Case Study 3: MyTutor — ‘Lucy’ Conversational Interface
The UK’s leading provider of online 1–2–1 school level tuition wanted to better match the needs of their students to the differing qualities of their tutors. Through analysis of detailed user research we developed ‘Lucy’ over the course of a two week sprint, a conversational interface embedded into the company’s website that ensures that the needs and personalities of student, tutor, parent and business alike are delicately and efficiently matched.
My team consisted of fellow General Assembly UX Design Immersive students Florence Opera, Aaron Standen and myself. The group was self-directed; scheduling and task delegation was agreed through daily stand-up, discussion and constructive appraisal.
MyTutor’s mission is to share the knowledge, experience and enthusiasm of the UK’s brightest university students with secondary school pupils looking for one-to-one support. Their exceptional tutors have inspired thousands of pupils across the world. In return, tutors get a rewarding, flexible and well-paid way to work.
Having raised £1.8m to date from some of the UK’s top investors the company puts innovation at the heart of their fast-growing business aim to revolutionise the tutoring landscape.
MyTutor know that different students have different needs; they have different learning styles, have their own unique personalities and reasons for employing a tutor, and they understand that matching the right tutor to the student is necessary to provide optimum service.
MyTutor’s service receives glowing reviews from their users. At the time of writing there was almost 27,000 reviews transparently listed on their site, a staggering 25,500 of which have 5-star ratings. Clearly this is a fantastic response and the service is working for those who use it. New-comer students and parents first get an enormous list of every available tutor, enter a few top level filters, get another multi-page list then find which tutors they are interested in working with. They are then are offered a free no-obligation 15-minute interview with them via the company’s online platform to get a feel for whether there is a personality match or otherwise and book lessons with the tutor they like best.
To make the user’s journey to a successful free interview and subsequent lesson booking reliably easier, MyTutor tasked our UX team to develop a new student-tutor matching interface for their website that would provide a curated list of tutors best able to provide for a particular student’s needs.
We started by checking out the competition, both direct and indirect, in some competitive analysis. We found that the UK tutoring market is sizeable with 23% of 11–16 year olds using tutors, but that tutoring is considered expensive for most parent’s incomes. All direct competitors provided dauntingly long lists of available tutors and offered little or no help in choosing. At least, in spite of their problems, we could happily confirm that MyTutor were at the top of their field and would really raise the bar considerably by taking the extra step to provide a highly curated tutor list. By removing the travel costs associated with traditional tutoring MyTutor is able to offer extremely competitive prices. Notable indirect competition was found in US online therapy website https://www.talkspace.com/ who’s service also relies on a good match between user and service provider, communicating it through an insightful website.
Our team visited the MyTutor offices to hold a meeting with staff members with diverse key roles in the business. Aaron presented our research findings and our analysis of the brief. Each staff member answered our questions and offered their own valuable insight into the company, its dynamics and challenges, and into their personal experience working there.
MyTutor has up to 3000 tutors ready and available to teach a potential student at any one time and the journey to find a tutor incorporates only a few filters options; subject, level, gender, premium (most highly rated and experienced tutors) and second subject. A cursory use of these filters will narrow the total down to a list of about 500, which is still a daunting 28 pages of 16 tutors.
During our stakeholder interview both company founder, Robert, and part-time UX Designer, Connor, used the phrase “choice paralysis” to describe the tutor search list. The company’s 16 employees, who share the duty of answering calls, were feeling the burden of many parents resorting to phone calls for help. As a company who are establishing a successful new product with rapid expansion and investment they know that they need to find a solution to overcome this user pain point quickly to meet growing user expectations and ease the strain staff for phone enquiries.
Through discussion with the client team we got to the heart of the brief: to design an interface that provided a simple and enjoyable user journey to a closely matched shortlist of tutors who could be easily contacted to arrange introductory interviews.
Time to gather quantitive and qualitative information was tight. We only had a day to formulate, disseminate and analyse surveys, and another day to conduct interviews. As a parent myself I was able to utilise my network to share our survey on local parent forums which gained an considerable level of response and we received 61 responses that provided a total of 12 scheduled phone interviews.
Although we wished to speak with students we were only able to arrange interviews with parents and tutors, who would speak about their perception of their children or tutees. We split into groups, Florence and I conducting interviews by phone and Aaron preparing our presentation for a stakeholder interview session the following day.
Key findings from our interviews are illustrated in these pull-quotes:
“It’s all about getting tutees interested and motivated”
— Sanaz, 41, parent
“You really have to personalise tutoring, there’s no one size fits all.”
— Toby, 22, tutor
“Personalty connection is crucial”
— Joy, 49, parent
Affinity mapping the user interview data clustered topics and set out a structure of priority. Interviews stated that for both parents and tutors, rapport is crucial to motivating students and a successful learning outcome. During interview, after initially considering that good value in lesson time would probably mean getting straight to work, after their listing qualities they believe make a good tutoring experience, parents unanimously agreed that they would consider paying for time for the tutor and tutee to get to know each other as lesson time and money well spent.
Task Analysis & Personas
MyTutor, employing active and qualified UX personnel, provided three personas that they work from. Julia, 46, represented ‘the caring parent’, Anna 42, ‘the cautious parent’ and Tam, 51, ‘the structured parent’. During our stakeholder interview Anna and Julia, although having their own varying needs didn’t represent a challenge to current company systems and represent parents who are generally very happy and complementary of the service.
Tam on the other hand micro manages everything he is involved in and presents a considerable drain on the company’s phone support in his requirement of a highly curated service that the current matching system doesn’t provide. We chose him as primary persona as we felt by meeting his demanding needs we would satisfy Julia and Anna’s.
In conducting task analysis of Tam’s journey from landing on the MyTutor site to scheduling intro interviews we realised that a key component of his story was his child, the judgement of the service being the outcome from their lessons. The tutee was the crucial missing link in the triad of users, the parent, the tutee and the tutor.
We created the persona for Tam’s daughter Holly, a highly capable and ambitious girl currently in her first year of sixth form college at a girls-only grammar school. She struggles to make connections in crowded environments and finds it easiest to develop friendships through mutual interests. She is naturally academic but prone to feeling despondent from time to time due to her dad over exerting control over her schedule and restricting opportunities for her to develop socially. She’s bright but doesn’t connect to people easily.
Dusting off my old pen tablet I used in a former career as an illustrator I drew up a story board to depict a happy path to guide the development of our project.
- Tam is aspirational, he has his sights set on sending his daughter Holly to an Oxford University and he wants to employ a tutor to help push her already high grades up from an A to an A star to ensure a competitive edge come admissions time. Tam is mostly concerned about the credentials of the Tutor. He’s more interested in things like the University they attended and the grades they obtained.
2. In spite of Tam’s criteria of what makes a good tutor, through our matching interface Tam and Holly to work together to find an ideal tutor match, one that would satisfy both their needs.
3. By matching Holly to a tutor custom chosen for her she’s found her optimal learning environment.
4. She socially opens up, improves her grades, starts dreaming about achieving getting into an Oxford University and crucially for Tam…
5. … this eventually leads to that admissions letter from Christ Church University, Oxford, that everyone was hoping for! Success!
“Learning to choose is hard.
Learning to choose well is harder.
And learning to choose well in a world of unlimited possibilities is harder still, perhaps too hard.”
― Barry Schwartz, The Paradox of Choice: Why More Is Less
We needed to prevent choice paralysis it in such a way that a parent like Tam walks away feeling like they had a richer more bespoke experience without having to engage with anyone from My Tutor in person, Holly finds a tutor she clicks with and the phone lines at the MyTutor offices are less congested with parents’ queries.
Taking our research and some design challenges we returned to the MyTutor offices to have some fun! Design studio is a great way to generate solution ideas fast, engage stakeholders and have a great laugh while doing it. It was also a great opportunity for me to use my workshop leading skills to ensure a constructive and inclusive ‘anything goes’ atmosphere to get some really amped-up creativity going. Unfortunately I can’t show you the fantastic drawings the group whipped up during the warm up drawing game as I’m proud to say they are currently adorning the client team office. It was fantastic to have such a fun kick-off!
Of the many ideas that flowed from the session the ones that got the most votes at the end were positioned on a feature matrix to weigh up impact versus feasibility. The client team were keen on employing some kind of chat interface but needed to keep a personal feel. Various filtering functions were put forward, of which a thumbs up/thumbs down to rate and reject/replace tutors from the curated list was very popular. To communicate that crucial factor, tutor personality, it was noted that nothing beats a video, no matter how short and that two names popular tutors represented the two extremes of popular teaching styles, one gentle/encouraging and other fast/direct.
Prototypes, Testing, Iteration
The team took the feature ideas and thoughts generated in the design studio and set about proposing more structured solutions. To synthesise the popular chat idea with filtering and attentive service functionality that didn’t involve live human input, the idea of an illustrated character guiding the collection of tutee information through a conversational interface styled form was developed through the team’s own second phase design studio. The basic vertically scrolling text section with left and right facing speech bubbles is familiar to anyone who’s sent a text message or any other messaging app, satisfying the familiarity and learnability heuristic. Given the working name of ‘Lucy’, this light-hearted and neutrally-styled character went from the scribbled sketch shown on the left in the image above to the low-fidelity wireframe to the right. I expect no prizes for life drawing here.
To communicate progress a tag field is populated to filter tutor list results and visibly build the student profile each time a question from Lucy is answered. The low fidelity wireframe to the right was developed to afford streamlined, time-efficient interactions to satisfy Tam’s haste with no instructions necessary. User testing found a good reception but it was found that the directionality of questions and answered needed clarification. As we were supplied a complete UI kit by MyTutor we moved straight to high fidelity with a near symmetrical layout with Lucy illustrated to the left and the student profile built to the right of the central chat scroll box. Aaron and I developed tone of voice and a channeled journey to represent Holly early and guarantee her soft interests are incorporated in the filtering.
The tag cluster was iterated through user testing from sketch (1), where request for explanation of functionality was requested to low fidelity wireframes (2), where in test the text explanation was found too wordy, to high fidelity (3) where the design is familiar enough to tags used in many content curation apps to prove self-explanatory. An ‘x’ deletes a tag, an ‘add a tag’ text field provides fast and flexible input and Lucy remains to be called to go back to the questions if necessary.
The thumbs up or thumbs down like/dislike filtering feature that came out of the design studio that was popular with the client proved very unpopular with users who were uncomfortable judging another human with a negative stamp, particularly one that may be captured and stored and may be used against the tutor. This concept was iterated upon to become a familiar favourite/bookmark icon that saves a tutor to Tam’s logged in home page profile as per existing website functionality, yet with the added feature that the favourite choices influence subsequent searches.
Lucy — in Focus
So, to recap, who is or what is Lucy? Lucy is the missing link between the parent, the student and the tutor. She is a conversational interface. She makes it fun, quick and easy to gather all the important information necessary to make an informed match between a student and a tutor.
- The basics — Subject, level, price preferences
- Scheduling — Unavailable tutors are frustrating
- Learning and teaching style
- Interests — Bonding over shared hobbies
Basics, things that are important to Tam, like the subject, level of study and cost. Scheduling preferences was key, as it was important for us to find a way to avoid pairing a student with a potentially unavailable tutor which would obviously be frustrating for a busy man like Tam.
Tutor preparedness. We found that tutors and parents all expressed a value in and a benefit of having a “getting to know you” portion built into the tuition experience. We feel that by offering the tutor a “cheat sheet” of the student’s interests and hobbies they could accelerate the process of getting to know each other and gain the student’s trust faster.
Early user buy-in. throughout the on-boarding questions we’ve written in
value statements reminding the user of the benefit of not only a great match but also of education and higher learning in general.
Quick profile building. All of the information gathered is logged in a profile section. This results in an already well developed client profile waiting for them when they fully sign up to MyTutor.
Data gathering. Even if a user doesn’t sign up for the service after their chat with Lucy, the data gathered will be valuable for demographics and targeting potential clients in the future.
You can try out our clickable prototype for the Lucy, as accessed from the existing home page of the website via this link:
After our 20 minute presentation to MyTutor staff Sonal and James they expressed enthusiasm for the body of research we presented and for our solution to their problem. Our group look forward to further discussing our work with the MyTutor team in the coming weeks.
To further this project in a subsequent sprint we would first like to implement responsive design for mobile, as mobile usage is prevalent for users, then we would like to realise a way to show instant tutor results based on answers to Lucy’s questions appearing in real time to give instant feedback and progress to the interface experience. We would also collect extracurricular interests of tutors so the data be available for matching searches.
This article is an independent and personal work. All opinions presented herein are that of the author’s and are not those MyTutor.