Automated assessment tool

LEAN CANVAS BUSINESS PLAN

Double Semester ⋅ CMT313 Software Engineering ⋅ October - November 2023
CARDIFF UNIVERSITY
Group Project

The Brief:
As a team, plan, specify and prototype an automated assessment tool that helps university teaching staff run summative and formative assessments.

In this first phase, we turned the idea into a business case. We used a Lean Canvas to define the problem, the users and the value of the tool, and a risk register to identify what could go wrong before we committed to building anything.

The Problem

Running assessments at university scale is mostly manual work. For every assessment, teaching staff have to set the work, share it with students, track who has submitted, mark it and return feedback, often for several modules at once.

As cohorts grow, this doesn’t scale. Keeping track of submissions takes time, marking can vary between markers, and feedback often reaches students after it would have been most useful. Time that should go on teaching goes on administration instead.

Existing platforms handle parts of this process, but staff still move between systems and manual steps to get from setting an assessment to returning feedback. We saw room for one tool that covers the whole journey, designed around how teaching staff actually work. We used the Lean Canvas to test whether that idea could hold up as a business before writing any requirements.

My Contribution

This was a group project, and I owned four parts of the business case:

Competitor research. I looked at existing assessment tools to see what they already did well and where they left staff doing work by hand. This shaped where our tool would fit and what it needed to do differently.

Unique value proposition. I used the gaps from that research to define what would set our tool apart. Instead of competing on features, it would cover the whole assessment journey, from setting work to returning feedback, in one place.

Cost structure. I mapped what it would cost to build and run the tool, including development, hosting and ongoing maintenance. This helped the team check that the idea was realistic as a product, not just as a piece of software.

Legal risk assessment. I added the main legal risks to the team’s risk register: data protection under GDPR, student privacy, and who is responsible when an automated mark is wrong. These risks went on to shape our requirements. Role-based access, encryption and audit trails all appear in the security requirements we wrote in the next phase.

The Approach

We started with a Lean Canvas so we could focus on the core of the idea before anything else: the problem, the people it affects (teaching staff and students), and the smallest version of the tool that would solve it.

At the same time, we built a risk register covering technical, legal and operational risks. Identifying these early meant they became part of the design from the start, rather than problems to fix later.

The module also included a workshop with Red Hat on Agile methods and working in sprints. It changed how we approached the rest of the project, and we went on to build the tool in short, iterative cycles.

Lean Canvas for the Automated Assessment Tool, covering problem, solution, value proposition, customer segments, channels, revenue streams and cost structure.

LEAN CANVAS

Risk register for the Automated Assessment Tool, listing each risk with its impact, likelihood and mitigation.

RISK ASSESSMENT

Challenges

Finding something new to offer in a crowded market
Plenty of edtech tools already handle assessments, so defining a value proposition that stood out was harder than I expected. My first ideas were features that other tools already had. To get past this, I went back to the competitor research and focused on the specific pain points teaching staff had, rather than the features they were missing. That’s how we arrived at our position: one tool that covers the whole assessment journey.

The ethics of automated grading
Automated marking raised questions I hadn’t thought about before: whether it could be fair, whether it could be biased, and what rights students have when a machine is involved in their grade. Researching GDPR and academic integrity showed me that these weren’t side issues. They affected what the tool should be allowed to do, and they led directly to the security and access requirements we wrote in the next phase.

Key Takeaways

  • Good competitor research finds a gap, not just a list. Knowing what other tools do is only useful if it shows you something they don’t do that you can credibly claim.

  • Legal and ethical issues are design decisions. They decide what you can build, so they belong at the start of a project, not at the end.

  • The Lean Canvas makes vague ideas concrete. It forces you to answer the hard questions early, and I’ve used it since to scope my own personal projects.

Next Steps

With the business case in place, the next phase was to turn it into requirements for a minimum viable product:

  • A top-level use case diagram, built as a team, showing the main things the tool needs to do and who uses each feature.

  • User stories with acceptance criteria, written from the teaching team’s point of view, so each feature has a clear, testable definition of “done”.

  • Non-functional requirements covering security, scalability, reliability, usability and accessibility, many of them shaped by the legal risks identified in this phase.