More and more brands want to communicate more personally and efficiently at scale. And they want to use chatbots, voicebots and conversational display ads to do it. To be able to write successful dialogue for conversational interfaces such as WhatsApp, Messenger and Google Assistant, together with my team of seven copywriters I became a certified Conversation Designer at the renowned Conversation Design Institute.
We love conversational
First off, I might have to explain the Conversation Designer title. Since it has virtually nothing to do with design in the art director’s meaning of the word, but is all about designing, crafting great conversations between AI Assistants and consumers. But I can imagine the term might put you on the wrong foot. But now you know. Talking about AI Assistant: there’s not a country in the world where the number of Google Homes has risen so fast and so exponentially as in the past few years in the Netherlands. April 2020, approximately 1,5 million homes (!) had a smart speaker such as the Google Home. And besides that, Dutch consumers are really getting used to talking with (not ‘to’!) chatbots on websites and through WhatsApp.
No surprise there, that this rise of conversational would have its effect on the number of AI Assistants created by Greenhouse_Conversational, proudly one of the frontrunners in this domain. Because my team of copywriters had to be ready to take on all types of conversational challenges ahead, we completed our Conversation Designer training early October. And now we’re skilled writers for all types of conversational solutions from relatively straightforward voice search and FAQ projects to full-blown voice brand experiences and customer contact automation.
It’s a copy thing
Obviously, the area of Conversational feels as comfortable for copywriters as painters feel with a canvas and paint. Creating an idea for a dialogue that has to go from A to anywhere in three interactions or less is a challenge we were born to take on. I mean, we’re so used to writing dialogue for online video, radio and TV ads. Trained to do it in the brand tone of voice, nicely gift-wrapped in a great storyline. So designing a conversation is something you undoubtedly can trust us with. We’ve been doing it since the very beginning of marketing and advertising, right?
What do the AI assistant and human need to have a smoothly flowing conversation? (Copyright: Conversation Design Insitute)
But as we’ve discovered (again): this kind of writing conversational requires is in fact slightly different from what we’re used to. As always (at least partly) because context is king. Meaning that writing dialogue that you put into the hands of code, AI and an internet connection changes your writing process. Apart from the copywriting thing we know how to do, conversational writing requires technological, methodological and psychological knowledge to an extent I couldn’t have imagined beforehand.
Writing for robots
We used to write dialogue for actors that you could brief. I would sit in a sound studio or on a film set for hours and tell an actor to emphasize this or that word more because it was so essential for the impact we were trying to get with consumers. Now, a robot can in fact be briefed as well, just in a different way. Not on paper, but through code. Instead of voice coaching, I’m applying SSML. Code such as: ‘’ to let a synthetic voice take a short break between words to make it sound more human or to add drama. Both the actor and the bot need to understand the context in which dialogue is taking place. Both have to understand the timing, speed and pitch of what they’re saying. Conclusion: what us copywriters need to get the results we want is not that different from what we’re used to, but the way we do it is unique to the user context, the interface and what AI requires to make the dialogue flow smoothly.
The Conversation Design training and certificate by CDI Academy has become the national and international educational standard for conversational copywriting. In a still growing number of lessons, copywriters are being trained to write natural (human), user-centric and effective dialogue. Subjects (among others) are:
- Identifying and creating proper use cases.
- Psychology, tech, organization and copywriting of conversational projects.
- Developing user and bot personas, based on brand guidelines, data & insights.
- Sample dialogues, flowcharts and expert re-writes.
- Wizard of Oz testing and user tests.
- And many conversational copywriting techniques.
Sample dialogue: back-to-back conversation, one person is playing the consumer, the other is playing the bot, to decide which conversational flow will work best. (Copyright: Conversation Design Insitute)
One Breath Test
One of the conversational copywriting techniques we’ve learned is used specifically for voice: the One Breath Test by the Conversation Design Institute. If something that’s being said by the AI Assistant takes longer than one breath, then it’s too lengthy for voice, and it will be edited. The editing, then, can be done using the Jenga technique. Which, just like we do in the famous game, is about removing as many parts as possible, but at the same time making sure the construction (messaging/dialogue) will still stand.
Machine Learning is rapidly improving speech recognition and conversational possibilities. Which is why users will not hear “Sorry, I’m afraid I didn’t get that” as frequently anymore. In most cases people actually get a relevant response to whatever they’re saying. (Which is not really guaranteed in conversations with actual humans.)
But obviously bots still do misunderstand people. (But how often have you felt misunderstood by a human spokesperson or contact center representative?) And if AI Assistants don’t instantly get what you’re saying, they’ll give you a fallback response*, after which the conversation will move on fluidly. This fallback response is formulated in a unique, brand-specific way, owing to the brand and bot persona cheat sheet right next to the keyboard. Giving each bot its own idiolect (words, catchphrases, emoji’s) to add soul to dialogue that’s perceived as unpersonal and aloof simply because it’s coming from a piece of tech instead of an actual person. It’s the copywriter’s responsibility and skill to design AI Assistants that are emphatic, human, helpful, and thereby likable.
From campaign to contact center
Greenhouse_Conversational can design effective concepts, technology and copywriting for all each and every automated conversation imaginable; and for the full marketing funnel, from awareness to retention. Take the award-winning ‘Auping’s Bedtime Stories’ - a voice experience. Or how we automated FAQs for big recruiter brands Randstad and Tempo Team, by designing a chatbot that answered all easy questions 24/7 and consistently, and handing over the harder questions to human contact center agents. For Red Bull we made their ‘Athlete Challenge’ more personal and relevant by letting Olympic gold medal winner Kjeld Nuis coach them 1-on-1 through voice.
Evolution: how organizations grow on their way to Conversational maturity. (Copyright: Conversation Design Insitute)
Award-winning? Thirst for more!
As one of the more experienced copywriters in my team, Renske van den Bogaard has completed the Conversation Design training too. This year, she’s won a pretty big creative award: a golden Spin for the Donald Duck action we've made. Still, she really wanted to get her certificate. To me it felt a bit like asking Virgil van Dijk to take a soccer clinic, but in the end Renske really benefited from it a lot, saying: “Sure, I’d seen a number of principles and methods come by during projects, more like on the go. But during the training they’ve been placed in a broader perspective, and now I understand the psychology and tech behind it so much better. In a way that it’s also become easier for me to work together with tech and strategy specialists in our agency. At the end of the day, that will make what we do as a team more efficient and effective.”
Become a winning conversational company
Challenge Greenhouse_Conversational to find the best way for your brand to talk with its audience 24/7: firstname.lastname@example.org.
*Fallback response: when the user doesn’t answer at all, or in a way the AI Assistant doesn’t understand, the bot will react by shortly repeating the question (‘rapid reprompt’), or adding more information to the question (‘escalating detail’).