Context
Smart Create is a class creation tool I designed to help schools set up classes and students during rollover on 3 Essentials, a unified teaching and management dashboard for Reading Eggs and Mathletics.
What is rollover?
Rollover is an multi-step, annual process that prepares a school for the year ahead. Since user accounts on 3 Essentials reflect real-world school data, it must provide the right tools to support this process.
Problem
Teachers found it difficult to create classes during rollover, leading to heavy reliance on our support teams and greater operational costs for the business.
Solution
A two-step class creation tool that allows teachers to bulk paste their student list from a spreadsheet and match them to existing accounts on our platform.
Timeline
Feb - Apr 2025
Team
Myself
Head of Web
Support Lead
Senior Developer x1
My contributions
Concept
System flow
Wireframes
Hi-fi designs
Design system and graphic assets
Outcomes
67% average reduction in class creation time for support teams
This allowed our teams resolve more tickets ahead of deadlines and ease the strain on support.
Increased growth in user-created classes
By 2026, a significant portion of classes were created by teachers rather than our support teams.
Avoided staffing costs for EMEA rollovers
By easing pressure on customer support, we prevented the need for casual staff and directly supported the company’s FY25 operational cost reduction targets.
Scroll down for more or jump to a section
01 / BACKGROUND
Over 6000 schools needed our help
In January 2025, we were preparing for the first ever APAC rollover on the 3 Essentials dashboard. The business had to hire and onboard casual staff, and pull people from other departments to assist in rollover (myself included).
We were…
A few weeks behind schedule
Limited by our class creation tool
Band-aiding the problem
We spent the majority of our time creating classes and assigning the correct student accounts based off spreadsheets that each school provided us with.
Our goal
The Head of Web approached me to design a new class creation tool to improve the rollover experience, promote self-service, and empower our support teams. We aimed to build and test an MVP before releasing the feature to schools ahead of the upcoming EMEA rollover.
02 / INITIAL DISCUSSIONS
What if users could paste their students?
We initially explored a concept based on an internal tool. Through rapid wireframing and bouncing ideas back and forth, we formed an early concept and hypothesis around how the experience could be improved.
The concept
1. Fill out basic class details in a form
2. Copy and paste a student list from a spreadsheet into the form
3. System matches each entry to the correct student account and assigns them to the class
Early insights
Assigning students was the most time-consuming and challenging part of creating a class.
We saw an opportunity to leverage the fact that schools use a spreadsheet as their source of truth for classes, teachers and students prior to rollover.
We hypothesised that shifting the user interaction from assigning students one by one to a bulk paste from a spreadsheet could reduce human error and speed up the class creation process.
03 / PROBLEM FRAMING
There were 3 common pain points
While the concept felt promising, we hadn't yet validated whether we were solving the right problems. I spoke with Sales and Customer Support to identify common user problems and validate our assumptions.
Target users
Teachers and coordinators (senior teachers who manage the school's 3 Essentials account) are the primary users of the dashboard. We also had to account for our support staff, as they log in and assist schools directly.
Common pain points
1
Assigning the correct students was difficult
2
Creating classes was too slow
3
The entire process was overwhelming
We didn't have the resources for formal user research, so we relied on second-hand feedback and leads. Since the same three issues kept surfacing, we were confident to move forward with them as our focus areas.
04 / UNDERSTANDING THE PROBLEM
Why were users having these problems?
Meeting with the Support Lead helped identify the root causes of these issues, drawing on anecdotal evidence and observations from screen recordings.
Why was assigning the correct students difficult?
Students with identical names and similar year levels were hard to distinguish.
Nicknames, initials, mispellings, or strange naming conventions made it hard to identify students by their real name.
Teachers would create a duplicate account or assign the wrong account by accident.
Why was creating classes slow?
It took a long time to assign students one by one as most classes had 20-30 students.
Most schools had over 10 classes to create, which made the process tedious.
New students had to be created on a separate page before they could be assigned to a class.
A lack of filters made finding students hard.
Why was the process overwhelming?
Users had to navigate between multiple areas of the dashboard to complete tasks that were previously handled in a single, guided flow.
Users were still adapting to the new interface.
My learnings validated that our concept and hypothesis addressed the right problems, but to drive teacher engagement, the concept needed more clarity of information and better support for secondary tasks.
05 / SYSTEM FLOW
Student matching was our main priority
I collaborated with the team to prioritise the MVP's must-have features. We realised that student matching was a top priority to solve first, so I created a system flow to guide discussions and define its logic.
Insights
We can use factors like name similarity, year level, and account activity to power the matching algorithm.
There were three possible outcomes from student matching that the user would have to understand and review.
The matching algorithm wouldn't be perfect initially, so fallback options for users were necessary.
06 / WIREFRAMES
Sticking to a full-page layout
I created lo-fi wireframes to visualise how the feature would look, how it adapts to different class sizes, and how it would scale with future updates.
Lo-fi wireframes of Step 1 and 2
I proposed a full-page layout to give users enough space to review each student match with all options and relevant information visible. After reviewing with the team, this was the direction we decided to go with.
07 / INITIAL HI-FI DESIGNS
Designing for clarity and ease of use
During the design phase, I met regularly with the team to review designs and assess technical feasability. We aimed to keep build cost low while meeting user needs, and after a few iterations, we had an MVP ready for testing.
A simple two-step workflow
Users found it overwhelming to navigate multiple areas when setting up classes, so I consolidated their tasks into a single workflow.
Supporting secondary tasks such as creating new accounts meant that users could complete all tasks without leaving the flow.
From their source of truth, to our platform
Users found it time-consuming to assign students manually, so I designed a spreadsheet-like component to support bulk pasting directly from a school's spreadsheet.
I had to balance the spreadsheet's visual design with technical constraints, following the team's decision to use a third-party solution.
Making a necessary step clearer and easier
We anticipated that our matching algorithm wasn't going to be perfect in every scenario, so I created a framework to present all the possible outcomes for users to review.
Our algorithm detects students without existing accounts and creates them with the class, so users don't have to.
Student matching panels
Users needed enough information when reviewing students, so I designed a container that could accommodate the required level of detail while keeping the information digestible.
Each panel provided alternative options in case the algorithm got it wrong.
Providing a fallback option
We were aware that sometimes none of the matches or options our algorithm provided would be correct, so I designed a way for users to search for the correct match themselves.
Users can search the entire school to find and match the correct student account and override incorrect matches made by the algorithm.
Empowering support teams with admin mode
Our support teams were still responsible for creating classes for schools that needed our help, therefore I created an internal version of the tool to speed up their workflow.
Pasting data into a text field isn't an intuitive UI pattern, but since our old internal tool had the same interaction, we decided to retain it.
08 / MVP TESTING
Internal testing led to 5 changes
I co-led sessions with support staff to test the MVP, observing how the algorithm performed with real data and validating our hypothesis. This also allowed us to identify key areas for improvements ahead of GA.
Increasing the confidence threshold of the algorithm
We noticed false positives where students were incorrectly marked as matched and often went unnoticed.
Raised the confidence threshold to favour user intervention over running the risk of purely relying on automation.
Tweaking copy to encourage thorough review
Before
After
Since we observed that the algorithm produced errors, we decided to err on the side of caution.
I refined the messaging to encourage review of all students rather than only the ones marked with a '?'.
Added class information to panels and tweaked the layout for better readability
Before
After
We found that some students were still too difficult to distinguish due to the similarity in names and data.
This highlighted a need for more context, so I added an additional data point and rebalanced the layout to accommodate the extra content.
Displayed student year level changes to improve transparency
Before
After
Support staff were unsure what happens to a student's year level after matching and class creation.
This indicated that the interface wasn't communicating the outcome clearly, so I added a visual indicator to make this more explicit.
Added a tag and tooltip to distinguish auto-matched and user-matched students
Before
After
We discovered that it was difficult to differentiate between an auto-matched student and a student that was matched by a user.
This made it confusing to double-check matches and return after being AFK, so I added a tag to students who were auto-matched by the algorithm.
From testing, we found that our hypothesis was partially validated as the MVP helped support teams create classes faster. Heading into launch, we continued monitoring the algorithm for issues and prepared to analyse usage data from teachers.
10 / LAUNCH
Immediate results from internal use
After a few rounds of QA and additional testing with real data, we released the feature to APAC users two months ahead of the EMEA rollover. This gave us time to observe how teachers used the feature and to train support teams.
Short term outcomes
The admin version of the feature reduced class creation time to roughly one-third of the original, significantly easing the workload on support teams and eliminating the need for additional staffing during the EMEA rollover.
What's next?
Unfortunately, my time at 3P Learning concluded following an organisational restructuring. The team has since continued to build on the product's rollover tools, further optimising Smart Create as the algorithm improved over time.
By 2026, a significant portion of classes during the APAC rollover were created by teachers rather than support teams, marking a major improvement from the previous year.
My reflections
Something I learnt...
If I had the chance...
If I had more time...
