and receptive (positive potential)
22%
Positive or open: combined
79%
Round improvement confirmed
Students consistently rated Round 2 feedback as noticeably better than Round 1. Rubric refinement after the first round translated directly into higher-quality feedback in the second. No changes to the underlying AI model were required.
An overwhelming majority of students also expressed that they wanted to continue using Eduface in their studies. In open comments, students specifically requested expansion to other courses, independent access outside class time, and deployment during their thesis and internship phases.
What the questionnaire revealed
Six themes emerged consistently across the 13-statement individual questionnaire and the structured group evaluations.
• Positive attitude toward AI across the board: students entered the pilot with openness and left with reinforced enthusiasm.
• Eduface seen as more focused than general AI tools: students who had used general-purpose tools noted that Eduface felt more objective and more relevant to their course.
• Peer feedback and AI feedback are complementary: students did not see AI feedback as replacing peer review. They experienced the two as distinct and useful in different ways.
• Strong appetite for autonomous access: students expressed a clear desire to access Eduface at any point in their drafting process, not only at prescribed moments.
• Feedback stimulates independent thinking: rather than providing ready-made answers, the formative approach prompted students to reflect and re-examine their own work.
• Appetite for use in other courses: both students and lecturers explicitly identified other contexts within Rotterdam Business School where Eduface should be deployed.
Six feedback categories students consistently praised
Each project team submitted structured group evaluations naming their strongest feedback areas. Six categories appeared consistently across all 18 teams.
Feedback category
Why students valued it
APA citation and reference checking
Specific, concrete, and directly correctable before submission
Substantiation of key claims
Identified where conclusions lacked evidence or argumentation
Coherence across report sections
Checked whether KPIs, analyses, and recommendations connected logically across the full document
Model and visualisation improvement
Concrete guidance on completing and better presenting matrices and frameworks
Stimulus for independent reflection
Prompted harder thinking rather than providing ready answers
Inline document annotations
Feedback anchored directly to specific passages removed ambiguity from revision
What this pilot proved: implications for scale
Four clear themes emerged from both student and lecturer evaluations.
1. Rubric quality determines feedback quality. The measurable improvement between Round 1 and Round 2 proved this directly. Targeted rubric refinement translated immediately into more relevant student feedback in the second round, with no changes to the AI model.
2. LMS integration strengthens adoption. Connecting Eduface to Brightspace lets students submit and receive feedback within their existing workflow. Integration turns AI feedback from an optional extra into core course infrastructure.
3. Didactic framing amplifies outcomes. When AI feedback was paired with structured group discussions and guided reflection tasks, learning impact increased. Students benefit most when feedback is embedded in an active learning workflow.
4. Long-form, complex documents are fully supported. The pilot confirmed Eduface's capacity to generate coherent, structured feedback across reports exceeding 20 pages, precisely where general AI tools tend to lose coherence.
"Doing nothing is not an option. Students expect AI in their education, and gaining experience is essential to learn to work differently."
Recommendation from the final pilot evaluation, Rotterdam Business School, 2026
What comes next: expanding across Rotterdam Business School
Lecturers and students both identified a clear set of next steps.
• Thesis and final-year dissertations: high-stakes, long-form capstone projects present the same challenge as the final-semester reports.
• Internship and placement reporting: internship documentation involves regular written reflection against professional competency frameworks.
• Earlier years of the programme: Year 1 and 2 students can build feedback literacy earlier, reducing foundational errors that appear in later work.
• Wider expansion across RBS programmes: faculty leadership noted there is no structural reason to limit AI feedback to Business Administration.
• Brightspace integration (LTI 1.3): native integration with the Hogeschool Rotterdam Brightspace environment, enabling autonomous student access.
• Student-initiated feedback at any stage: students expressed a strong desire for independent access at any point in their drafting process.
Frequently Asked Questions
Was the 79% positive response across all students or just those who found it useful?
The 79% figure combines two groups: 57% who found the AI feedback directly useful, and 22% who were neutral but open to its potential. It represents the proportion of students who were either positive about AI formative feedback or open to exploring it further. The remaining 21% expressed reservations or were not yet convinced.
Did the pilot use a control group?
The pilot ran as a full-cohort deployment rather than a controlled experiment with a comparison group. All 18 project teams received Eduface feedback in both rounds. The evaluation focused on student reception, feedback quality, and appetite for continued use rather than a comparative grade outcome study.
How long did setup take before the pilot could start?
The rubric configuration was completed before the first feedback round began. After the first round, targeted refinements were made based on initial student and lecturer feedback. This iterative refinement is what drove the improvement between Round 1 and Round 2.
Can Eduface handle reports this long?
Yes. The pilot specifically confirmed Eduface's capacity to generate coherent, structured feedback across reports exceeding 20 pages. The platform processes the full document rather than sampling sections, which is why students valued its ability to check coherence across report sections, something individual section-level feedback cannot do.
Is Eduface GDPR-compliant for Dutch institutions?
Yes. Eduface runs on proprietary GPU infrastructure located in the Netherlands. Student data stays in the EU and is not shared with external AI providers such as OpenAI. Dutch institutions can start a free pilot directly. Eduface is also an approved supplier on the Jisc/CHEST framework (UK) and the HEAnet framework (Ireland).
Ready to run a pilot at your institution?
Eduface is an approved supplier on the Jisc/CHEST framework (UK) and HEAnet framework (Ireland). Dutch institutions can start a free pilot directly. EU data hosting and GDPR compliance are standard.