Checklist to build a chatbot that doesn’t suck

Chatbot Emoji Conversation
Mar 23, 2021
Morgane Werner
Facebook LogoLinkedIn Logo

Are you call center agents constantly spending time answering repetitive customer questions? Are your customers waiting minutes to get answers? Are people reaching out to you on channels that you don’t have the bandwidth to manage? These are all pain points that could easily be resolved with an AI chatbot.

Also known as virtual assistants, AI chatbots have become a must-have for solving common customer experience and service issues. AI bots are able to resolve many types of customer requests, from basic Q&As to qualifying and transferring requests to human agents. 

Unlike your customer service department, bots are available to answer any question 24/7. As Ian Jacobs,VP Research Director at Forrester and co-author of the report, How AI and automation drive better customer service experiences, says, “If you need an answer at 2 am, is it better to have a 40% chance of getting the right answer from a chatbot or a 0% chance of getting any answer from contact center agents because they aren’t working?”. And at Mindsay, we’re seeing chatbot accuracy rates much higher than 40%.

10 questions to answer before starting to build your bot

Here are the questions you need to answer to be sure a chatbot is a right solution for your business.

1. What business issue are you trying to solve? 

Identifying a problem or a business optimization that can be addressed by the bot is the first milestone of your project. The chatbot you want to create for your website has to work as per the specific business requirements. 

So, before you build one, it is vital to outline your goals and define the needs. You also need to know the type of problems you want solutions for.

2. What use cases do you want your chatbot to answer?  

Once you have identified the main goal, you need to design the scope of the project. Defining the precise use cases will help you build the architecture of the bot as you need to know exactly what the AI chatbot will do and why that is important.

Here are some use case examples:

  • Cancel a flight ticket
  • Track an order
  • Find a product

3. Which chatbot building platform is the best?

You have your use cases, now it’s time to choose the chatbot platform! Several choices are available: ready-to-use platform if you want to save effort and time. But if the business use case you’re trying to solve is unique or complex, we advise you to go on a platform where you can entirely customize your chatbot.

Key features to look for when choosing a chatbot platform:

  • Low-code or not, depending on your technical skills
  • Human handover possibilities
  • Data protection

4. What content do you have? 

Prior to creating a chatbot, you must list all of the questions most frequently asked of the customer service. But when you first create a chatbot, be careful not to try to do too much. It’s better to focus on the most recurrent use cases. 

Now, think about the content you already have to train the natural language processing (NLP) of your AI bot and how much you'll have to build - this will help you estimate the amount of work!

5. Do you have qualitative APIs? 

There are essentials if you want your bot to provide real-time information (such as flight information), connect to other services, or let your bot redirect users to a live agent.

6. What system integration do you need for your bot? 

Pick the systems you'll need to integrate: Zendesk to redirect to agents or create tickets? Musement to book activities? Typeform to add a modal form to your bot? - and make sure the connection is feasible.

7. On which channel(s) will be my chatbot? 

Where will your bot be available? Will it be integrated with WhatsApp? Can customers engage with it via SMS, Facebook Messenger, or only on your company’s website?

An AI-powered chatbot can be connected to any website, many messaging apps, or even on social media. Your decision must be based on the use - there's no need to have a bot on Facebook Messenger if you don't have any followers.

8. What language(s) do you want your AI bot to understand & speak? 

Usually, start with maximum two languages, and pick the ones that are mostly needed by your users. Then improve your bot by implementing new languages. Don’t worry, you will be able to add more than 10 languages with a Mindsay chatbot!

9. Is it hard to build a chatbot? 

If you have tech skills, you can always build a bot on a platform that is not specialized in your industry and without any support nor care. Otherwise, pick a provider that knows your business challenges and that will guide you from A to Z.

10. What KPIs do you want to follow to measure chatbot's ROI? 

Last but not least, a bot is far from immutable. Collecting data and following key analytics will help you improve your bot continuously based on its use. Without KPIs, you won't be able to measure your ROI.

Now that your automation project is clearer, it’s time to roll up your sleeves and start building your chatbot!

Photos gratuites de arrière-plan bleu, avenir, doigt
Time to build your AI bot

How to create a chatbot from scratch?

Building a customized low-code chatbot has never been easier since low-code bot platforms emerged. To ensure you build the right AI chatbot for your business, here is the checklist we advise you to follow:

1- Identify the objectives you want to achieve with the bot

2- Select the channels your customers are using most, or the ones you’d like to develop

3- Choose use cases that will help you answer your customers’ questions

4- Create the conversation flows of the typical questions and answers the bot will handle

5- Train your AI chatbot

6- Test the chatbot as your customers would use it

7- Deploy your chatbot on your channels

8- Constantly explore new ways to improve your chatbot so you can provide the best customer experience.

You made it till the end! Now that you’ve gone through all of the steps, it’s time to build your own chatbot 🤖.

Psssst, did you know that you can build a chatbot in just a few weeks? We’re here to help you achieve this goal! At Mindsay, we don't just build conversational AI bots, we maintain and improve them and we ensure great customer care from the beginning of the project and all along. Check out our customer stories to see for yourself!

Related articles

Get started with bots

Get Started