A. List of data-points (market data information on individual stocks) that users evaluate on a regular basis; uncovered both from the survey and dashboard.

B. Next, the team wanted to understand the competitive landscape to determine how our competitors were handling responses regarding market data-points. Specifically to understand how our competitors treated questions 1 & 2. I evaluated 9 different skills in our landscape, like Bloomberg and the Motley Fool; as well as 4 skills outside of our landscape such as ESPN, Spotify, and others.

C. Our team believed that to differentiate ourselves from competitors, CNBC's Alexa skill needed to leverage our unique brand into our conversational experiences. Years of research supports that CNBC's audiences engage with certain personalities such as Jim Cramer. Each personality brings a distinctive tonality in how they speak and describe market data.

I observed several CNBC anchors to understand their tonality.

A. Ideate:To start I listed out all the data-points that we defined earlier in the process and designed all valid responses for each data-point. These responses were designed after gaining insight from the competitive analysis and CNBC's anchor observations.

B. Creating a System: Once we had all the responses for each datapoint. The product owner and I broke down each response to identify variables, i.e company, datapoint, monetary value, verbs (that described the monetary value) and timeframe.

This would allow us to create a modular systems where we could interchange certain variables; for both design & engineering efficiencies.

C. Incorporate:CNBC Into the Conversation: Once we had all the responses for each datapoint. The product owner and I broke down each response to identify variables, i.e company, datapoint, monetary value, verbs (that described the monetary value) and timeframe.

This would allow us to create a modular systems where we could interchange certain variables; for both design & engineering efficiencies.

A. During the first phase of the activity, I showed participants cards; each card containing a datapoint (front), and our designed response (back). I asked them to describe the datapoint, and then asked them to assess the response I designed.

B. Next was figuring out what were the value thresholds for verbs that describe; Very Positive, Positive, Negative and Very Negative.

For example, do we only use "Very Positive" verbs such as "skyrocketed and surged" for values greater than "x" percent? We weren't sure so we asked SME's to shed some light on this.

C. An example of a response for a datapoint, volume, going through iteration once I had feedback from our SME’s.

Once our responses became validated, next was to support certain edge cases, i.e "What is Apple's stock price", in the event of a 52-week high, holiday/event, or a trading halt. Those edge-cases needed to be supported to avoid keeping the user from being misinformed. This all became documented through various conversational diagrams. I created conversational diagrams, collaborating with my engineering lead to ensure that the conversation and system logic happened in an order that made sense.

A. Early wire-flows I designed for the conversation experience. The wire-flow shows how a GUI + response changes over the time of day.

B. High-fidelity mocks of the GUI experience, indicating that users can receive updates on popular market events such as earnings season.

C. Most recently, to expand this experience, our team incorporated 'News About a Stock', a feature I designed that delivers users news headlines about any stock.

D. Four months after the launch of the feature, both engagement and growth on Alexa have had major increases.