Reverb Insights Dashboard
Reporting dashboard for a product that doesn't exist yet.
Inventing a Product
Reverb Insights was meant to be a product by which large-scale publishers (think Time Inc., Hearst, Conde Nast, etc.) could leverage Reverb's natural language processing engine to drive recommendations across multiple properties. The consumer-facing components would vary widely across our prospective customer base, so we began with the reporting dashboard.
Initial meetings revolved around typical early-stage startup questions like "what does our platform do?" and "what would customers want to use this for?".
We concocted a series of pain-points we imagined big publishers would have in reaching their audiences, and set out to create tools that would help solve these problems.
Navigation Map
The idea was that our customers would run all of their content (their corpus) through our system, which would analyze every word of every article, find connections, and organize them into groups according to concept. For example, an article about Steph Curry could appear in any of the following concept segments: basketball, NBA, Golden State Warriors, Oakland, Davidson Wildcats, Christian, etc. Ultimately this information would be used to provide relevant article recommendations.
Once the publisher's corpus had been ingested, the system would run on the publisher's network, keeping track of traffic and (eventually) generating profiles of site visitors.
The dashboard shows information about content in blue, displaying things like the number of articles that exist in a certain category, and the most popular articles in each content segment.
Yellow charts indicate information about the audience, generally focusing on displaying the concepts that are driving the most traffic.
Since we had no actual customers during development, we put an emphasis on flexibility. A global view of the entire corpus was always available, but some more specific segments were dynamically generated, and others could be created manually.
Information Design
One of the most prominent charts we needed was one showing the amount of traffic to a series of different concepts. We also wanted to show trending data: in this example, the lighter-colored bars indicated traffic from the previous week.
There was no upper limit to the number of concepts that could be displayed, so we needed another option that wouldn't require so much vertical space.
Layering the current and trending data worked well, effectively doubling the data density. The arrows were added to indicate trend direction, but some people still found this presentation obfuscated last week's numbers.
The final solution layered the trending information a different way, which delivered enough clarity to satisfy the team.
Final Mockups
