Conversational AI in the Age of Hyper Customer Reviews
Consciously or unconsciously, we are living in the age of hyper customer reviews. Customer reviews have become the new currency for companies who have the know-how on capitalising customer reviews as a competitive tool. Customer reviews as a standalone tool is not competing head-on with traditional corporate branding and marketing efforts — rather it enhances the branding and marketing in terms of providing actionable insights (e.g. what the customers are suggesting), real time feedback (e.g. most customer reviews are fairly recent experience recollections), and high contextual depth (e.g. there is a lot of meaning behind customer reviews that can be extracted).
In this is article, I hypothesise that in the age of hyper customer reviews, conversational AI will become more prevalent in meaning-making (i.e. drawing deeper conversations with customers, rather than just one-way write-and-submit review) and product-expansion (i.e. potential co-creating new product with customers through deeper interactions)
Based on the hypothesis above, I am posing a few forward looking statements, and corresponding questions on Conversational AI:
- Human has a natural tendency to generate imagination and insight as we go through our subjects of interest. Specific to Conversational AI: What will be a good approach to capture this generative data insight and feed back into the dataset?
- There is a lot of research about human-in-the-loop in machine learning. I am supportive of this movement/ research. Instead of looking through the lens of the more popular Generative Adversarial Network (GAN), can we make AI more collaborative. Specific to Conversational AI: Can there be such approach as General Peripheral Insight Network (GPIN) whereby human helps to generate more insight to feed into machine?
- The expansionist view touches on human ability to imagine and take on a creative curve into unknown. Specific to conversational AI: What is expansionist view (as in product-expansion) in the age of AI / machine learning?
A Better Way to Designing and Developing Products in the Age of Conversational AI
Interactive.
Generative.
Iterative.
The three words above summarise product characteristics in the age of Conversational AI (note: this is not meant to be exhaustive, but an attempt to formulate the core of characteristics).
In the age of Conversational AI, designing an interactive product become an expectation / norm, rather than an exception. For example, Amazon 4-Star stores display a couple thousand product lines with an average 4-start-plus customer rating (that’s based on customer reviews) on Amazon.com. The display prominence is given to lines currently trending up. A case in point for conversational AI and customer reviews: Conversational AI consume and process a vast amount of structured and unstructured (mostly!), and abstracting these customer reviews sources into something meaningful and actionable.
In the age of Conversational AI, designing a generative product become a competitive edge. The gap between product development (conceptual, roadmap, etc.) and product delivery (actual making of the product / service) is in fact narrowing. In fact most advanced product companies already started to merge these two functions to tap into spirit of rapid new product generation. A case in point for conversational AI and customer reviews: Conversational AI has capabilities to automate communication (in this case, potentially analysing sentiment and polarity of customer reviews real time), and create personalised customer experiences at scale (in this case, potentially scaling A/B testing by interacting with customers). What do these capabilities translate into? There are several paths, and I will discuss one path here. From the interactions data extracted from customer reviews, we can unearth customer insights by using Sentiment Analysis (opinion mining), Phrase Net (connecting key words in a text using lines to show linkages), and Word Tree (displaying of the words in qualitative dataset, where frequently used words are connected by branches to the other words that appear nearby in the data).
In the age of Conversational AI, iterative product design and development are possible now. A case in point for conversational AI and customer reviews: Conversational AI becoming even more relevant because through understanding customer intent, and its corresponding sentiment analysis (of valence, polarity, and reasoning), we can keep iterating and pivoting product features/functions according to different customer personal/segments.
A Better Way to Designing and Developing Products in the Age of AI
After a decade of working with product managers, interaction designers, and software engineers; I have come to realise the combinatorics power of design thinking, design sprinting, and agile development. Henceforth, prior to deep diving further into Customer Reviews and Conversational AI, I do like to take a moment to share a better way to designing and developing products in the age of AI. The intent of the diagram below is to bring to the front the combinatorics power of design thinking, design sprinting, and agile development — keeping this sharp and short: The D2A AI Product Development Model.
The D2A AI Product Development Model taps into the interactive, generative, and iterative nature of AI products:
Interactive = Design Thinking.
Generative = Design Sprinting.
Iterative = Agile Development.
The outcomes from Design Thinking and Design Sprinting feeds into Agile Development.
Side Notes on Design Thinking
Design Thinking focuses on empathising customer needs through re-framing problems from customer perspectives.
Side Notes on Design Sprinting
Design Sprinting focuses on answering business and behavioural questions with customers.
Side Notes on Agile Development
Agile Development focuses on iterative process to meet changing customer needs while ensuring sustainable development cycle.
Designing and Developing Conversational Customer Reviews Bot
This section covers the practical case study of designing and developing conversational customer reviews bots through the D2A Model. Let’s call this bots Sparky.
First, Design Thinking helps to sharpen our focus on the customers needs, wants, and aspirations. It is about empathising with the customers, in order to define and ideate Sparky products possibilities, prior to taking on the prototyping the product concept. At this stage, most of the activities revolves early stages of toying with abstract Sparky product ideas bubbling through going through human eyes (i.e. examining the subtle and non-subtle cues of the customer reviews), and machine learning data (i.e. examining the word clouds, sentiment analysis, etc.).
Taking the cue from Design Thinking’s abstraction of Sparky product possibilities, Design Sprint helps to sharpen our focus on speed, iteration, and pivot of these Sparky product possibilities. A 4-day or 5-day product development sprinting helps to focus the dreamers (e.g. interaction designers, product anthropologists, etc), and implementers (e.g. business sponsors, product managers, etc.) on the customers problem at hand. The speed of mapping the problems and solutions helps the business to pivot, if needed, towards a more feasible, viable, and profitable product — a minimum viable product (MVP).
With a Sparky MVP in sight, the baton will be passed on to the software teams to develop an actual MVP conversational customer reviews. Agile Development helps to sharpen our focus on iterating the product as we learn through further customers feedback.
TAKEAWAYS
As customer reviews continue to play a bigger role in corporate world (part of branding, marketing, customer service, and customer success effort), so will the voice of the customers who understand the importance of their reviews (this is their new currency). Sparky demonstrates that customer review bots not only have a place in this highly competitive corporate race, but they also will spark (pardon the pun) greater appreciation of customers in this age of AI.
The following will be in other articles where I will explore the following:
- The sociological view of human conversational in the digital age.
- The customer reviews as the new currency in the age of machine learning.
- The counter-intuitive concept of General Peripheral Insight Network (GPIN) whereby human helps to generate more insight to feed into machine.
So look out for it.
END NOTES
This article is part of the Product Data Scientist Pocket Guidebook Series (Please check out a similar guidebook series with more focus on machine side of data science — Data Scientist Pocket Guidebook Series).
Hopefully, this will be a handy reference that will help you navigate the basics on trending and challenging data science and machine learning topics. Ideal for aspiring data scientists and machine learning engineers who wants to get pro-tips and case studies.
