AI Retail web assistant

DEMOCRATISING WEB INTERACTIONS

Position.

For many large retail businesses, conversion rates are mediocre. Over time, many pages become bloated with too many features, verbiage and too many things vying for one’s attention. Perhaps there’s a better way of interacting with the web, one that can afford ‘everyone’ the same benefits with the same amount of effort.

Problem.

Older users and those with a variety of disabilities clearly do not experience the web the same way as everyone else. Many digital tools are learnt on a trial-by-error basis. Discovery and task-based scenarios are often difficult and time-consuming, resulting in incorrect decisions being made due to interpretation and location issues, resulting in confusion and poor brand experiences.

  • Website too difficult to use

  • Navigation is either hard or complex

  • Specific content is hard to find

  • Chatbots are perceived to be good for basic written communications

  • Half of all users utilise web chat

  • No loyalty incentive

  • Lots of users didn’t trust they could find a good deal so phoned Call Centres

Possibilities.

Identify a typical user-issue and explore how an AI/VA might improve
matters for all users.

  • Create an experience that can work for all users, is potentially a faster route to success and conversion rates

  • Finding the right content becomes natural/requires little effort (creating ‘dashboard mental model’, raising awareness of e.g. renewal dates)

  • VA will learn about user and present relevant and timely content

  • Improve emotional reactions and engagement with brand

  • Agent will be helpful rather than cross-sell/provide irrelevant information

  • Agent to be more like ‘call centre’ associates (listening/intelligent)

  • Responsive and attentive

  • Agent finds genuine best deals, creating trust and alleviating call pressure on Call Centres

  • Create a better VA than current examples

  • People with previously reduced access to our content and services would finally have unbridled access

  • Anyone with an accent or speech impediment would not have an issue using this system which would provide subtitles for each word spoken asked or given

  • VA could even be international, acting as a translation device if needed

  • Technology is out there, needs adapting to a high-level AI paradigm

  • Inclusivity is the right thing to do

How might we improve upon current VA models?

How might we improve BT’s CX?

Exploration document: VA proposal hypothesis / Why a VA? / Current AI usage / AI development agencies / Chatbots / Concepts / Conversational retail experience / VA architecture

Concept 1, VA as Digital Lifeform

  • VA represented at top of screen as digital lifeorm

  • Actionable options appear underneath - User can speak-select an option or click with their mouse

  • Chosen option appears in speech bubble below, with voice detection icon active/or not

  • User account details below

Concept 2, VA as Character

  • VA represented at top of screen as an animated CGI cartoon character

  • VA audibly asks a question, user speaks or click-selects an answer

  • Users response becomes active below VA’s speech bubble

  • Stage area is where content is presented by VA for User to choose, browse or ignore

Contact me