10 Steps to Define Your Chatbot Strategy
As any software project, chatbot development goes through a set of standard stages: Chatbot Strategy, Design, Development, and Testing. Usually, the better you define your strategy, the smoother and faster your project will go.
Chatbot strategy has a lot in common with mobile and web project development. However, it has its own features to consider. Here are 10 steps to implement these features and succeed in building a chatbot strategy.
Step 1. Define Your Goals
Before you develop a chatbot, you should outline your goals. Usually, companies create chatbots to drive sales using messengers, improve brand’s online presence, provide users with a personal human-like assistant, or automate specific tasks such as customer support or the processing of user queries.
On the other side, your business goal is not the only priority. You should also align your chatbot strategy with your customers’ needs. Sometimes they can even overshadow your own goals. By solving the problems of your users, in this case, you create a business opportunity for yourself.
There are some specific tasks that chatbots can do better than other interfaces. In particular, conversational design allows bots to collect important information from users and provide a relevant, useful service to customers.
The true benefit of the chatbot, however, is its time and effort saving skills. Bots are able to guide users and help people solve problems without wasting time on searching and browsing through websites or apps.
While defining your goals keep in mind chatbots’ strengths and plan how your business can benefit on them.
Outcome of this stage is a project goal statement.
Step 2. Understand Your Users
Understanding your users’ needs, behavior, and expectations is one of the keys to success.
If there are different user types within your brand target auditory, it’s necessary to identify them all from the early start. When it’s done, you can figure out who your bot interacts with and how the bot can enrich relations between these people and your brand. Classifying your audience is one of your major tasks, because such an insight will help you keep your chatbot strategy and product focused and help you deliver effective experience.
Usually, your database and Google Analytics are the best available sources to learn your user base. Using the data you collected on your website or/an app, you will be able to study the patterns of interaction between your audience and brand - demographics, geo location, time and frequency, etc. Based on these insights, you can decide either to target one audience cluster or build a conversational interface that covers several user groups.
Outcome of this stage is the portfolio of your key audience segments.
Step 3. Learn from Competitors
It’s important to analyze your competitive landscape when you start any project. Even though it can be tricky to use competitors as a source for inspiration in building a chatbot strategy.
Chatbots’ popularity is going up. But the amount of real-life examples is not enough. Nevertheless, it’s a good idea to try different chatbots, regardless of the industry. Testing allows to try conversational interface and, possibly, come up with an idea how take advantage of it.
The most important part of this step, however, is to identify your unique proposition and make it drive your bot. Sometimes, a website or an app can have many features, but miss the ones of core value. The bot, on the contrary, may have better chances to broadcast your unique services over the world. This is why it’s important to focus on the things you do better than your competitors in the first place.
Outcome of this stage is the summary on competitors product and market strategies.
Step 4. Pick a Platform
There are many platforms where chatbots can reside. For example, popular messengers such as Facebook Messenger, Slack, Telegram, Skype, Kik, Line, etc. Also, you can build a bot for web, email, and SMS.
The best way to decide on the platform for your chatbot strategy is to see what your customers use. You may not have relevant data on your users, of course. Then, you can address to available public information on the demographics, age, trust, etc. of messengers’ users to match it with the data on your target audience.
In fact, you may need to build a chatbot for more than one platform. The good news is that modern frameworks for bot creation help developers scale one chatbot for several platforms at a click. Compare it with custom development for several mobile OS and you will see a viable opportunity for saving on cost.
Outcome of this stage is the list of platforms for your chatbot project.
Step 5. Capture Requirements
There are 2 ways how you can capture the requirements for your chatbot project.
If it’s easy to identify the user groups for your chatbot, you can apply a standard framework for user stories. Such framework forces you to think from the user’s perspective and define a separate set of requirements for each user group. A user story has a format similar to this:
As a <type of user>, I want <action/some goal>, so that <outcome>.
Example: As a mom who travels with kids, I want to get fare alerts from airlines so that I can book a flight ticket with a good discount.
If your bot is focused on completing small tasks, but targeted at a larger audience, it is better to use jobs to be done framework. This framework is focused on the event or situation, motivation and goal, and the intended outcome. Job stories have a following format:
When a <situation>, I want to <motivation>, so I can <outcome>.
Example: When I plan my vacation ahead, I want to get fare alerts from my airline so I can buy money-saving tickets.
Outcome of this stage the document with project requirements.
Step 6. Prioritize Your Desires
The process of capturing the requirements unlocks your creativity. However, you shouldn’t forget the speed to market principle. The faster you launch your chatbot, the faster you get feedback from your customers.
Moreover, it’s always better to start with a small project and improve it over time, rather than to invest much before you can validate your hypothesis.
As soon as you capture the requirements, sort them out by "Must Have" and "Nice to Have.”
“Must Have” requirement is any requirement that absolutely has to be delivered for the project to become successful. This is a sort of requirement that brings the biggest value for your customers.
“Nice to Have” requirement is a complimentary functionality. It’s rather desired, than needed.
In general, sorting out process should go side by side with feature complexity analysis. If some of “Nice to Have” features appears time-consuming, it makes sense to postpone its implementation to speed up the first issue of your project.
Outcome of this stage is the prioritized list of requirements.
Step 7. Consider Brand and Build Your Bot’s Personality
Chatbot is an additional way of interaction between your customer and your brand. This is why this experience must be consistent with the other elements of your brand’s style.
Why you should think about the bot’s personality and tone of voice while building a chatbot strategy? Because instead of visual interface, your bot will use conversation. In this context, the tone of voice you apply should resonate with your brand’s communication style and the expectations of your target audience.
If you want to add more personality, you simply create the bot’s character with a story. However, the bots with personalities should never imitate humans. It’s more about a personal touch. Writing a script, keep the bot’s personality in mind. Thus, your chatbot’s messages will convey its emotions, behavior, and temper.
Usually, AI chatbots have more room for personality development. Applying this technology, you open the door for humor, quirks, and felicitous bot replies.
Outcome of this stage is the script for a bot.
Step 8. Design a Conversation Flow
We’ve already pointed out that a chatbot is not about visual interface, it's all about conversation. Either your bot is supposed to be AI or rule-based, depending on your chatbot strategy, its conversation flow should be elaborated carefully.
First of all, you should have a proper onboarding to introduce bot’s functionality at the start of a session. It’s important to minimize user effort and build only clear and unambiguous bot messages.
Every sentence your chatbot sends should to be carefully thought through. Avoid gender-specific pronouns and open-ended questions. To build a more natural conversation flow, diversify you bot’s replies as much as possible.
Building a dialog tree you should make sure that every possible user request is covered by a part of your chatbot’s logics and user won’t get stacked at some point. If you think that you may miss something, simply leave the option to contact a human agent to avoid a dead end.
Outcome of this stage is the conversation tree chart.
Step 9. Select Appropriate Technology
Nowadays there are many tools that help to jumpstart a bot project. In fact, the right choice in terms of tools will help you save time and money and launch your project faster.
There are 2 major categories of these tools. The first one is available do-it-yourself platforms. These tools work best for simple chatbot projects. The other one includes different NLP engines that help developers enrich their chatbots with natural language understanding features.
DIY platforms allow to integrate a source code that will send data from your web server to your bot. This is how it can display this data to your customers.
The major feature of these platforms is a so-called CMS style functionality that helps you draft a dialog and configure dependencies using visual interface. Often, such platforms provide a set of integration capabilities with third party components and messengers for faster deployment.
Among these platforms //chatfuel.com/ is one of the most popular. But it is applicable only for Facebooks and Telegram bots. There are similar tools such as Motion.ai, Flow XO, Botsify, and other platforms with visual interfaces for building a chatbot.
The more complicated your project is, the more complex tools you need to implement it. If you want to empower your chatbot with NLP capabilities, you will need to integrate it with NLP engines such as API.ai, LUIS.ai, Wit.ai, Watson Conversation, or Amazon Lex.
These tools help bots understand user input in natural language and trigger required actions from your source code. In other words, your chatbot will be able to hear what your customers say.
Outcome of this stage is a technology stack selection that will help you reach your goals in the most efficient way.
Step 10. Take Analytics into Account
You will want to know how good your chatbot is. To monitor its performance, you need to choose a proper tool for analytics. The tool that will help you keep an eye on the way your customers interact with the bot.
There are plenty of tools designed for chatbot analytics. Compared to a well-known Google Analytics, monitoring platforms for bots mostly focus on conversations.
Relying on bot analytics, you can adjust the scripts for complicated questions, control new/repeated/abandoned conversations, identify features with high engagements, track user information and input, analyze failed responses.
There are several worthy analytics platforms:
- Botanalytics is a good tool for gathering conversation history and monitoring real-time data. It has advanced features that allow to compare different conversation flows and identify the features with high user engagement.
- Botmetrics is an open source tool that helps you track basic metrics about bot performance.
- Botlytics is a relatively simple SaaS platforms that monitors bot activities and provides practical insights.
All in all, the right tool for analytics is the way to make informed decisions and continually improve a chatbot.
Outcome of this stage is the selected tool for analytics.
Moving step by step, you'll get a solid understanding of why you need a bot. You will create your chatbot strategy. This journey will help you find out what a bot should do and how it should work.
Once you are done, you will be ready to start your chatbot development project. With this strategy, you will discover opportunities and unlock a new channel of communication between your brand and your customers. Who knows, maybe one day your chatbot will earn you a million or solve serious problem on a global scale.
Would you like to have a custom chatbot strategy developed for your business? Let's talk about your project.