Ever imagined what it’d be like if your customers could reach out to you on WhatsApp?

They wouldn’t have to undergo the boring, painstaking process of calling you on your customer care helpline, writing to you via your website or directly emailing you at contact@abcd.com. With the WhatsApp Business Solution, your customers are only a WhatsApp message away from getting the help and support they need from you.

Given that WhatsApp occupies a large portion of an individual’s time with their mobile device, it is critical that companies acknowledge this behavior and design initiatives to fully leverage these growing trends.

Once you implement the WhatsApp Business Solution, your customers need not go through a long IVR process, only to realize that they have to wait for another 20 minutes to get hold of a customer executive. They can simply send their concerns to the company using WhatsApp and the query will be addressed within seconds by an AI-powered WhatsApp Bot.

Now you may be thinking, “Great! I definitely need a WhatsApp bot. But how do I get started with building one?”

Well, when it comes to creating a WhatsApp bot, the first and foremost consideration is design. There are a lot of UX considerations to keep in mind while designing an automated experience for WhatsApp.

We have already discussed the process of developing chatbots and designing an intuitive UX, in our guide to designing smart chatbots. In this article, we will draw upon our experience with creating WhatsApp bots at Haptik to provide a step-by-step guide that should be a good starting point for anyone trying to design a WhatsApp bot.

UX Considerations/Limitations to Keep In Mind for a WhatsApp Bot

A WhatsApp Bot is fairly different from most other kinds of chatbots you might be familiar with. Typically, UI elements such as buttons and quick replies are a key aspect in streamlining a chatbot experience. UI elements not only help to propel the conversation forward but also help the user to achieve their end goal in the quickest way possible.

But that is not the case with WhatsApp bots.

While designing a WhatsApp bot, one has to operate within certain UI limitations and build an experience sans quick replies, buttons, and carousels.

As of today, WhatsApp only supports text. Therefore, any WhatsApp experience needs to be designed based on textual inputs from the user. This means the bot flow will be user-driven (Users typing and asking questions) and not bot-driven (The bot suggesting intuitive quick replies and buttons).

This also implies that the WhatsApp bot should handle diverse unstructured inputs by users since that is the only way to move a conversation forward.

It is also important to first build simple use-cases that do not require a lot of back and forth between the user and the bot. The more the number of interactions, the higher the risk of bot errors, which might lead to a bitter first-time experience.

So how does one create conversational flows in WhatsApp?

While designing conversational flows in WhatsApp, there are primarily four aspects to keep in mind.

1. Menu
2. Questions
3. Tips
4. Bot Errors

We will examine each of these in turn.

1. Menu

At the start of the conversation, the user should be shown a concise menu. The menu should contain the top functionalities of the bot. Like any chatbot experience, this menu is no different.

Each item listed on the menu should be clear and easy to understand.

The menu selection can happen in two ways.

1. The user needs to either type the menu item. For eg. Benefits of WhatsApp. Since a user is required to type, the menu item should be concise so that it is easy for a user to type.

2. The user can simply type the item number to move forward, like in an IVR system. (Users are habituated to use numbers to proceed).

The menu serves as the first message that will be seen by the user. Therefore, sharing a useful tip right at the beginning is a great opportunity to educate the user.

An informative tip like ‘Type main menu’ to go back at any point in the conversation will come in handy for the user.

2. Questions

Questions are a great way to steer the conversation in the right direction. The questions need to be relatable and intuitive. Forcing the question just to move the conversation forward might lead to a substandard conversational experience.

In fact, questions or commands are the only way to keep the conversation going on WhatsApp.

Let’s take a look at the Types of Questions:

a. Yes/No Questions

Questions that trigger a yes/no response are an easy way to engage the user and push the conversation forward.

Example: Would you like to learn more about Chatbots?

The answer to this question might seem straightforward. But that is not entirely true. A user can say Yes and No in so many different ways.

Yes: Ya, yup, yeah, yah, ok, sure, alright

No: Nope, nah, not really, not interested

Tip: The bot should be trained to understand agreement vs dissent. There are some obvious cases which might fail with this approach — for e.g. ‘Why not’.

b. Suggestive Questions

Suggestive Questions are a perfect replacement for buttons. It is ideal to present options and let the user define the next course of action.

Example: Are you interested in chatbots or voice bots?

A good way to emphasize the selection is by using bold.

Good news: You can use * bold * text on WhatsApp.

Example: Are you interested in chatbots or voice bots?

You can easily observe the substantial improvement in readability.

3. Tips

Providing tips throughout the conversation is a must. A seamless two-way interaction on WhatsApp is only possible if the user is aware of how to use it.

Let us now consider the Types of Tips:

a. Menu

Type ‘go back’ to see the main menu.

OR

You can simply type ‘Menu’ to select any other option.

b. Question Tips

Tip: You can also ask me questions like ‘example’

4. Bot Errors

The bot break aka error messages should be carefully worded. It should first acknowledge that the bot cannot help with the query. It should then focus on redirecting the user back into the conversation. Either redirect the user to the main menu or into another conversational flow.

The possibility of Bot breaks or errors will be slightly higher in WhatsApp Bots due to the potential of unstructured flows.

To sum up…

Keeping in mind the framework we’ve discussed, we must remember to keep the design of the WhatsApp bot as simple and as easy to understand as possible. The simpler the design, the more intuitive it’ll be for your customers, thereby resulting in a better overall experience for them. And let us not forget that the successful implementation of a WhatsApp bot solution will result in significant cost savings for your business!

Even for an experienced bot-builder, creating a WhatsApp bot for the first time can be a challenging experience. There are limitations in terms of the specific subject matter, and one has to constantly operate within a narrow framework in order to achieve the best possible results. It’ll be interesting to see the elements/features that WhatsApp decides to support in the near future. We’re hopeful that they’ll make it easier for developers to build bots compatible with their ubiquitous platform.

If there is one aspect of this entire process we’re 100% sure about, it’s that WhatsApp bots are here to stay!

This article has been penned by Nameer Khan, Program Manager at Haptik.

This article was originally published on https://haptik.ai

Designing a WhatsApp Bot: What to Keep in Mind was originally published in Chatbots Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.

In recent times, we have seen both sides of the chatbot phenomen — failed attempts by a few global brands to create an ‘AI’ powered chat experience, and successful ones by some others leading to millions in revenue. Yesterday, at the landmark F8 2017, Facebook launched Messenger 2.0 with the impetus to encourage brands to further enable & enhance chat support.

The steady rise and adoption of chatbots has helped Facebook Messenger increase interactions between businesses and people. With a staggering reach of 1.2 billion users, there are 2 billion messages exchanged between brands and their customers on a monthly basis. It is safe to say that people are warming up to the idea of using Messenger over other platforms to interact directly with brands as highlighted in the infographic below.

The first day of the conference ended with the keynote of David Marcus, VP of Facebook Messaging Products:

“If you direct your existing or future customers to Messenger instead of directing them to a mobile web page or an app, you will definitely see lift, if you build the right experience.”

The basic tenet of this premise stems from the break in the user’s journey, created by redirecting him/her to an app or a website (avg. 40% drop off after Facebook ads redirect to third party platforms). If a business can target, address, and acquire potential clients on a single platform, it already has the recipe for success in place, and further efforts to drive traffic to a website or app would be futile.

Prior to the release of Messenger 2.0, there was an air of excitement as well as uncertainty about chatbots.

1.Is it practical to build your own AI from scratch?
2.How will users find bots on Facebook?
3.Does every chatbot need to be customized? Or does one size fit all?

Messenger 2.0 has released features that unravel the answers to a myriad of questions everyone was asking.

Should AI be built from scratch?

Building a core artificial intelligence engine which can be deployed for use cases such as augmented reality and NLP is next to impossible.
Yesterday, Facebook announced several developments in terms of AI.

First off, let’s talk about AR studio. It is one small step for augmented reality, and one giant leap for developers. Developers, for the first time, will have access to a mainstream platform to integrate AR with their projects.

Smart Replies, another new feature debuted at F8, will help small business owners focus on their businesses by letting the bot automate responses to the most frequently asked questions, such as business hours, directions and contact details based on the information provided on their Facebook page(s).

Facebook’s ‘M’ is on track to taking AI to the next level by understanding user intent and context from natural language, and thus, effectively recommending relevant results, suggestions, bots, and tools.

These tools are not easy to build. They need Terabytes of data, years of data crunching, heavy duty computing systems, best machine learning algorithms and top talent around the globe.
Yeah, you guessed it right, this is something only tech giants such as Facebook and Google can do justice to.

What does that mean for chatbot makers? Stop investing in AI?
No, we need to utilise these world class platforms provided to us by global leaders. These platforms are not solutions, but means to reach the same. In markets like India, with it’s complexity by region, industry and demographics, it is imperative that localization and customization is built on top of these platforms offered by Google and Facebook.

That’s exactly what we do at my firm, Mpower.chat. We have proprietary NLP (Natural Language Processing) tools and platforms aimed at creating training sets in controlled environments, especially when there is limited user interaction data that is actionable. We then club this with leading NLP platforms to deliver top-notch results.

How will users find bots on Facebook?

Facebook is focusing on making Messenger easier to navigate. It’s adding a new feature called ‘Discover’, through which you can look for different Messenger tools, bots and brands. There’s also a new tab for games alone.

Does every chatbot need to be customized?

In my opinion, use cases such as customer support and purchasing products will mostly be similar across businesses, especially in the same industry.
Chatbots are an additional touch point in the customer journey. Not only do businesses need to build new experiences such as NLP for chatbots, but they also need to be able to recreate the existing experiences present on web and mobile applications directly into the chatbot.

At Mpower.chat we have created a WordPress like infrastructure. Our core chat engine works like a central CMS, and added features such as cart architecture, online payment & lead management systems are used as plugins to create added functionalities.

CHAT EXTENSIONS: THE WOW FACTOR

“Currently experiences with chatbots and businesses are mainly single player and one-on-one,” Marcus explained. “We want to change that with chat extensions by empowering developers to bring their experiences in existing one-on-one and group chats.”

With the example of Spotify, Marcus showed us how the true potential of chatbots can be unleashed in peer-to-peer and group chats, by creating business and developer-made experiences such as suggesting and playing live music within the chat itself.

Chatbots are thought of as humans interacting with bots. But think of a possibility when humans are interacting with humans and bots are assisting and enhancing their conversations.

So, what are bots? Human-like computers?

In the simplest terms, you could say they are AI based applications to automate processes. Today, chatbots are available on messaging platforms, websites and personal assistants such as Siri and Google Home. But tomorrow, as technology grows, they will be part of Augmented Reality and perhaps, human-like personas in Virtual Reality.

The infographic below, showcases the growing success of chatbots across industries. There is enough evidence to show businesses that introducing automation through chatbots leads to significant improvements in customer support, sales and employee productivity.

Well, the proof is in the pudding or in this case, the numbers. A far cry from when Messenger’s journey first began, it has now come a long way since being the bashful new kid on the block, putting all rumours and second-guessers to rest. Messenger has made it clear that their renewed focus is towards bringing communities together, and it shows — with truly phenomenal advancements like making an everyday smartphone camera the first vehicle for widespread AR.

Though these advancements seem like Facebook has reached for the stars, in truth, we know we’ve just scratched the surface (as is the case with all newly constructed technologies). Still, it’s a giant first step towards actually realizing some pretty big VR, AR, and AI goals. What we have only been able to fantasize about, and watch in sci fi movies, is that much closer to reality (nerds, rejoice!).

Let’s see if Facebook can become to the world, what WeChat is to China or (if I had to take a guess), if they go above and beyond. One thing’s for sure, it’s a great time to be alive and in the Bot Business!

Drop by Mpower.chat and write to me at nameer@mpower.chat to learn more about what we’re up to. Thanks to Chantelle Menezes for editing and Hrushikesh Mande for suggestions.

Facebook Messenger 2.0: The Game Changer for Chatbots was originally published in Chatbots Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.