Revolution in Conversations Powered by Business Intelligence and Analytics
‘Problems exist in the absence of conversation.’
According to Gartner Research the value of the Business Intelligence and Analytics (BI&A) market was worth US$16.9bn in 2016. Yet, this huge market hardly caters for Conversational BI&A. As more and more people are starting to believe conversational interaction will become the dominant interface worldwide, now is the time to start thinking about Conversational BI&A.
The smallest unit for measurement is called a conversation-step or dialogue-step. The conversation-step is in text format, which may have been converted from voice. It contains the date and time stamp, plus the verbatim narrative, and in the case of a choice it also includes the options. Other data or media orchestrated within the conversation-step are also included.
The conversational-string is the collection of conversational-steps in the sequence from the start of the dialogue to the end-point. An end-point maybe a premature end of the conversation or could be an outcome such as a completed transaction or a hand-off to a person.
The conversational-string may be linked to a Channel (such as Facebook or Skype) and may transverse across Channels. In addition, the conversational-string may transverse across different Chatbots such as a hand-off to a specialist Chatbot.
Conversational BI&A is being provided by the tech titans such as Amazon, Facebook, Google, WeChat, Apple and Microsoft. Specialist offerings have also emerged such as:
Adobe Voice Analytics: Adobe Analytics Cloud can now track the performance of voice-enabled intelligent assistants like Alexa, Siri, Google Assistant, Cortana, and Samsung’s Bixby. This is Adobe’s first voice and conversational analytics capabilities within its cloud service.
Dashbot: San Francisco; Raised US$2m Funding: Chatbot consumer metrics, sentiment analysis, conversational analytics, and response effectiveness. Dashbot natively supports Facebook, Slack, Google Home, Alexa and Kik. It also supports a generic API for any conversational interface.
Botanalytics: San Francisco; Raised US$302k Funding: Aimed at Chatbot developers to obtain analytics about its conversations with consumers. Identifies bottle necks, filter conversations and the level of engagement. Botanalytics supports Chatbot platforms from Microsoft, Azumo and Converse.
Bot Metrics: San Francisco: This provides a middleware for consumer conversational analytics for Facebook, Kik, Mobile, Web and SMS. The Bot Metrics middleware can be set-up on Amazon, Heroku or Linux.
Manner AI: London: This provides a consumer split testing and personalisation capability for Chatbot engagements. The analytics and visualisation enables responses to be analysed and improved using deep reinforcement learning. The goal is to optimise conversion rates.
Botan.io: This provides analytics for Chatbots using the Telegram platform. Their focus is on consumer segmentation and profiling and determining which events are the most popular features of the Chatbot. Botan.io uses AppMetrica for consumer-centric analytics.
Octane AI: California; Raised US$1.5m: Have a feature called Decision Data that tracks the performance of quick replies and buttons used in the dialogue which Chatbot users choose most often chatting with a bot.
The challenge for Conversational BI&A is to action the insights gained. This is particularly difficult for natural language processing, when it is purely evolved through machine learning. The ability for machine learning to unlearn and relearn is highly problematic, especially as the conversational threads become longer and deeper.
Those Chatbots that deploy human controlled machine learning and blended the conversation with scripts and events are more likely to use analytical insights to develop more efficient and effective conversations, especially for conversational commerce and conversational advertising. Amazon is a natural market leader in this space as it is so focused on successfully growing conversational commerce using Alexa and Echo. According to Consumer Intelligence Research Partners, LLC (CIRP) released analysis of Amazon Echo from Amazon, Inc. in May 2017, there are over 10.7 million USA customers using Echo.
The reason why Conversational BI&A works so well for Amazon is the way they have developed conversational commerce around the power of each conversational-step. Down at this level of granularity, there are up to four key elements Amazon uses for orchestration within the conversational-step:
- INTENT: An intent performs an action in response to natural language user input (e.g. Book Hotel); in other words, the intent is looking for purpose of the conversation.
- UTTERANCES: This is the spoken or typed phases that invoke an INTENT e.g. “I’d like to book a hotel”.
- SLOTS: This is the data that needs to be input required for FULFILLMENT of the INTENT.
- FULFILLMENT: This is the mechanism for achieving the FULFILLMENT and includes links into Amazon shopping cart, payments and delivery.
Combined with these elements of the conversational-step is other data such as:
- Customer account details.
- Channel for the conversation such as Echo.
- Geo location of the conversation.
Clearly, Amazon Conversational BI&A is already powerful enough to drive more and more conversational commerce. The other tech titan that is already highly successful in conversational commerce is TenCent’s WeChat in China.
One other area of Conversational BI&A relates to RegTech Chatbots pioneered by Df2020 using scripts for simplifying and streamlining conversational workflow. In their case they are using productivity business intelligence for conversational-step and conversational-pathway utilisation and optimisation, decision risk mitigation and outcome analytics. Df2020’s user generated Chatbots are new forms of knowledge assets, which the conversational data is used for audit, compliance, metrics and patterns for knowledge learning at the edge.
‘As the “Revolution in Conversations” continues to change everything, this new type of Big Data combined with Business Intelligence is going to be an accelerant for change.’