eeco
UI / UX
Mobile Development
About
Client
Eeco is a news app developed for a young startup team that had a genuine vision of letting people digest only the news they want and only when they want. When this seemed like a yet another content curation project, the client had a huge vision for the future which is the need-of-the-hour, 'FAKE NEWS DETECTOR'.
The client was very clear in making this app a 'first-of-its-kind'. While other applications only deliver the news articles tagged with the interest categories of the user obtained earlier, this app was meant to 'curate' the content for the users throughout their journey. Beyond this, the application had to nudge the user to digest the articles only when they have free time amidst their daily plans & schedules.
The
Requirement
Solution
Firstly we were very much convinced to take up this news application development project owing to the interesting features it demanded. Curation had to be done by implementing Machine Learning algorithms. For prompting the user at the right time with the right articles, integrations with the user's calendar on Google seemed to be the right start. For spotting and demoting the fake news among thousands of articles available online extensive datasets and training models had to be in place.
Approach

As usual, our ideation team started brainstorming to validate the features and its contribution to the app's intended purpose. On successful validation, feature enhancements were discussed to deliver the best for the next big thing in the 'News & Magazine' category.


Since one of the main features of this app is to help the user digest the news even in the shortest free time one could have, we were more cautious in delivering the best User Experience possible within the limitations of this application.


The Google Calendar integration system was developed to look for the free time of the user between his scheduled plans. An extension of this feature will be analyzing the users' usual commuting times through public transport for the articles to be delivered and the ones through self-driving commute when the articles won't be prompted for reading.


For example, when a user has 10 mins of free time, one or more news articles that could be read well within 10mins alone will be suggested. The user's reading speed data will be analyzed to further deliver the right articles just not to leave reading a news midway.

Analyze the users commute to decide when to deliver.

For the curation algorithms, Python Pandas was used by the developers with the help of existing datasets available with the client. Provisions were made to let the user 'Thumbs-up' or 'Thumbs-down' the article shown to him. Based on the user's feedback the algorithm learns by itself to improve the suggestion accuracy. In the long run, the users will be segmented and tagged based on the common characteristic models they share.

Python React JS elasticsearch MySQL
Results
The curation accuracy of the application improved by more than 80 % after 30 days of usage with an average of 7 articles consumed per day per user.
45 % of new-sign-ups were through referrals from existing delighted customers.
2 x improvement in sales volume when compared to market average.
Client
feedback
We chose RootQuotient after having a bad experience with an agency earlier. Their sound knowledge in latest technologies was helpful in achieving complex technical goals.
Cathy
CEO