Gypsy Circle is a future tense social network that helps friends connect before they arrive at their destination. Input a trip and see friends who are also visiting as well as friends who live there. Once you input a trip, those friends will receive push-notifications alerting them of your travel schedule.
Challenge: When Gyspy Circle came to SSI, they had already launched the first version of their mobile app on the app store. The first version failed to perform properly as it was slow and was crashing frequently. The biggest reason behind their product's bad user experience was complexity of the algorithm. There were two items on the travel feed algorithm that were taking about 20-30 seconds each (Trip Overlaps, Friends Visiting) with high CPU and memory usage on server side.The user experience kept deteriorating for people having more friends and more trips as the algorithm had exponential complexity. Trip Overlap Time was estimated to be 23 seconds and Friends Visiting Time was calculated as 15 seconds which was very slow.
Solution: SSI did a detailed evaluation of the application and server side code. In order to cope with the challenge of complex algorithms, SSI had to change the database architecture and migrate the data as per the new schema. The biggest challenge was to change every part of the application so that it became compatible with the new schema. As a result of mapping new algorithm Trip Overlap Time was brought down to just 0.3 seconds and Friends Visiting Time below 1 second.
SSI re-wrote major parts of the mobile app, modified data models and designed new modules for application to improve performance, robustness, scalability, user interface and experience.
Result: Application was launched with lots of positive press and marketing. Great user adoption was observed in first few weeks of the launch. Application has been performing exceptionally well as user base has been growing exponentially.