Product & User Experience

MikMak Analytics

Content Performance and Usage Analytics


MikMak was tracking user behavior by collecting page-level events in Google Analytics, but we found it difficult to develop insights and report on the behavioral trends we needed to iterate on our video content and to evolve the user experience of our iOS app. Furthermore, we were about to introduce varied feeds and test different versions of our videos, so it was critical that we advanced our event pipeline to account for feed position and variations in our creative.

During our beta phase, we tracked simple metrics such as new and returning users, session instances and duration, and actions like viewing product details or adding merchandise to cart. We also began iterating on user segmentation through identifying patterns of repeat visits of three minutes or longer. However, we did not yet have the sophistication to ascertain what content led to user engagement and loyalty


As we reevaluated our approach to analysis, we developed top-line questions about behavior in the MikMak iPhone App:

Content Insight Questions
Which content leads to loyalty? causing a new visitor to come back 3 times or more
Which content leads to product intent? causing a recurring visitor to take action on a product
Which content leads to conversion? causing a visitor with product intent to buy
Which content leads to user-evangelism? causing users to encourage others to become users

User Insight Questions
How do we recognize users who may become recurring users?
How do our efforts to engage them perform?
What do recurring visitors have in common?
How do our efforts to convert them perform?

These questions drove re-design of our events and reporting — both in our analytics tooling (custom events in ObjC + Segment + Mixpanel) and how the MikMak API would deliver content and user tags to the iOS app. For example, in our new video taxonomy the MikMak platform passed data about where and when a video with a particular host-product combination could be watched by a particular user. This allowed us to experiment with how products and on-screen personalities might independently effect app loyalty and purchase conversion.

As our data accumulated, the design of this analytics program would afford MikMak the ability to define and segment our audience based on Loyalty (based on duration and frequency of visits), Engagement (derived from proportion of videos watched, content-seeking vs. grazing behavior, number/depth of product and social actions taken). We could also analyze Visit Patterns and Funnels to determine what behaviors led to purchase and what activities indicated emergence of user loyalty.

In addition to providing a framework for experimentation with our content, merchandise assortment and app experience, our short term goal for this analytics program was to report on the app’s performance as we launched publicly to our investors and the brands whose products we sold. In the long term, this program laid the groundwork for scoring content so we could develop personalized video feeds for our users.

MikMak Analytics Solution.png



During our beta phase (February – May 2015) I performed all analysis and reporting using our Google Analytics implementation. As the app experience evolved I led the design of our analytics strategy in partnership with the CEO and VP Engineering — authoring our framing questions and hypotheses, and managing two data-science consultants to define our data taxonomies. Alongside engineering, I oversaw the implementation of our complete data pipeline from content-tagging in our CMS, to app events, to collection in Mixpanel. Once in place, I personally coded custom reporting dashboards in Mixpanel’s developer platform (an HTML & javascript portal with direct access to data and templates).