Third Space learning
Third Space Learning provides schools with online Maths tuition. I was tasked with improving the experience of teachers using the portal to manage the programme.
User research helped me understand the decision making process when choosing lessons and an expert review revealed major usability issues.
In the design and user testing phases, I moved from paper to mid and high fidelity prototypes to design a solution that improved usability and encouraged behaviours that would lead to the best outcomes for pupils.
The final solution simplifies the onboarding and lesson selection process for these time-poor teachers, from a complicated and disorientating experience spread over multiple pages to something that's self contained, quick and easy to use.
I was responsible for interviewing customers, designing the new information architecture and interface, user testing, visual design and copy.
I was briefed to improve the user experience of the teacher portal. Choosing the right lesson is a critical component to the success of Third Space Learning so this became a big focus of the brief. If inappropriate lessons are picked, pupils aren’t going to make the progress they should be, and schools won’t see the results they expect and won’t retain.
Customer interviews helped me understand the mindset of teachers who choose lessons. I learnt that:
I saw first-hand one major pain point while conducting observations at a school. One pupil was so upset that they hadn't received their regular tutor that they could not continue with the session.
Later, I discovered that the platform allows you to assign pupils to a slot in the session on a weekly basis and each slot is tied to a specific tutor. If you don't add the pupils in the same order each week, they'll get different tutors.
The problem was that teachers couldn’t see which tutors occupied which slots, which led to incidences like the one I witnessed.
Going through the experience as a user (based on the scenarios identified through the user research), I identified some poor usability issues that were encouraging the wrong behaviour.
To see noticeable progress, pupils should use us for at least one school term. However when adding pupils, the default number of weeks was set to 1. This encouraged teachers to swap pupils in and out each week and because of the issue mentioned earlier they ended up with different tutors.
We provided a suggested next lesson for teachers based on the pupil's previous lesson, however this didn't take into account the teacher’s teaching strategy so cannot be fully relied upon. We had also created session reports to give teachers feedback on the previous lesson so that they could use this info to guide their lesson selection. However, the platform design allowed teachers to select the suggested next lesson at the overview level without them reviewing the session report from the previous session or sense checking the lesson's content and considering alternative lessons based on how the pupils were performing.
You had to go to the 'Curriculum' section to view the contents of our lessons. After finding a suitable lesson, you needed to navigate to 'Plan Sessions', relocate the lesson you had earlier identified and select it. This seemed massively time-consuming and cognitively intensive as you had to remember the name of the lesson that you wanted.
You could only use 1 filter at a time making it hard to narrow down the 270 lessons to a couple of relevant options.
Categories and Topics are treated as distinct elements whereas in fact they have a hierarchical relationship, where a category is made up of multiple topics.
A user flow helped me focus on the ideal path that we wanted users to follow.
Now I understood the users and had the ideal user flow which catered for the different scenarios, sketching allowed me to explore a better integrated lesson selection interface.
Stitching all my ideas together, I generated a final sketch:
I transitioned my sketches into a clickable axure prototype which I iterated on through 3 rounds of user testing. I developed 2 key scenarios to test; a new school adding pupils to timeslots; and a teacher choosing a suitable lesson for their pupils. My first round of user testing with former teachers (conducted via Skype), validated some ideas and highlighted issues to be addressed:
Users went to 'Pupils' to pick lessons for pupils rather than 'Sessions'. In the next round of testing, making ‘Pupil’ more specific by renaming it 'Pupil Profiles’ seemed to resolve the issue.
I designed the 'Assign pupils' interface so you only pick one pupil per slot. This was to help create the mental model that everything within each tab is tied to a pupil. However, one user ignored this and tried to select multiple pupils. From this, I realised that it's better to allow this as it’s quicker even though it slightly dilutes the concept.
I realised that clicking on each pupil you wanted to give a lesson was inefficient so I added a ‘Select for all pupils in session’ shortcut.
I learnt that users wanted to be able to write blanket comments to go to all tutors teaching a particular lesson so I added this in the next iteration.
After addressing the usability issues raised, I moved onto designing a high fidelity version of the key pages based on the existing style guide.
By this point, the start of the new school year was fast approaching so we had to start the build based on what I had produced. I was however able to continue with testing once we had something live on our beta environment.
This revealed a couple of issues, particularly when it came to choosing which pupils you want in a session. It wasn't clear when you’ve added too many pupils so I made the copy more explicit and introduced a prominent sticky counter which shows you how many empty slots need filling for the selected session time and how many pupils you currently have selected.
I also added logic to prevent the error occurring. The check boxes of the other pupils you can choose from become disabled once you’d selected the maximum number of pupils for the slots available.
For such radical changes, there were relatively few post-live issues that needed to be addressed. One issue I tried to resolve involved users who were accidentally cancelling their sessions.
The teacher portal redesign gave us the opportunity to create win-win situations by making the 'right behaviours' effortless and steering users away from the 'wrong behaviours'.
Pupils' progress is dependent on the teacher choosing the most appropriate lesson. If teachers don't see progress they won't continue to use us. By restructuring the site, teachers can now use information about how the pupil performance to inform their lesson selection and the new filtering solution allows these time poor teachers to quickly find a suitable lesson.
Previously teachers were only selecting pupils for 1 week at a time which caused problems with pupils receiving different tutors each week. By designing a system where they put pupils in for the whole term as the default (with the ability to change this), we eradicated issues that had previously bought pupils to tears.