A talent for words: The ❤️ of chatbot user experience design
“If I’m selling to you, I speak your language. If I’m buying, dann müssen Sie Deutsch sprechen!” — Willy Brandt, former West German Chancellor
It’s not what is said. It’s what is heard (Frank Luntz)
Both humans and computers talk in a dialog. Indeed, arguably a chatbot’s most attractive feature is just that: it’s conversational.
A chatbot user experience (UX) offers new possibilities to excite and delight users if you can get that conversational content designed the right way.
“The difference between a good experience and a poorly executed one can be in the small details, such as how long the text is and whether the user has to scroll to read all of it.” — Amir Shevat, Designing Bots: Creating Conversational Experiences
How a conversational UI writer adds value
We can think of a chatbot conversation as having three parts:
- An intent (the ask or task of the user. This is not explicitly expressed).
- An utterance that the human inputs (what the chatbot smart technology processes) to articulate their needs.
- The prompts (I will include system messages, button labels, and so on in this category) that the chatbot responds with.
Of course, you need to choose a chatbot platform that has natural language processing (NLP) capability to understand what your user intends to do and make sense of their utterances before responding smartly, preferably in their natural language of choice.
Your choice of NLP should be tolerant of phrases, partial sentences, and spelling and grammatical errors, but then it’s the conversational UI writer’s turn to shine.
“Chatbots communicate in dialog; humans communicate in dialog. What could be more natural?” — Lilli Cheng, Microsoft
If you have a passion for language and writing good UX copy, then your services have never been more welcome. New career opportunities beckon. The need to design and create a great bot and messaging content is driving growth in demand for conversational UI and chatbot text writers.
Personality: An unbroken series of successful gestures (F. Scott Fitzgerald)
First of all, establish the personality of your chatbot. Because chatbots are about interacting with people in a “human, but not too-human way”, develop a style and tone of voice that fits your image but that will also resonate with users.
“To be understood intuitively is the goal — the words on the screen are the handle of the hammer”. — Elizabeth McGuane, Intercom
Simple guidelines for conversational copy
What follows are some considerations to ensure your chatbot “sounds right” as you craft an interaction with that right conversational style and tone. This all begins with knowing your users and what they want to do: the context.
Content-wise, bear in mind these points:
- Keep chatbot talk real! Design and build real text for the prompts. Avoid Lorem Ipsum-style placeholders (or worse!). “Filler” text changes the nature of the interaction and the experience. Imagine if a voiced question was met with “Lorem Ipsum” in response.
- Be concise with those chatbot prompts. It’s a conversation, not a speech. Text shown on messenger and bot UIs on mobile devices must be glanceable and listenable on the go. Get to the point — how can the chatbot help? What can it actually do for a user? Focus the chat around the agency. Don’t ramble or distract from the task or the ask!
- Use button text or other widget labels supporting the chatbot prompts that are action-oriented. Use the imperative form of a verb so it is clear to the user what selecting an option will do. Asking a user to select or say “OK” is not a great action label. Keep it short.
- Start your prompts with a capital letter and end with a punctuation mark (for example, a period or question mark), provide for an inflected voice tone with questions, and so on. Phrases and contractions, as well as full sentences and words, are fine; it’s how we converse in the real world.
- Be consistent with your words. Consistent does not mean “the same”. Don’t use single text prompts dumbed down so they are devoid of personality or context, but craft some nuances for re-use on the chatbot platform.
- Use domain-recognizable terminology for your chatbot, but also leverage synonyms for those terms, slang equivalents, abbreviations, and so on. For example, someone might ask a bank chatbot how much credit they have left, but also ask are they “broke”? It’s best to totally abstract away any tech jargon, system talk, or any GUI terms themselves and keep things plain without being boring.
- Take care with humor. It can contribute positively to the user experience, but it is a matter of context of use. Even in Japan with banking bots for millennials, humor might have a place, but it depends!
- You must be aware of cultural nuances generally and be respectful of differences without being stuffy. Using emojis with chatbot prompts can be a good way to build user rapport, but again be conscious of the context, and of platform internationalization capability.
- Admit defeat gracefully when the chatbot responds to utterances that cannot be understood or things that are just impossible to perform. Declaring you’re only a chatbot is part of affording user comfort (and you might even say you’re made of donuts), but don’t continually demand more information or just give up.
- Users are especially intolerant of voice-driven chatbots that fail to understand what they mean and unlike GUI app failures they will not blame themselves. Tell them what the chatbot can do upfront. Redirect the user to a help center or human support representative. Try something like,·“Sorry, I don’t understand. I have an easier time with a few simple words, such as ….”. Don’t shout, “System error! Contact your sysadmin!”
- Never blame or “scream” at the user by using all UPPERCASE! text or “loud” voice responses. Yelling during any conversation is unlikely to entice further engagement but does the opposite.
- Write chatbot messages that give an indication that the system is working in response to an utterance. For example, “Hang on, while I look that one up . . . ” is much preferable to the white line of death in a paused, or even hung, conversation.
- Keep any user assistance that might be offered within the conversation concise, contextual, and step-driven. Any user assistance must be about completing the task or ask at hand. You can provide links or tell the user about examples or more content online if needed.
“Conversations are about people. Making conversations feel more personal is possible with only a few data points. It implies being intentional and creative around what we know around the interlocutor, and progressively build and evolve from there.” — David Boardman and Sara Kao, IDEO
Beyond Clippy
In general, ensure that what you write is supportive of the overall user experience in content, style, and tone; reflective of why people are turning to chatbots in the first place. With thoughtful conversational-style content creation (such as using scripts), backed by user validation, a great experience can be created, iterated, and tested for your chatbot before it goes live with the conversation.
Nobody wants to be the next Clippy!
Some Recommended Reading
- Taming Text (Grant S. Ingersoll, Thomas S. Morton, and Andrew L. Farris)
- Designing Voice User Interfaces: Principles of Conversational Experiences (Cathy Pearl)
- Designing Bots: Creating Conversational Experiences (Amir Shevat)
Ultan Ó Broin (@ultan) is a user experience and product consultant with digital customers of all kinds in the U.S. and EMEA. They are a member of the editorial board of MultiLingual. Eighties music and hair.
All screen images are by Ultan Ó Broin.
Multilingual chatbot designer? This story can also be read in Spanish: Consejos de escritura para el diseño UX de chatbots.
