At Lekta AI, we’ve been building chatbots since early 2017. While each project is unique, we found that certain universal principles can be applied regardless of the business domain, target audience or communication channel. Over the years, we’ve developed a 4-step battle-tested process that helps us deliver consistently great conversational experiences for our clients and their customers.
This is the second in a series of four articles on the topic that can serve as a comprehensive resource for anyone looking to build a bot. If you haven’t read Part 1: Requirements Discovery, you might want to check it out too.
You’ve now gathered requirements, have a product vision and can move on to the next step of our process — designing your chatbot’s personality! Regardless of how tempting it might be to jump right in and start writing dialogues, you should focus on the big picture first and decide on what you want your chatbot to be like.
Why focus on personality at all? Glad you asked.
Customer experience is becoming everything.
While it’s true that your bot should be functional first and foremost, how it will solve users’ problems is no less important. Research reveals that personality plays a key role in your chatbot’s success, leading to positive user experience, engagement and retention (Tuva Lunde Smestad, 2018). If done right, it can breathe life into your chatbot and positively impact the overall perception of your product. If done wrong, it can even cost you a customer!
This doesn’t surprise us — Salesforce State of the Connected Customer Report (2019) shows that “customer expectations hit all-time highs”, and that connected, consistent and personalized experiences are the new norm. In fact, 57% of interviewees have switched to a competitor because they provided better customer service, and two-thirds are willing to pay extra for the superior treatment.
It looks like customer experience is the new battlefield, and how you handle your customers is becoming everything. More and more brands compete solely on the basis of customer service, and according to The Walker Study, customer experience will overtake price and product as the key brand differentiator in 2020.
Your chatbot is your brand ambassador.
Likely, you’re investing in a chatbot, so it can lighten your load and take over a chunk of your customer support. That means, your bot will be the first point of contact (and the first impression of your brand) for the majority of new leads. It will be, in a way, your brand’s ambassador, whose performance will reflect back on your company and have a major impact on how you are perceived. The stakes are high — get it wrong, people will turn away from you, get it right, and you won big time. Not only have you scaled customer support, but its world-class quality will make a lasting impact on many.
So, how do you ensure that your new brand ambassador can live up to the task? Just like your employees are trained on your company values and communication style to provide consistent and quality support for your customers, your bot’s personality should be on-brand too. At the very least, it should be an extension of your brand and your CS team, carrying through the same tone of voice and principles. At most, it can elevate it to the next level, become an entity in itself, the talk of the town, and the new standard for future bots to come.
People will assign human traits to your chatbot anyway.
People will assign human traits to your chatbot, whether you like it or not. Anthropomorphism is real and can’t be helped — we attribute human motivations, beliefs, and emotions to animals, objects and more recently to bots.
So, even if you don’t design a personality for your bot, people will assign it one themselves. So would you rather have a say in how your bot is perceived, or let your customers decide?
A predefined persona helps everyone be on the same page.
Last but not least, defining your bot’s personality helps everyone see eye to eye, especially in disputable matters. This is particularly important when you have multiple writers working on bot dialogues. A predefined persona helps them stay in character, and makes questions such as “How would our bot react in this situation?” easier. No guidelines to fall back on can result in inconsistent responses and your bot suffering from a multiple personality disorder.
How to develop your bot’s personality?
Now that we’ve covered why personality is important, we will share with you our tips and tricks on how to develop it. But first things first — bot persona description doesn’t need to be overly detailed or extensive. The purpose of the exercise is to define key traits that our bot should represent, and lay foundations for its mannerisms, reactions, visual design, name, gender, etc. In the end, all of these attributes will have an impact on your customers’ experience and their perception of your product.
We usually develop chatbot personalities together with our clients and follow a workshop format. Our personality workshops consist of 3 parts: recollecting product vision, defining personality using The Big 5 framework, translating personality into specific bot behavior.
Recollecting product vision
As a first step, you should set the right tone for the workshop and make sure that everybody is on the same page when it comes to the overall product vision. While this workshop is creative in nature, a certain level of agreement needs to be reached upfront, so that everyone operates within the same frames. Most of these frames were already discussed and pre-determined during the requirements discovery phase and will be brought up again for the sole purpose of laying the groundwork for the creative works.
We usually start off with your bot’s raison d’être. What is your bot’s purpose? What is its primary use case? Answers to these questions should definitely be taken into account when defining your bot’s personality. An insurance claim bot will probably need to be more serious in nature than a Netflix chatbot whose main job is to recommend your next watch.
Next up, let’s recall who our target users are. In the end, the level of humor, playfulness, and choice of words should be adapted to your customers. One way to approach this task is to analyze how your customers speak, what words and expressions they use and mimic them. Another exercise that we like to do at this stage is an empathy map. This helps everyone step in the customers’ shoes and understand how they usually feel when they’re interacting with your brand. Are they under a lot of stress because the service you’re providing stopped working? Or maybe they’re excited to buy what you have on offer? Keep in mind, that your bot’s personality will need to be balanced enough to address all these business processes and different customer moods that come with them.
Finally, (and as mentioned before) your bot’s persona should be on-brand. Ideally, you should familiarize yourself with your client’s brand book before the workshop, so that you’re aware of the brand’s story, personality, mission, and key values. It’s worth looking at the brand’s emotional benefits — that is, how should a customer feel after interacting with your client’s brand? What might come in handy are standards set for your customer support team too. What words are they encouraged to use, and how are they supposed to behave? Here we might be looking for tips such as “always show empathy” or “advise, not enforce”. These standards and brand guidelines will be a tone-setter for your chatbot and its behavior.
Defining your bot’s personality with The Big 5
Once everyone is on the same page, you can finally get your creative juices flowing and begin working on the personality itself. We’re lucky to have a few psychologists on our team, and while we worked with several personality frameworks before, the big 5 has become our all-time favorite. It’s an empirically supported, five-factor model assessing personalities based on 5 dimensions: openness, conscientiousness, extraversion, agreeableness, and neuroticism. We will briefly explain what these dimensions mean, so you get a better idea of what this framework measures. You can also get a more detailed view of The Big 5 here.
- Openness — represents our openness to changes and new experiences
- Conscientiousness — reflects how responsible, organized and hard-working we are
- Extraversion — refers to how we recharge, and how outgoing and sociable we are
- Agreeableness — entails our tendency towards being cooperative, trusting and empathetic
- Neuroticism — encompasses our tendency towards stress and anxiety
Now, you may wonder if there’s any point debating these characteristics at all. Shouldn’t every bot be open, conscientious, extraverted, agreeable and non-neurotic? Not necessarily. For starters, you want your bot to stand out and be slightly different than all bots out there. Secondly, different use cases call for different bots. As is the case with humans, where a highly extraverted person is more likely to do well in sales and marketing, while a more introverted one would be a better fit to resolve technical issues, so is the case for bots. You don’t want your bot to be overly excited when your customer is calling to say that your service is down. You also don’t want it to be too agreeable if your customers tend to ask for home service when you know that their problems can be solved online. Let us juxtapose the high and low scores of these traits and take a closer look at how they may exhibit themselves in a conversation.
As you can see, there’s room for everyone on this planet — even more neurotic and not-so-agreeable bots you wouldn’t think of at first.
Translating personality into behavior
Now that you’ve agreed on what you want your bot to be like, it’s time to translate these traits into a specific behavior. Remember, that you have limited tools to showcase your bot’s behavior. Predominantly, you’ll be relying on dialogues, but don’t forget you have emoji (or even memes!) at your disposal.
Now, let’s think of what type of behaviors your bot will usually be engaged in. There are quite a few standard situations every bot will have to deal with. To begin with, every bot will need to introduce themselves and find out why the customer is calling. Every bot will invariably fall off the happy path at some point too, misunderstand the user and cause her agitation. Your bot needs to be prepared for these instances and solve them with grace.
During the workshop, we usually focus on these 4 flagship areas: welcome message, self-service, misunderstanding and agitation, and write a few sample dialogues, so that everyone can see how a newly created bot persona would act in those situations.
Following the workshop
Following the workshop, it’s good to sum up what you achieved and agreed on, as likely it will be a lot! We usually send a client a short document featuring their bot’s personality and a few sample dialogues. This summary will be useful for anyone working on the project, especially the dialogue writers.
Speaking of dialogues, now that you have your personality you can finally begin writing them! If you’d like some practical tips on how to write well, check out Part 3, Chatbot Conversational Best Practices.
Alternatively, you can put more effort into your chatbot and develop its world. You could come up with an entire story: where it’s from, what its hobbies are. Don’t forget, that you will also need a name (an article on best naming practices is coming too).
Do comment if you have any questions, and make sure to check out Part 1, Requirements Discovery, if you haven’t already.
