Danielle Morris
Conversational UI
Creating a digital whisky companion that educates consumers on scotch and The Glenlivet brand, and connects with them wherever they are
SECTOR
E-commerce
client
Glenlivet - A a renowned distillery producing the best-selling single malt globally
challenge
Create an automated conversational experience that uses dialogue to answer users’ whiskey queries in real time. Building deeper personal relationships between the brand and the consumer, helping the average consumer build their confidence and knowledge of whisky.
timelines
Three weeks
the team
Senior UX Design & Research (me), content writer, developer.
activities
Sketch personas (based on consumer types)
Chatbot prototype (using Motion.ai and LUIS natural language processor plugin)
UX writing
User stories
Landscape analysis
User flows (conversational paths)
Stakeholder workshops & prioritisation
Established conversational principles and tone of voice (with the brand's SM manager)
Guerilla user-testing (in-person)
outcomes
A working Facebook chatbot experience filled with content, stories and tips that users could only discover through conversation. A proof of concept prototype for internal use by the client for stakeholder demos.
The main learnings were around the constraints of natural language processing and the complexity of plotting even a simple conversational path to cater for all possible interactions and still keep it sounding human.
content-led strategy
The Glenlivet is the best-selling single malt whisky in the world. As the global digital agency partner for the brand, we (Zone) were appointed to develop a content-led CRM strategy that reinforced The Glenlivet’s positioning as the original single malt – the one that all others are measured against.
Whisky is a complex and highly involved category. Consumers range from novice to connoisseur, and aficionados typically have a range of different bottles in their drinks cabinet. Personal recommendation is a key driver of consideration and purchase intent. Could a chatbot operate in that space?
.jpg)
An automated conversational experience
Help the average consumer discover the pleasure of drinking & build their confidence and knowledge of whisky
Demonstrate The Glenlivet’s knowledge and expertise in the single malt category
​
Use conversation to answer users’ whiskey queries in real time... building deeper personal relationships with our audience
learning by doing
Due to a compact timeline & budget constraints, we were operating on a skeleton team of just one UX, one developer & one content writer.
​
Chatbot experience was lacking at the agency and it was 2017 so AI wasn't a thing... yet.
​
After a quick scout around for non-expensive and platforms we could use out of the box, we settled on Motion.ai - a drag-and-drop interface that works on multiple channels e.g. Facebook and web.
We combined this with LUIS, a Microsoft plug-in that parses responses ("user intentions") looking for trigger words and selects an appropriate chatbot response - natural language processing.
​
It also records user entries with unsuccessful matches which we could use to iterate the flow.
what we did
Landscape analysis - there were a surprising number of whiskey chatbots out there already!
Established conversational principles and tone of voice with the client and the social media manager.
​
Sketch personas capturing the different consumer types, on a spectrum from novice to knowledgable enthusiast. Focusing our thinking around what content is appropriate, what products they would buy, and what product is the next step up for them. How can we help a novice find an aspirational but affordable Glenlivet bottle by flavour, for themselves or as a gift? How can we help an enthusiast identify their "dream dram" and direct them to an event where they can try before they buy?
​
Gathered content - Motion.ai is relatively limited in terms of pre-made modules you can use - we needed to be clever with the logic of the user flow options & the editorial copy. We looked for opportunities to introduce more natural language processing (or pre-made modules that simulate this e.g. yes/no responses)
​
Mapped out the full user journey and chat experience, to test using the walkthrough method and enable prioritization sessions with the client to establish which conversational paths would be released first.
​
Built a working Facebook prototype for guerilla testing and for internal use by the client for stakeholder demos
_edited.jpg)
enjoy A wee dram...
We released a Facebook chatbot experience filled with content, stories and tips that users could only discover through conversation. The companion that helps you learn and enjoy whisky in new ways, no matter where you are...
​
To give you a flavour, here's a short video of part of the experience (from within the Motion.ai app).
​
The client were thrilled with this proof of concept that encapsulated the quality and heritage of their brand, and have continued to iterate over the years (and incorporate AI).
