PEARLII
Using A.I and your own smartphone to replace the dental check up
With a strong team, funding pitches in place, a prototype websites and a neural network algorithm in the process of being trained, the biggest obstacle for Pearli was a way for people to take selfies of their teeth that are of a consistently usable quality for dental assessment. Put simply, we needed to teach people to how take photos of their own teeth. The photos needed to be of a standard that the algorithm could use. This was complicated by the fact that the algorithm needed 5 photos of 5 specific angles to capture enough visual data to assess the users dental health.
Deliverables & Constraints
-
Creation of photo-taking instructions.
-
Research artefacts.
-
Creation of onboarding prototype.
-
2 week sprint.
-
No previous research on Millennials (from Pearlii).
Roles
The User
Problem statement & Scope
Three UX/UI Designers dividing research, synthesis & prototyping equally.
Pearlii was pivoting towards millennials, a large tech-savvy group open to trying experimentation with innovative technology.
Millennials would like to use new tech for their oral health needs but the difficulty of taking selfies of their teeth remains an obstacle.
-
Desktop Research
-
Client Meeting
-
Problem Statement
-
Competitor Analysis
-
Survey
-
One-on-one Interview
-
Contextual Inquiry
1. Painpoints & Personas
2. Customer Journey
3. MVP Matrix of features
4. Process flows
Desktop Research & Competitor Analysis
We researched how other dental apps worked. How were they different? Who was their market? What technology where they using? What did their on-boarding process look like? To all these questions we asked 'Why?'. We where mining for gems of good UX.
​
We found that in Australia, there wasn’t a single app that did what Pearlii did. In the U.S - which we were not able to access - there is a competitor who’s process was to take dental photos in pairs mainly because of the higher usability of photos that are taken by someone else. There were pieces of the onboarding process which we could borrow from other apps, however, primarily we were going to have to build this from scratch by ultra-fast trial and error.
Client Meeting: 'To succeed, we need people to be able to take usable photos of their own teeth.'
We discussed the companies strategic position, it’s next steps and where our sprint would be able to add the most value. The pieces were in place, what they needed now was proof that people could actually take and upload photos effectively on their own. It was up to us to experiment and find a novel solution to this problem.
Survey: We Asked Millennials About Their Dental Health Habits
Key insights:
-
Money and time are the biggest obstacles to going to dentist
-
Most only go to dentist when there is a problem
-
They're very open to using A.I and other new technology to access their dental health
One-on-One Interviews
Key insights:​
-
Money & time where again biggest obstacles to going to the dentist
-
Most had not been to dentist in years
-
They only visited the dentist when they felt their was a problem
-
Going to dentist was seen as a unpleasant experience
Contextual Inquiry: The surprising difficulty of taking a good dental selfie
It was time to understand from the ground up how people take dental selfies.
We decided to keep it as raw as possible. How would people take photos of their teeth when shown the photos that they had to replicate?
​
Task: we gave people a set of photos that where deemed usable dental photos. We ask them to take 5 photos based on these photos. They used there own smartphone to ensure familiarity with a device.
Our hypothesis was that whilst it would take multiple attempts, seeing the photo that they needed to replicate, the average user would be able to find successful ways to replicate the photos through trial and error.
​
We were wrong.
​
Key Insight:
People who attempted to take photos of their own teeth with no instructions, demonstrations or any guidance except the photos they had to replicate demonstrated high levels of anxiety, confusion, frustration and anger. They often gave up and often communicated that if it was just them ‘they would have given up by now.' And half just gave up.
We Built A Photo Taking Process From Scratch
Given the previous attempt at letting millennials figure it out on their own, we decided we needed to create some written instructions as a structural starting point.
To create instructions we first had to learn what to do and how to do it. Essentially, the task we had given people yesterday, we now had to do ourselves. We had to find a way to consistently take successful dental selfies.
First, we worked with individuals in the first contextual inquiry round to see what ways would work. We found that standing in front of a mirror, reversing the camera was a much easier way to capture the correct angles using a smartphone. It would utilise the focus of the primary camera and show the screen photo being taken clearly in the mirror. We eventually scraped together a set of instructions that had some success. We then became the test dummies, making sure that we could consistently take accurate photos. We then experimented with different written instructions until the instructions were simple enough that we could all follow them in a step by step process.
Usability Testing: Words Are Not Enough
We went back out and did a usability test and asked millennials at a local university campus to take photos of their teeth using the instructions. We explained what Pearli was all about and whilst there was a definite sense of awkwardness with strangers asking questions about teeth we managed to persuade people to attempt the task.
We found that written instructions on their own where still confusing, however, when we used the instructions to demonstrate to them visually by doing it ourselves, they were able to copy us with a high success rate. We gathered that while the verbal instructions where important for detail, a ‘how to’ visual guide would be neccessary in our onboarding prototype.
Introducing Megan Swanne
We synthesised the survey, interviews, contextual inquiry and usability testing insights into a primary persona.
Mapping A Macro Journey & A Micro Journey
We then mapped out a macro level relationship Megan has with her Dentist according to the research. We then mapped out a micro journey of Megan attempting to onboard without effective instructions. We chose these journeys because they illuminated where Pearlii could add value (macro level) and they illuminated the experience we were working to prevent by designing an effective onboarding process for Megan.
Ideation & MVP Matrix
First, we isolated pain points & opportunities. Second, we did some unbridled ideation with how we could build features based on these pain points & opportunities. Third, we ran a draft MVP (internal to the group). Fourth, we ran a final MVP with 2 of Pearlii's Co-founders to make sure that we built the best possible prototype.
Process Flows
We then created various process flows based on the structure provided by pearlii, our desktop research into apps and our research into what Megan would want.
Lo-Fi Sketches
Based on the Process Flows and MVP Matrix we drew multiple versions of what the features might look like on the prototype.
Medium Fidelity Prototype
We built a medium fidelity prototype on Figma.
Usability Testing
Equiped with instructions and handheld mirror we went to a local park in the city and asked Millennials to use the prototype as if a friend had just recommended the app.
Their task was to complete the on-boarding process which included taking photos.
​
Note: Given we had proven that Millennials could consistently take successful dental selfies if they followed our process, what we where testing was our instructions ability to instruct them to position the phone correctly. We still did not have photographic instructions so the medium fidelity was heavily dependent on written instructions. If they positioned the phone correctly relative to their teeth and mirror we counted that as succesful.
​
Key insights:
-
Written instructions should be less wordy.
-
Some found taking usable selfies challenging (but we inferred this would change with high definition photos instructions).
-
Instant results should be feature of the app.
-
Intro page had too much information and needed to be broken up.
-
On-boarding questions about dental health should be done consecutively and near start (people preferred to get it done in one go).
-
Pricing was confusing.
-
Details of what exactly was a 'dental assessment' was unclear.
High Fidelity Prototype
Using Figma again, we iterated and created our Hi-Fi Prototype making changes according to the feedback. For the Hi-Fi it was important that we had a visual guide/instructions. So we took step-by-step photos of the method which we had been demonstrating to people in person. We also placed all the onboarding questions together as suggested by the previous usability testing.
Usability Testing
This was the first usability test where we could rely on detailed photo instructions.
​
The feedback?
​
“Simple questionnaire.”
“Simple instructions and easy to understand.”
“Very intuitive.”
“Very easy to use.”
“I like that you can review them (questionnaire answers).”
“Honestly looking at pictures is more helpful than reading the text.”
“This is actually really cool!”
“I would really use it!”
​
Area of improvement:
​
- some people thought the written instructions could be even more succinct
- more explanation of the 'results page'
High Fidelity Version II
Just some quick iterations based on the previous round of usability tests
