Design Roles
Matt Weldon - UX Lead - wireframes, copywriting, information architecture
Jeff Heppert - UX Designer - visual design, information architecture
Jon Jecker - Product Manager - investigated technical problems with Auto-Unlock 1.0, helped brainstorm UX solutions, proposed troubleshooting flow
Spencer Carrol - iOS Engineer - redesigned Auto-Unlock system logic
Auto-Unlock 1.0
Auto-Unlock: Your front door unlocks for you automatically when you arrive home!
When your phone enters a “geofence” around your home, the August app wakes up and unlocks your door as soon as it detects a bluetooth signal.
Auto-Unlock 1.0 tutorial and geofence settings.
Technical Problems
While exciting in concept, the Auto-Unlock feature proved unreliable for many customers from 2014 to 2017. The feature was a major driver for purchases and customers really wanted it to work. But it often didn’t, which resulted in frustrated, vocal customers.
Numerous attempts were made by engineers to re-architect the feature, but a solution eluded the team for some time.
A large body of customer feedback and customer care call logs accumulated, and the team had heard countless anecdotes from frustrated users. In 2017 PM Jon Jecker reviewed and distilled the customer problems into three main categories:
Spurious Auto-Unlocks. Sometimes Auto-Unlock would unlock a door seemingly at random, even in the middle of the night while a user was sleeping. While a rare occurrence that only seemed to happen with a small group of customers, this was very bad as it could immediately ruin a customer’s feeling of safety using August. We took this issue very seriously.
Auto-Unlock happened too late. A not-uncommon problem for some users, sometimes Auto-Unlock worked, but not until the user was already at their door for a few seconds. Timing needed to be perfect with this feature - otherwise why not just use a key?
Unwanted / Unexpected Auto-Unlocks. Sometimes a user would describe a scenario that made them think the feature was broken but in fact the feature was working as designed. The most common example of this would be users who came home by parking their car in their garage, only to find that their front door had been unnecessarily unlocked by the Auto-Unlock feature.
A Clear Mental Model Was Lacking
In addition to the technical challenges with the feature, it gradually became clear to me that our messaging (or lack thereof) was doing a poor job establishing a mental model about how Auto-Unlock actually worked, despite the tutorial.
Some users, while explaining their troubleshooting process, assumed the feature worked like “Star Trek” in which the door would unlock any time they stood in front of it.
There was little in-app messaging about the feature, ironically because the team feared giving away too much proprietary information, and the only provided control was the ability to set the geofence size.
Engineering Redesign
In order to fix the spurious unlock issues (which turned out to be caused by “GPS drift” for users that lived between two equally distant cell towers) and Auto-Unlock timing issues, iOS developer Spencer Carrol gradually devised a new architecture for Auto-Unlock, and discussed with the design team frequently.
One of the changes Spencer decided to make was to change the triggers for whether or not to unlock the door when detected.
Insight: A New Mental Model
Teaching The Mental Model
Auto-Unlock 2.0 onboarding tutorial.
Reinforcing The Mental Model
Once a user had enabled Auto-Unlock, a key addition I added to the flow was to introduce notifications at key moments. This took the form of three basic but important “training wheels” notifications:
A “Welcome to Auto-Unlock” notification triggered right at the end of the tutorial. “Auto-Unlock is now set to Home mode and will not attempt to unlock your door. We’ll notify you the next time you’re away.”
An “Auto-Unlock is now set to Away Mode. We’ll unlock your door the next time you arrive home.” notification whenever we detected the user had left their geofence.
A notification when the lock was detected. This took one of two forms:
“Auto-Unlocking [door name] at [home name]. Auto-Unlock is now set to Home mode”
“Auto-Unlock was attempted, but we were unable to unlock the door. Auto-Unlock is now set to Home mode.”
These simple notifications were intended to set expectations among users as to how exactly the feature would work.
Settings and Troubleshooting
We decided to also give users more control over Home / Away status just in case it got out of sync, added a “history” feature to help with customer care calls, and provided users a way to give feedback when unexpected behavior occurred. We used this feedback to validate whether the new version had fixed the issues.
Outcome
After years of effort and iteration, August’s Auto-Unlock feature now has a 98% success rate as a result of the re-architecture and UI changes.
iOS App Store and Amazon ratings for the August app increased substantially shortly after the release of this feature.