User Experience Design, Information Design, User Interface Design
Taboola platform allows online content publishers to market digital content through a recommendation engine which is served via site widgets and feeds on publishers and websites across the web.
Network Insights allows publishers to compare articles to overall traffic in the Taboola Network and identifies topics that are interesting to readers. Network Insights spots emerging and waning interest in topics that informs the publisher which topics to cover.
Publishers have a suite of tools that help inform editors, writers, and social media managers to make decisions about their articles on a daily basis. Most of these tools do not provide a way to identify topic trends that are emerging or declining in interest.
To allow publishers to make more informed decisions about topic coverage by leveraging Taboola's rich dataset, surfacing emerging and declining topic interest, and keeping publishers in the loop by sending notifications when a notable event occurs.
As the sole product designer, I led the design initiative to take Taboola's rich dataset and design network insights with proper data visualizations that made sense to the end users at the publishers.
Network Insights started as an internal idea. It had to be validated by publishers before considering the feature. Through conversations with publishers and internal teams, we collected data to get the assurance that it was worth building and prioritized what publishers wanted from the minimal viable product.
To see what repeat visitors (i.e., loyal readers) were interested in on their website and across the Taboola network.
Trends of loyal readers on the publisher's website and across the Taboola network.
A breakdown of categories and topics that are interesting to loyal readers.
Related terms connected to trending topics.
Actionable alerts based on thresholds.
Types of alerts:
Big Gain: Your coverage on topic X has increased Y% from last week.
Opportunity: Topic X has spiked since yesterday with a Y% increase in interest in our network.
Decreasing Interest: Topic X has declined since yesterday with Y% decrease in interest in our network.
Surprise Interest: Topic X has increased interest in our network amongst your loyal readers but not getting any traffic on your site.
Early Fatigue: Interest in Topic X is still attracting interest in our network, but you have not written about it since last week.
Following user research, I started writing user stories with the product manager to break down and prioritize the requirements of Network Insights. I also started exploring data visualizations that would best reflect the data. We also needed to consider the application posture — power users who want to analyze the raw data and form their own conclusions. Based on the variety of publishers that would be using Network Insights, we knew the design required a hybrid of easy-to-digest visualizations and granular data that would allow a deep dive into the data.
I designed the layout with a bird's eye view of loyal readers and popular loyal reader categories at the top half and the more detailed topic data at the bottom of the page in a sortable table to allow publishers to delve into the topic analytics. A loyal reader is defined as a person who has visited the website twice within two weeks.
We beta tested with a few publishers to validate our design and gather feedback. It went through several iterations to clear up confusion with the data visualizations. The feedback was generally positive, but there were a couple of lingering user experience issues that were out of scope for the first launch that we wanted to address.
Network Insights topic data was not live — it was about 24 hours behind. This lead to some confusion and also made the data slightly less useful.
No data visualizations for topic data.
In the second iteration, Network Insights Live, we tackled the two issues mentioned earlier. I explored data visualizations that would best represent topic data and decided to go with connected bubble charts.
Topic bubbles are color coded by category.
The size of the bubble is relative to the portion of traffic it is receiving in the Taboola network.
The lines connecting each topic bubble has a relative thickness in relation to how strongly the topics are related.
After launching the second iteration, we discussed adding an interior bubble within the topic bubble that represented the publisher's captured traffic. However, in most cases, the publisher's captured interest would represent 1-5% on popular topics and representing that in relative terms would be confusing — publishers would see a small dot most of the time. We tried it out but ultimately decided against it.
After completion of the second iteration, we saw an uptick of usage. Publishers were pleased that we built a tool that helped surface topic trends and a user-friendly way to explore topic data on their website and Taboola's network as a whole.