A new approach to analyzing social media data and identifying niche communities and influencers:

Project is currently in stealth mode
being tested with exclusive beta clients.


I lead a team of engineers building a content aggregation and analysis tool to help brands understand their online communities. We've designed architecture for handling large scale data collection, filtering, and sorting and integrated with IBM Watson to use machine learning to enhance content understanding. 


THE CHALLENGE
While leading the creative team at mllnnl, I was looking for ways to target highly niche communities with micro-influencer campaigns. Unfortunately, Instagram's API does not support multi query searching. This meant that searching for niche influencers in a specific category and location was nearly impossible to achieve. In addition to this, as a performance driven agency, we weren't keen on hiring influencers based on their likes and follows - a metric which felt indirect to what we were trying to achieve.

Early wireframes for Audience.ID
OUR STRATEGY
To achieve our goal we needed to scrape Instagram and collect as much data as possible. Using this data, we could parse influencers based on multiple criteria and identify their audience interests as a better metric for campaign success. With that in mind, our engineer and I began designing a proper schema, and a minimal interface for an online tool our community managers could use to scrape the content they needed to identify key influencers.

Our first iteration of the product used by our internal team to scrape, analyze, and engage with Instagram content
LEARNING AND PRODUCTIZING
While our goal at first was to achieve specific campaign goals, we ultimately decided that we were better off turning this project into a complete platform that would enable community managers on multiple teams to collaborate on building micro communities and campaigns.
Our first launch of the Audience.ID platform
BUILDING. TESTING. SHIPPING. 
After building the first iteration, we began to learn and listen from our initial clients and continue to build and test new features. This lead us to cleaning up the UI, adding a machine learning component, and reorganizing the community filtering aspect of the product. As we began building more code, we adapted our internal process to be more organized and agile. This included writing and documenting robust API and developing a organized scrum workflow.
Audience.ID today. Completely redesigned and integrated with IBM Watson
LEADING A FULL STACK PRODUCT TEAM
In addition to my role as user experience and product designer, I was also responsible for hiring our developer team and managing our product and feature workloads. Using Toptal, I hired several software engineers, helped design our robust API, tracked features and workload, and managed our remote team of engineers. 
Below are a few screenshots of the inner working of Audience.ID
Back to Top