Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Chatbot interaction design — IxDA London July 2017

879 views

Published on

Now that chatbots are seeping into the mainstream, interaction designers are starting to receive briefs to design and improve chatbot conversational experiences on apps, websites, and more.

That begs the question: What is interaction design for chatbots?

These are some thoughts from my recent work writing UX guidelines and recommendations for a large high-street client. I'll address some basic thinking and approaches to designing chatbot experiences, focusing on practical rather than future-forward applications.

Published in: Design

Chatbot interaction design — IxDA London July 2017

  1. 1. Boon Yew Chew IxDA London Lessons from writing chatbot UX guidelines 26 July 2017 Uncovering chatbot IxD basics
  2. 2. Lessons learnt Popular chatbot literature isn't helpful for design. Make your own guidelines.
  3. 3. Chatbot interaction design Facilitating conversation- based interactions through dynamic text and visual- based dialogue
  4. 4. Side-stepping misconceptions Not everything needs AI Chatbots failure rates are high Personality is only the tip of the iceberg Bot conversation involves a learning curve
  5. 5. Let's design to... build confidence meet expectations enable (re)usability
  6. 6. Let's avoid... designing for novelty majoring on personality being trapped by the tool
  7. 7. Discovery: 2 things to understand User expectations and needs AI strengths & weaknesses
  8. 8. Broad perceptions and expectations People expect chatbots to be... useful for assisting in basic tasks and queries efficient and get things done faster straight to the point ...but not to... answer all their questions fully understand their responses be very intuitive or reliable offer opinions be too informal or expressive Chat Bots: A consumer research study, myclever.com (2016) Chatbot Strategy: Market research report, Fifthquadrant (Nov 2016)
  9. 9. Expected qualities of a chatbot Chat Bots: A consumer research study, myclever.com (2016) Chatbot Strategy: Market research report, Fifthquadrant (Nov 2016) Knowledgeable Helpful Informative Problem solver Responsive Friendly Polite
  10. 10. Uncovering specific user needs Input only trial Chat logs Call centre logs Top tasks survey etc.
  11. 11. Novice vs. expert users “Just give me the answers. I’m not interested in chatting.” “It’s nice that I can ask it things, and get the answers right away.” New to chatbots Will default to common sense approaches May give up early and not persist May be open to chatter Strong expectations for chatbot behaviour May attempt a few ways to get answers Speaks in a way that’s appropriate to task Less patience for chatter
  12. 12. Understand AI strengths and limitations Firstly, build / test and validate the experience Understand cognitive technologies, and how it impacts the experience Know the difference between effort to understand vs. effort to answer
  13. 13. Do an audit Does this chatbot experience match user expectations and needs? What’s getting in the way of the conversation? What’s making the conversation feel odd, pointless, wrong? Is the AI weakness at fault? What is it good at? Is it the content that needs fixing?
  14. 14. Writing guidelines and recommendations Strategy Craft Interface
  15. 15. Chatbot strategy 101 Does it need to exist, and why? What problem does it solve? Who is it for? How will it support users? How does it fit within the wider experience? Strategy
  16. 16. H I G H U S E R E X P E C T A T I O N L O W U S E R E X P E C T A T I O N C H A T B O T W E A K N E S S C H A T B O T S T R E N G T H D E L I V E R A N D D E L I G H T S U G G E S T A N D G U I D E L I M I T A N D C L A R I F Y P A R T N E R A N D R E D I R E C T
  17. 17. Confidence is the currency of conversational experiences Complex responses Open-ended questions Expressive personality Too much assistance Limited knowledge Clinical personality SWEET SPOT V E R Y C L O S E DV E R Y O P E N Craft
  18. 18. User confidence comes from Knowing what the bot can do Feeling understood Successfully completing tasks Ease of use Craft
  19. 19. Communicate limitations and capabilities Offer structured, suggested answers Employ conversational repair techniques Gradually provide tips and suggestions Provide short, concise responses Provide shortcuts and quick commands Support chat saving and resuming Recall and reuse user data Integrate useful services Supporting novice and expert users Craft
  20. 20. Three layers of chatbot experience Conversation (text, rich media) Container (chat window) Platform (website, app, Facebook messenger) Interface
  21. 21. Consider the ecosystem Platform 1 2 3 Universal CTA Footer Promo Hero In-page "How can I help?" "Who would you like to speak to?" "Try me out" "Let me introduce myself" "What about this can I help you with?"
  22. 22. A landing page! Platform 1 2 3 How does it work? Who is it for? Why is it good? What can it do? ...
  23. 23. Layout strategies, behaviours Container
  24. 24. Keep it simple Container
  25. 25. Emulate real conversations Conversation
  26. 26. Introduce complexity gradually Conversation tips and suggestions asking for feedback rich, dynamic UI complex logic
  27. 27. Good defaults Conversation examples up front structured prompts concise and simple avoid repetition
  28. 28. more advanced stuff... Conversation self-correction multi-line handling context tracking 'undo' transparency, privacy and security ...
  29. 29. Work to the AI from the user experience, not vice versa “better use cases – those real-world tasks and interactions that determine everyday business outcomes – offer the biggest payoffs” https://hbr.org/2017/04/ai-wont-change-companies-without-great-ux
  30. 30. Let's avoid making more human puppets
  31. 31. Thank you. boon.chew@gmail.com @boonych

×