PitchBook Data, Inc. is a financial research and analytics company. It helps venture capitalists, private equity professionals and professionals involved in M&A make smarter investing decisions.
From our 2.1 million article data set, we had 4 deliverables:
- Sentiment analysis algorithm
- Interaction flows
- User experience
- Information visualizations
- Led 3-member team through vision, UX design, and execution
- Helping two programmers build front-and-back-end algorithms
- User research
- Wire framing, wire flows, and prototypes
- Liaising with PM, developers, university, and customer success team
I ensured that while we built the UX and algorithms separately, we touched base on core user needs.
As with any design project, I faced a number of constraints. The salient constraints were:
- VC users wouldn't be directly accessible.
- We had access to an imperfect 2.1 million article data set that required cleaning.
- Asynchronous work with PitchBook Data teams.
We received a number of user personas from PitchBook as part of the user research process. The ones below are the most relevant to our scenarios.
In addition, I interviewed PitchBook’s customer success managers, product managers, and developers. My salient numerical findings were:
The final statistic was especially important. This meant that I had to come up with a design solution to make our work accessible to users. Other important findings were:
- Heaviest users are analysts at the lowest levels
- Most of the sales go to companies or teams; rarely individuals
- Market sentiment is a subordinate metric to hard numbers
- The specific visualizations are very utilitarian, as VC’s are always in a hurry
The initial flow into the Signals app was as follows. I
Before creating a modified wireflow, I had to create the wireframes for the product itself and model how the product would be accessed at different points in the platform. I created a few paper wireframes and more complete wireframes on Balsamiq.
My modified user flow radically simplified access to the data. The SentiSage userflow is highlighted in blue.
I also created a wireflow combining the wireframes I created with an the user flow.
We decided to visualize our results in as line graphs, as it is one of the easiest chart types to understand and commonly used in the finance industry. The gif below shows Apple's drop in sentiment after the iPhone battery scandal, drawn from our data set.
The above is how the graphs look in the final prototype. It has a few salient features:
- Drastic changes in sentiment (+/- 0.2) are tagged.
- Sentiment is expressed in natural language (neutral, negative, very positive).
User experience: Integration
I integrated our product at two different widgets to maximize access and increase the <10% access rate to Signals.
- “Signals” widget: The most logical place for the product to sit, as Signals deals with ephemeral indicators like search engine referrals and social media likes.
- “News curated for you” widget: Provides easy integration into a default widget and subverts PitchBook's highly top-down information architecture.
Within the news widget, our product is expressed as a Likert scale label for that article, along with a spark-line.
User experience: Full widget
Sentisage incorporates net sentiment, positive sentiment, and net sentiment. Negative sentiment is prioritized because that's what's more interesting to a VC.
User experience: External access
News access is provided for each data point. The user chooses the data point and can click on the most positive and negative sentences for further exploration.
User experience: Time scaling
Users can dive deep into the data based on the following features:
- Easy changes to time scale
- “Handles” to scale
- Automatic graphical scaling
Time scaling allows the user to contextualize sentiment based on time.
User experience: Company comparisons
The company comparison widget provides VC users a way to look at different companies they might want to invest in. For example, an analyst might want to compare metrics such as stock price, growth rate, sentiment for Facebook with Twitter and Instagram when considering investing in tech.
The salient features are:
- Compare companies
- Use drop down menu to draw from PitchBook companies
- Color coding
“I was very happy to have you share your insights and process with everyone. You should be proud of what you have been able to accomplish and I hope to be able to harness your insights into new and exciting product opportunities.”
- Michael Mott, Senior Product Manager at PitchBook, Team Datallica supervisor
I led my team to present our work to an audience of 40 PitchBook employees, representing the product team, data team and UX team at the company. PitchBook’s Chief of product also joined the meeting from London. Our main contact and PM at PitchBook Michael Mott gave us his feedback as below:
Once the product is integrated in PitchBook’s platform, it will help their users gauge a trend of market sentiment for different firms. They will base their crucial investment decisions on these and other data points. PitchBook would also introduce this feature during demos and POC as one of the selling points of the platform.
Mott also mentioned that this will help them get more subscriptions for the platform.