Conversation Intelligence Concepts

The Goal

The goal of this concept car was to deep dive into a fantastical future-state of our AI functionality.  We wanted to explore weaknesses in our existing offering and idealize a repackaging of our Conversation Intelligence product for discussion.

‍

The Process

We started with research on where the product was falling short.

We quickly found a gap between the value our paying customers described vs the perceived value a trial user described. It was not clear what aspects of our platform were related to this product and what benefit it would bring.

‍

The Concepts

Problem 1: Value Prop is not clear

Proposed Solution: 

Double down on messaging: 'Save time by automating the classification of your leads.'

‍

Implications:

• Rework Keyword Spotting feature to work as a set of automation rules based on any criteria vs transcription keywords. [Shipped]
• Build word cloud visualization so users can quickly see trends on phone calls and in text messages. [Shipped]
• Highlight conversation-based notifications (ie. when a customer says X, send me an email).
• Highlight conversation-based tags (ie. when a customer books an appointment, tag 'Appointment Set') [Shipped]
• Enhance tag-based reports & report filters, assuming users will often build automation rules on tags.

‍

Problem 2: Data is not immediately actionable

Proposed Solution:

Make the value of the transcription-based data as clear and actionable as possible, making assumptions based on research to give users the insights we expect them to glean from the data.

‍

Implications:

• Create widget on dashboard to show insights from Conversation Intelligence.
• Create 'Customer' as third phase of the lead funnel and spend time honing our algorithm to predict customer. [In Progress]
• Build report that illustrates how leads move through the funnel. This can be used across marketing to show how users can see Customer count per campaign/ad vs Raw Lead count.

‍

Problem 3: Setup is complex

Proposed Solution:

Make research-based suggestions, and allow for advanced users to customize.

‍

Implications:

• Redesign automation rules flow to improve usability & create suggested templates that lower cognitive load. [Shipped & Shipped]
• Suggest automation rules when user has conversations that match common criteria.

‍

Output

To disseminate the concept and get feedback, I mocked up what a marketing landing page would look like for the product, with quick user goals, value props, and product shots. See below for pieces of that mockup that were broken down into individual slides for deeper discussion.