Should you make your chatbot a “Star Wars” fan?
Should you make your chatbot a “Star Wars” fan? Should you even invest in its personality? The topic has been widely covered on the web, but here’s a new spin to it. We interviewed Anna Andraka — a bot personality creator and copywriter, who also happens to be… a playwright.
Anna tells us why your bot’s personality is key to creating an engaging conversational experience for your customers, and explains how to develop it.
Anna worked with Lekta on a project for the largest telecommunications company in Poland — PLAY; she co-created PLAY bot Danny’s personality and wrote most of his dialogues.
Could you tell us if and why bots need a personality?
People tend to project human traits onto animals and objects all the time. Notice that we usually speak to our dogs as if they were human. We beg our phones to work and scream at our iPads if something goes wrong with them. The same applies to bots: when your customers start a conversation with a bot, they humanise it and expect a human-like interaction.
Another thing is that the world is just full of bots these days. If all of them were dull, they’d be very similar to each other. Hence, investing time and resources into developing your bot’s personality is yet another way to differentiate your brand.
You should also remember that your bot is not only a communication channel but also a brand representative. Or even an ambassador!
Just like you’d want your human brand ambassador to know your company values, your bot should live and breathe them too.
How do you go about creating a bot’s personality? Where do you even start?
Remember that, while talking to users, bots only operate in dialogues. The aim of creating a bot’s persona is to be able to create responses that don’t exclude each other.
Bot’s persona is what its attitude towards books, movies or sports is.
Think of it as a brand book that you’d always have in the back of your head.
You should start by coming up with a general vision for your bot. The way I see this, the process is similar to creating a fictional character. Think of many playwriting exercises out there — for example, 50 questions, that help you define your character’s views. Only when you have that vision, should you be adding in more specifics, such as character traits. Remember: what people do in professional life can have an impact on how they behave, and how they are. The same goes for bots!
As a rule of thumb, a bot should be consistent with the brand. This means that your bot should be in line with a wider communication strategy applied to brand communication. A bot can enhance and strengthen your brand’s voice, so this isn’t just about creating a fun communication channel, it’s about creating a consistent experience for your customers.
Well, it does sound complicated! So when we finally come up with that, how do we “inject” personality into the bot? How do we pick the right words?
Let me show you this on Danny’s example. While working with Play and Lekta, we wanted Danny to be an abstract being, born from code and algorithms. Even Danny’s avatar is an abstract shape!
Right from the beginning, everyone who talks to our bot should understand that Danny is not some kind of a wanna be-human being. Hence, its name — “Danny” derives from the word “dane”, which stands for “data” in Polish. Then, we decided on its character traits: Danny doesn’t treat himself seriously. It’s funny, self-deprecating, but also open and direct.
You have to create this whole persona, so you can come up with answers to users’ questions that go beyond standard, service-related queries.
Luckily, the Play team is ambitious and open to unconventional ideas, so we were able to create a truly extraordinary bot personality!
Now, how does this whole process translate into what Danny says? I’ve got some examples for you:
User: Do you like movies?
Danny Bot: Well, I think I might be from space, so how about “Star Wars”?
User: $^#&*#!
Danny Bot: Whoa, that wasn’t nice! But it won’t get to my head, because I don’t have one ;) / Ouch, I feel an error coming!
User: Tell me a joke.
Danny Bot: I don’t need any jokes to make people laugh… especially when I don’t understand what they’re saying :) / OMG, Danny! They didn’t program any jokes for you!
On that note, people often start testing chatbots’ capabilities by asking them surprising questions. How do you prepare for that?
Well, users’ creativity can be truly disarming! They will ask the most random questions: from “Do you have a cat?” to “Hey Bot, is there a Ms. Bot?”, or “What’s on Mars?”. This is natural — we’re testing the capabilities of a bot and what it can understand. As a bot creator, you can never predict all possible questions people will ask.
But the basic question you have to ask yourself is whether you want your bot to discuss off-topic questions at all. There are different opinions here, but I think choosing to do so helps bring your bot to life. So, how should a bot respond to off-topic questions? Well, we come back to the notion of personality here — either a bot likes films or not, either it bikes or it does not. These preferences are then translated to responses. The challenge here, though, is that we can’t recognise all the questions 100%. It’s important that a bot’s answers are specific enough to make the user feel understood, but at the same time generic enough to fit as a response to a broader range of questions.
You see, Danny knows things because it is a virtual being living in a digital space. So it is aware of “Star Wars” — there’s no way to miss it since it’s all over the digital world — but it shouldn’t claim that it’s a fan. Why? Well, we don’t want to risk users asking it about a favourite prequel, or who Rey’s parents really are — the whole conversation would quickly get out of control.
General and witty answers are the best: Danny’s a robot, so it would probably like Lem’s “Fables for Robots” or maybe Spielberg’s “AI”, and it should say that when asked about its favourite books or movies.
Are there any pieces of dialogue you’re particularly proud of?
What gives me the biggest pride and joy is when I see that the dialogues I wrote work well in real conversations, and not just on paper.
That is, when I see that users enjoy the chat, when I myself am surprised how cleverly it matched the answer to a question, and when the conversation is going on for a while and the bot doesn’t fail. This is when I get the feeling of a job well done.
As a playwright, did it ever cross your mind that you’d be creating a fictional character of a bot?
Well, not really! But when an opportunity for a bot job came up, I was extremely excited to see how principles of playwriting can be translated into the business context. As it turned out, there are a lot of similarities between creating a fictional literary character and a bot.
But there are also differences, of course.
In a theatre play, the dialogue is fixed, but in the bot context, the situation is completely different. A bot lives and breathes its own life, it’s unpredictable, and so is the direction of the dialogue. This sometimes happens in theatre too, when the audience gets engaged, but it’s a rarity. I’d say this is the main difference.
What skills should you look for in a bot copywriter?
I’d say a mixture of playwright and marketing skills, and language intuition. Remember that usually, a bot is some sort of a marketing tool, and there are certain do’s and don’ts in that space you need to manoeuvre around. Commercial goals related to bot’s personality can present a difficult challenge for someone outside of the business world. So, an artist or a playwright may not necessarily always be your best choice. A happy medium between art and business acumen in a hire is what you should look for.
And if this person also happens to have a strong language intuition, then you know you’ve found your bot copywriter. Both in case of a voice and text bot, you need that intuition to predict what will sound well and what would fall flat. Dialogue always needs to have a certain smoothness to it.
Also, remember that jokes that are funny and witty in a text message, may not translate well to a voice channel. Emojis and emoticons add a lot, and sometimes something really funny on Messenger can turn into an ultimate dad joke in the voice channel.
What is one final piece of advice you’d like to give our readers?
I know that “personality” is a scary word. But remember: it’s not about creating a whole human being from scratch. It’s mostly about coming up with a communication channel that is both useful and fun to use, so it stands out and makes people remember your brand.
So, in the end, all it takes is to ask yourself: would I want to talk to this bot? Would I enjoy talking to it? If the answer is “yes”, then you know you are on the right track.
