Candid is a clear aligner company who offers at-home teeth straightening. I joined Candid in October of 2018 as a product designer. My role was to identify problems or areas of opportunity to improve our customer experience. This project was the first project I worked on with a pretty awesome team.
- Me (product design)
- Sudy, head of research
- Mary, product manager
- Joseph, front-end engineer
- Diaphel, back-end engineer
Candid's lead time from a customer booking an appointment or ordering a starter kit was 2-3 months. Our customers consistently told us they needed some form of communication about where they were in the process.
At first we questioned what might be the real problem. So we took a look at our current CRM drip for customers in the top and mid funnels. The emails were sending properly, we provided enough context, but our open rates were low. So we took a look at our support requests. There was definitely a problem, ~27% of our customers had emailed support and ~21% had open support tickets. Since our start kits and our studio associates prompt all of our customers to create an account to fill out consent forms, upload photos, and purchase their aligners when the time came, we decided the best place to surface information about a customers progress was in the customer portal.
If we provide a time line showing next and current steps, customers will be less confused and know what to do next. We will know it's working if aligner conversion isn't affected negatively and the percentage of open support tickets goes down, the time through funnel goes down, or the number of emails to customer support goes down.
Our crazy user journey map of the prototype screens. We did include alternate routes to account for 3 major edge cases.
In order to make sure this was a good use of our engineering time, we decided to first user test our idea. So I designed and built a usable prototype of the new timeline experience. This prototype had interstitials to indicate the passage of time. Because we had to user test an experience that could take 4-6 weeks to experience, we decided to do in-person moderated user tests. We found six people who were interested in clear aligner treatment and rated them based a number of criteria for the experiment.
Of the six people we tested with, only one had a tough time understanding the timeline and it's purpose. The other five were able to understand what was happening and could resonably assume what to expect or what would happen next. Four of the six knew how to self-help within the prototype should things not go as "planned".
A few detailed shots of the screens.
Because the lead time was so long, we needed to figure out the best way to run our A/B test without sacrificing too much important data and keeping to our fast timeline. It's do or die in the early startup world and sometimes accuracy is sacrificed for survival. We chose to run our test for 4 weeks, rather than the full 2-3 months. This meant we had to reduce our statisitcal significance threshold from 90% to 80% in order to make a decision.
Screen recording of the prototype walkthrough.
At the end of the 4 week A/B test, we had reached 80% statistical significance and decided to assess the test. Our criteria had been to keep aligner conversion from being negatively affected while reducing either the percentage of customers with open support tickets, the time through funnel, or the number of emails to customer support. We succeeded on 3 of those 4 fronts.
While we mostly nailed this design and implementation, it did highlight a few areas we needed to work on. The most negative affect was the number support emails, we believed and later confirmed that this was no longer an influx of support questions about where customers were in the process, but meatier questions regarding things like how to take their intra-oral photos and teeth impressions. We also noted that the affect on time through funnel wasn't a drastic improvement. After taking a closer look at the time it takes customers between ordering their at-home kit and getting their aligners, we noted two specific problem areas: 1) the time for customers to return their impressions and photos (and do them properly, a high percentage of customers had to redo them) and 2) the time it took for aligners to be produced and shipped. You can read more about the work I did to improve the impression and photo process here.