What is a chatbot? And how to design a chatbot that converts
What is a chatbot?
At the most basic level, a chatbot is a computer program that simulates and processes human conversation (either written or spoken), allowing humans to interact with digital devices as if they were communicating with a real person.
To date, mobile messenger apps have become hugely popular. In fact, messaging apps now have 20% more active users than social media platforms. With 55% of consumers preferring to communicate with businesses over messenger apps over other communication channels.
“People are now spending more time in messaging apps than in social media and that is a huge turning point. Messaging apps are the platforms of the future and bots will be how their users access all sorts of services.” — Peter Rojas, Entrepreneur in Residence at Betaworks
We are rapidly hurtling towards a world where chatbots are the virtual face of customer service and sales teams, PAs and engineers. That’s why you need to be able to design a chatbot that converts.
Where customers completing a full sales cycle without ever leaving the chatbot environment is the norm, rather than the exception.
So it makes sense that the demand for chatbots is growing.
Right now, 55% of consumers would rather conduct all of their customer service interactions on a messaging platform. So, what happens if your chatbot’s UX is clunky? Or if you don’t have one at all?
You can bet you’re missing out on a lot of opportunities to create meaningful interactions with your customers. That’s why it’s crucial to design a great chatbot with fantastic UX that your customers will actually enjoy interacting with!
But where do you start? How do you even design and launch a chatbot?
How to Design a chatbot that converts
Best Practice Chatbot Design Checklist
- Relevant functionality and a distinct purpose
- An on-brand chatbot persona
- Punchy copywriting & engaging tone-of-voice
- Personalised messaging
- Appropriate wait times
- Simple, easy-to-locate CTAs
- Useful prompts and re-engagement options
Ok, let’s dive into these in a bit more detail!
Tip 1: Define your chatbot’s purpose and functionality
Before you start to build your chatbot, you need to answer a few crucial questions.
First of all, what do you want your chatbot to do? This might seem like an obvious question, but there’s a huge variety of options open to you, and you need to make sure your chatbot serves the right purpose.
Where will your chatbot be placed? Will it be part of an app? Will it be on your Facebook page? Will it only be present on product pages? Or will it be present on the homepage, or even across your entire website?
Think about the types of users that will be visiting those pages, which stage of the Pirate Funnel they are at and what type of information or service they might require from your chatbot. From here, you can set the purpose of your chatbot. Here are some examples:
- Giving users information about pricing
- Troubleshooting issues with a current service or product
- Collecting feedback on a product via a short survey
- Processing customer complaints
- Connecting a user with a real salesperson
- Suggesting products a user might like
- Scheduling a meeting
Think about the “what, who, when, why” and “where” of your chatbot.
Tip 2: Develop your chatbot’s persona
Your chatbot needs a friendly persona with a human touch. Although most users will be aware that they are interacting with a chatbot and not a real person, humans still value that real, “personal” touch. Drift’s 2018 “State of Chatbots” report noted that 43% of users they surveyed would prefer to deal with a real-life assistant. But the beauty of chatbots is that it doesn’t have to be an either/or situation. You can use a chatbot that answers simple questions or provides information, but that passes the user on to a human when they’re needed.
Take a look at Cleo, a UK-based money managing app. This chatbot has a distinct “female” persona and uses British phrases such as referring to saving money as “tucking away”, to appeal to a UK audience.
The more that we, as marketers, can design a chatbot that imitates the natural flow of a conversation between two friends, the more likely users are to have an enjoyable experience. And this all starts by building the right chatbot persona.
First of all, your chatbot’s persona needs to align itself closely with your overall brand’s tone of voice, but it doesn’t need to match it exactly. In fact, it needs to be more “helpful” that your core brand persona, because its sole purpose is to help visitors to your website.
It’s a good idea to decide on a single, easy-to-recognise individual persona rather than your brand as a whole. This will make your chatbot seem much more personal, as it feels more natural to have a conversation with one individual than a whole organisation.
For example, if you run a dog grooming business called Fluffee, you might want to design your chatbot’s persona to be called “Dan from Fluffee”.
Tip 3: Incorporate your brand’s tone of voice (and punchy copywriting)
Your chatbot’s tone of voice will mostly be dictated by:
- Your brand’s tone of voice
- The persona you developed in the previous step
But there’s still a lot of cool ideas you can play around with!
Consider using emojis to brighten up your copy, or appeal to a younger audience (just try not too sound too much like a Dad who’s just discovered his emoji keyboard). After all, emojis are a core part of the modern digital language, with over 900 million sent every day, without text, over Facebook Messenger. Here’s how Sprout Social uses them to appear friendly and approachable:
You can also experiment with using a few well-placed GIFs as responses or prompts too.
When designing your chatbot, keep the copy as short and to-the-point as possible. On a desktop, chatbots generally appear in a small pop-up dialogue box in the bottom right corner of the screen. If a user is interacting with your chatbot on a mobile device, there will be about as much space as a typical text or WhatsApp conversation. So remember that real estate for your dialogue is in high demand. That means no lengthy paragraphs.
Think about UX here. You don’t want the user to have to scroll back to read the chatbot’s last message, so keep your sentences short and snappy. As always at Growth Tribe, we recommend writing like Hemingway!
Another good rule to bear in mind is the ‘Old-School Twitter Rule’. That means keeping each message to roughly 140 characters. You can send multiple, short messages, but there’s still something to keep in mind…
Tip 4: Make them wait (but not too long)
This is a pretty simple tip. When you’re talking to a friend on a messaging platform, there’s a natural pause between messages. They don’t all come through at the same time in one big, unreadable rush. Your users don’t want that either.
By incorporating natural gaps, or ‘message delays’ between messages as you design your chatbot, you give the person interacting with your chatbot the time and space to read and process the information you’re sharing with them.
Delays that are too short will overwhelm your chatbot user. But delays that are too long will cause unnecessary anxiety. Just think about how you feel when you see those “typing” dots on your screen for too long.
A good rule is to allow two seconds of delay for each one line of text. But, as always, the best way to decide how long to leave is to test, test and test again!
Tip 5: Personalised messaging
Personalisation is a great way to engage with your users and deliver those all-important ‘wow’ moments, and it can go a lot further than simply using the first name.
In fact, 71% of consumers have said that they would prefer adverts that are personalised to their shopping preferences.
Basic personalisation that you can start to use straight away could include statistical, demographic preference prediction which offers users products or services based on what other people in that demographic bracket like, or information from a database such as location, appointments or prior purchases.
Try some of these personalisation ideas:
- Create personalised shopping experiences much in the same way a real-life personal shopper would. Ask your user questions about what they’re looking for, tastes, preferences, sizing or location (depending on your product) and design a varied number of personalised outputs based on the data from their responses.
- Once you’ve got this data, you can use it to enrich the information you already have on your customer’s buying behaviour and build detailed customer profiles for retargeting.
- If you already have customer profiles matched with customer accounts, you can use the chatbot to prompt them to login. From here, you can offer them personalised offers based on the preferences and purchase history you already have. Take a look at how ASOS have done this with their chatbot.
Sephora, the makeup store, takes personalisation to a new level by allowing the user to “try on” makeup looks by opening their camera within the app. Once you’ve found a look you like, it prompts you to browse the products you’d need to recreate it in real life!
In the future, AI chatbots will be able to learn an individual’s preferences and situations and use that prior information to offer carefully tailored suggestions and advice.
Tip 6: Simple CTAs, basic multiple choice and plenty of re-engagement
When it comes to great chatbot CTAs, you need to make sure you don’t rush the user. Again, the most important rule to bear in mind is that it needs to be conversational. The more complex your product or service, the further down the sequence the CTA needs to be. You can’t start to rush your users towards a CTA before you’ve answered all of their questions, or they’ll get frustrated!
To get the best interactions with your chatbot, give your users a small selection of options that they can click on when it asks them a question. This ‘multiple choice’ style of conversation will prevent any issues with open-text answers and will allow you to design personalised responses based on the option the user selects.
You’ll also need to decide how your chatbot “closes” the conversation. Depending on the type of service it provides, there are lots of options to keep a user engaged. If the chatbot’s purpose is troubleshooting, it might as “Is there anything else I can help you with?” once the issue is resolved.
A good example of a highly-engaging Facebook Messenger Chatbot is the one by Spotify.
First, it greets me by name, before going on to show me a neat trick that allows me to create Spotify playlists in Messenger, with my friends. Simple CTAs prompt the chatbot to tell me more information about this tool. It then asks me what type of music I’m looking for, and includes multiple choice options such as “mood” or “decades”. When one is selected, it delivers a playlist I might like, then prompts re-engagement by repeating the question – therefore keeping me in the loop!
And here’s an example of a good chatbot gone bad. In the lead up to the 2016 election, NBC Politics had a chatbot on their Facebook page which provided users with customised search results. They’ve since stopped the experiment but kept the chatbot live. The result? A rather rude chatbot which explains that it no longer works, asks if you would like to see the latest edition, then claims that the action is impossible. Frustrating? Incredibly.