OBI4wan Chatbot Checklist

OBI4wan • 21 minute read
Chatbots & AI
Featured image

More and more organisations and professionals are dealing with chatbots. Creating a chatbot (or having one made for you) is really not that difficult! However, you probably might want some help with it. For anyone who wants to know more about chatbots or is looking for interesting insights and hands-on tips, we now have this cool chatbot checklist!

In this chatbot checklist, we’ll take you on a journey through the chatbot landscape. You’ll learn more about the wonderful world of these visual assistants based on the best practices from professionals, along with advice from our own chatbot experts!

Did you know, for example, that it helps to give your chatbot a personality? Or that the analysis and testing phase of the chatbot are extremely important for ensuring that the implementation and go-live of your chatbot goes smoothly? You’ll discover the best tips for every phase in the development of your bot.

Is there a ‘golden tip’ that you know about that we haven’t covered here? Then feel free to let us know, and we’ll add it!

1. What is the goal of your chatbot?

    #1 Determine a specific goal #2 Define clear tasks
    #3 Choose the right chatbot #4 Be specific
    #5 Keep it realistic, start small!
    #6 Make sure your chatbot is, and remains, scalable
    #7 Use data to predict the success of your chatbot
    #8 Pay close attention to your analysis
    #9 Be aware that a chatbot is not always the solution
    #10 Decide whether to develop the chatbot yourself or have it developed for you
    #11 Immerse yourself in chatbot technology: get inspired! #12 Make a development plan

2. A flawless start for your chatbot

    #13 A team with the right skills ensures the best results
    #14 Service employees are most likely the best chatbot builders
    #15 Do some ‘missionary work’ within your organisation
    #16 Involve your employees from the start
    #17 Involve stakeholders and ask for feedback

3. Make your bot conversational

    #18 Give your chatbot a personality
    #19 Apply the principles of conversational human voice (CHV) in the dialogue
    #20 Write your dialogue as naturally as possible
    #21 Apply human communication rules to chatbot dialogue
    #22 Take the customer by the hand
    #23 Write ‘one-breath’ text
    #24 Adjust conversations per user group
    #25 Determine what languages the conversations should be available in #26 A/B test your dialogue
    #27 Define business rules
    #28 Don’t limit yourself to flowcharts, make
    conversations that are more comprehensive

4. Testing and linking to external systems and tooling

    #29 Take the time to test
    #30 Enrich your chatbot by linking it to external systems
    #31 Have your chatbots and employees work in the same environment
    #32 Always provide an opportunity for human contact
    #33 Distinguish between virtual assistants and employees

5. The go-live and optimisation of your chatbot

    #34 Choose a moment to go live
    #35 Spread the word!
    #36 Make your chatbot even smarter
    #37 Categorise questions
    #38 Your chatbot can make mistakes
    #39 Don’t expect miracles
    #40 Interim evaluation

6. Evaluation: what did your chatbot achieve?

    #41 Measure the performance of your chatbot
    #42 Check customer satisfaction (NPS)
    #43 Have the goals been achieved?
    #44 Expanding your chatbot
    #45 Add new chatbots to your team
    #46 Happy? Share your success with others!

1. What is the goal of your chatbot?

# 1 Determine a specific goal

A chatbot without a goal does little for you. What do you want to achieve with your chatbot? Important properties and ingredients for the success of a chatbot are speed, quality and personalisation. With good application, a chatbot can quickly deliver the following:

chatbots resultaten OBI4wan

#2 Define clear tasks

Defining tasks is one of the most important steps in the development of a chatbot. What do you want to use the chatbot for? What problems and challenges can a chatbot solve for your organisation? How can a chatbot help your customers better? Think of things like opening times or requesting product information. You can include these tasks and goals in a development plan.

#3 Choose the right chatbot

The type of chatbot you choose depends on the problem or question you want to solve. Chatbots are often used for:

  • Answering frequently asked questions;
  • Referring website visitors to the correct page;
  • Routing calls to the right employee;
  • Referring to the correct information source;
  • Answering questions outside opening hours (for example during the weekend or in the evening hours).

Based on the goal, or a problem you want to solve, you can easily determine which chatbot your organisation needs!

#4 Be specific

It’s not yet conceivable that a chatbot can take over all the problems within your organisation, so choose specifics — it will increase the success of your chatbot. By starting with just one task first, it’s easier to measure what’s happening, whether it yields efficiency and whether customer satisfaction has increased.

#5 Keep it realistic, start small!

If you’re not completely confident about the technology, start small. For example, you can start with using a bot outside of opening hours. This type of bot is very accessible. It’s not really necessary to change your entire customer service in one go. Divide it into small, manageable steps and measure the result, so you’re able to take more efficient steps and then grow from there.

#6 Make sure your chatbot is, and remains, scalable

Make sure the chatbot can grow along with your organisation. To achieve this, make a list of requirements that the chatbot must meet for your organisation. Do not settle for a simple chatbot that can only provide standard answers to questions, but invest in a platform that ensures that your chatbot can “stay with the times” and continue to work for you in the future. Because the world of technology
is changing so quickly, more and more possibilities come up. Find a platform that fits your organisation and that also has integrations with other platforms, such as Google and Genesys.

#7 Use data to predict the success of your chatbot

By analysing past customer service messages, you’ll get a clear picture of how a chatbot can best support your organisation. Once you’ve done the analysis, you’ll know exactly which questions are suitable for automation and how many questions a chatbot can handle on its own. A chatbot feasibility analysis can help with this!

#8 Pay close attention to your analysis

Think carefully in advance about which questions you want to include in the chatbot and pay close attention to the analysis, so you’ll have a strong basis from which you can continue to build. A choice that you make in the beginning has an effect in the long term. Of course, you can always make adjustments during the process, but that’s not always efficient.

#9 Be aware that a chatbot is not always the solution

A chatbot is not always suitable for every organisation, or for every kind of question. Routine and predictable tasks can be easily automated, but for complex and ambiguous tasks, the interaction between people is still important. In order to truly understand customers and be able to respond to their individual needs, we still need human empathy, creativity and emotion.

#10 Decide whether to develop the chatbot yourself or have it developed for you

Do you have the capacity to develop your chatbot yourself? Or do you prefer to outsource the development of your chatbot? Before you get started, determine what works best for your organisation. This of course will also depend on your budget and your team composition. There are basically two options: You can outsource the development to a specialised chatbot developer or build your own chatbot using a chatbot platform.

#11 Immerse yourself in chatbots: get inspired!

A lot has been written about chatbots, especially lately. Improve your knowledge by reading white papers or success stories from others, or visit an event where chatbots and artificial intelligence are central. Also, talk to other professionals who have developed a chatbot and ask them about their own experiences.

#12 Make a development plan

Once you’ve set your goal, it’s important to make a good development plan. In this plan, make sure you include exactly what you want your chatbot to achieve, in what steps and within how much time? Which stakeholders within your organisation will you need for this? Good planning is half the battle! Use a planning tool, such as Trello.

2. A flawless start for your chatbot

#13 A team with the right skills ensures the best results

We’re seeing more and more job titles that are related to the development of chatbots. It starts with a good foundation of skills, which, when combined, ensure the best results. Call in help from these kinds of people: Business Analyst, Product Owner, Conversational Copywriter and, optionally, (with the more technical platforms) a Developer.

#14 Service employees are most likely the best chatbot builders

Service employees know the customers through and through. They know how to deal with customers, what the recurring problems are and they are good at communicating. With an easy-to-use platform you can enable your service staff to build up chatbots themselves. Also, at a later stage, when the chatbot is live and needs to be monitored, the role of service agents is crucial in improving performance.

#15 Do some ‘missionary work’ within your organisation

Sometimes there’s a lack of clarity among colleagues regarding what a chatbot can do, or there are unrealistic expectations. So, do some “missionary work” to make the bot more familiar to your colleagues. You can do this by talking, explaining and showing examples. They may seem like steps that you could easily skip, but they’re very important in the process of making everyone familiar with the technology.

#16 Involve your employees from the start

Involve employees who work with the chatbot daily from the start. They need to know exactly how the chatbot works, what tasks they need to solve and what they should or should not do. This not only facilitates cooperation and greater acceptance of their new ‘colleague’, but it’s also essential for the proper functioning of the chatbot.

#17 Involve stakeholders and ask for feedback

Involve colleagues from all over the organisation in the development of your chatbot, especially colleagues who have little to do with the chatbot or the subject of chatbots in general, because only then can you get a complete and representative picture. Be open to feedback, and have colleagues test the chatbot. Based on the information you get back from this, you can optimise your chatbot and chatbot conversations.

3. Make your bot conversational

#18 Give your chatbot a personality

Chatbots, just like customer service employees, are a part of the ‘face’ of your organisation. Many organisations therefore give their chatbot a name and personality. Should the chatbot be male or female, what should it look like, how should it behave, what is its tone of voice? These are the questions you can ask yourself when designing the personality of your chatbot.

#19 Apply the principles of conversational human voice (CHV) in the dialogue

Make the experience with chatbots as human as possible, also in digitised customer contact. The importance of the ‘human touch’ in communication is reflected in the research by Charlotte van Hooijdonk and Christine Liebrecht in the Conversational Human Voice (CHV), the human voice in digital contact.
CHV contributes to a better reputation, higher satisfaction, more involvement and a positive word-of- mouth. You can apply CHV in chatbot dialogue by integrating the following three elements into conversations: personalisation, informal language and inviting rhetoric.

#20 Write your dialogue as naturally as possible

Think of chatbot dialogue as if it were a real conversation between two people. Write dialogue as if you were talking to someone you know and give it a personal touch. If it fits with the context of the conversation, you can also add cool or surprising elements or characteristics of your organisation or brand.

#21 Apply human communication rules to chatbot dialogue

To ensure that people and bots understand each other, we need to look at our own unwritten communication rules that have evolved over thousands of years. For example, one rule concerns the amount of what we say: don’t omit important things in a conversation, but don’t add irrelevant things. When it comes to chatbots, we have the same expectations that we have when talking to people.

#22 Take the customer by the hand

During a conversation, it’s important that the chatbot guides the customer through the entire process and invites him or her to continue the conversation until he has been helped completely. To keep your customers engaged in the conversation, it’s best to end the sentences of your chatbot with a question.

#23 Write ‘one-breath’ text

In order to make your chatbot truly conversational, you’ll need to take into account that the answers are not too long. You can do the ‘one-breath test’ for this. If you read the answer out loud, does it fit into one breath? If so, then it’s perfect! Do you need to breathe more often? Then divide the text into smaller pieces.

#24 Adjust conversations per user group

You can treat different user groups differently and adjust your answers, depending on who the chatbot is talking to. Consider, for example, returning versus new visitors, or current customers versus potential customers. Also use a tone of voice that suits your target group.

#25 Determine what languages the conversations should be available in

Is your target group international? Keep in mind that the dialogue should be available in multiple languages. You can then translate your conversations into multiple languages, and your chatbot can also learn these languages.

#26 A/B test your dialogue

You may sometimes doubt what works best with your chatbot. For example, imagine a customer asking a very long question that the chatbot cannot understand. Is it better to transfer the customer directly to a live service agent or should the chatbot kindly ask the customer to shorten his question? To find out what works better for your customers, you can test both scenarios by means of an A/B test.

#27 Define business rules

Another important point is establishing business rules, where you tell the chatbot where it can get information from, when it can use external systems and when it has to ask the customer for additional information. In other words: what can the bot do, and what not do? Bots are getting smarter nowadays and can be self-learning, yet it’s still important to set business rules, so you keep control of what the chatbot does.

#28 Don’t limit yourself to flowcharts, make conversations that are more comprehensive

Don’t limit yourself to flowcharts, which take the ‘human’ out of the conversation. Create realistic and comprehensive conversations that are flexible and match how customers engage with your organisation.

4. Testing and linking to external systems and tooling

#29 Take the time to test

Keep testing your chatbot, at all times. Is it consistently doing well? If not, what’s going wrong? Can you give it extra examples to sharpen and improve the service? By doing a test yourself, you get a better sense of what the chatbot means for your organisation and customers. Take the time for this step, because your chatbot’s success has a direct impact on customer satisfaction.

#30 Enrich your chatbot by linking it to external systems

A link with an external system provides a 360-degree view of the customer and can help the chatbot to perform certain actions. This link is not necessary, but it does enrich the chatbot! For example, you can link your bot to your CRM system, where there’s more customer information. You may have to address certain customers differently than other customers. This distinction can then be made.

#31 Have your chatbots and employees work in the same environment

For best results, have your chatbot work in the same online customer service (webcare) environment as your customer service employees. This way, your service employees will know exactly what the chatbot is doing and can intervene at any time. In addition, you can easily use your chatbot on new channels.

#32 Always provide an opportunity for human contact

To avoid customer frustration, make sure there’s always an employee available to pick up a question if the chatbot can’t quite answer it.

#33 Distinguish between virtual assistants and employees

Make sure your customers can distinguish between talking to a chatbot or to a human customer service representative. By making a clear distinction between the two, you’ll prevent any confusion for the customer. The transition from chatbot to a live conversation with an employee must also be clear, so the customer always knows what he or she can expect.

5. The go-live and optimisation of your chatbot

#34 Choose a moment to go live

Has your chatbot been tested extensively and is everything working properly? Then it’s time for the launch! Do you opt for a soft launch or are you going all the way right away? Choose the best moment of the launch that suits your organisation. This is of course branch or season dependent.

#35 Spread the word!

In addition to choosing the moment, communication, and possibly a campaign, are of course also important. Search the press, use social media channels and newsletters to get the message out. Your marketing colleagues would be happy to help you with this! Also, don’t forget the internal communication with colleagues. They would also be happy to contribute by announcing the launch within their own network.

#36 Make your chatbot even smarter

The chatbot has to learn to understand what you mean. Where bots used to work primarily through a script, nowadays we have Artificial Intelligence and technologies to make bots smarter. This means the current bots have the potential to find out the purpose of the message (intention), but also to filter data from the message (entities), such as an e-mail address or customer number. A bot can therefore be trained to understand messages.

#37 Categorise questions

Most organisations that bring in a chatbot are already in a service process, where online messages are handled in a tool. By categorising questions and extracting examples from your own process, you can feed the chatbot with the different messages and different ways in which the customer asks a question.

#38 Your chatbot can make mistakes

Like every new employee, a chatbot can make mistakes in the beginning. If you introduce your chatbot step by step during a ‘soft launch’, you can remove potential functional errors without exposing too many users. By manually assessing whether the chatbot has done something right or wrong or if the chatbot didn’t take action where it was actually needed, the chatbot can learn how to estimate when it can be part of a conversation better.

#39 Don’t expect miracles

Give your chatbot time to learn and prove itself. The bot will learn new things every day. Thanks to the development plan that you have created for your chatbot, you can define milestones. Have the goals not been achieved (directly) or is the bot still not working optimally? Be patient! You can compare a bot with a new employee and not expect miracles.

#40 Interim evaluation

PSchedule a weekly or bi-weekly evaluation time to measure the work and performance of your chatbot. Also include the goals that you have set. Are the goals still realistic? Perhaps the chatbot can handle even more than you expected. By doing this evaluation, you get a better sense of what the chatbot means for your organisation and customers.

6. Evaluation: what did your chatbot achieve?

#41 Measure the performance of your chatbot

The number of conversations that your bot has had is an important indicator for measuring the effectiveness of your chatbot. Also check which part of the conversations has been successfully completed by the chatbot and which part has been transferred to a human employee.To determine how good the chatbot is in recognising the intention of the customer, you can look at the number of correctly submitted requests and statistics about Natural Language Processing (NLP).

#42 Check customer satisfaction (NPS)

In addition to indicators that measure the performance of your chatbot, it is important to investigate the effect of your chatbot on customers and employees. Has the NPS score changed positively? Have the waiting times become shorter? Do customer service representatives feel more motivated because their work has become more interesting and less repetitive?

#43 Have the goals been achieved?

Go back to the beginning and see which goals or KPIs you have set in advance. Did you want to save costs with the bot, a higher NPS, or growth in productivity? Evaluate whether the goals have been achieved. Hopefully you also did some measuring in the meantime and were able to make adjustments. Be critical of the goals set and achieved, and learn from them if the results are disappointing. What can you do better next time, is there anything you could have done differently?

#44 Expanding your chatbot

After a few successful months, start exploring the new chatbot to see if you can optimise it even further and maybe even include more AI in the chatbot. If your chatbot now works with buttons, you might be able to expand it with text fields. Or use a bot at the back and front of the service centre. A chatbot can be used at the back to label questions, for example. At the front, it can add value by answering simple and recurring questions.

#45 Add new chatbots to your team

An important factor for long-term success is the scalability of a chatbot. Maybe you want to start with a simple FAQ chatbot? Or maybe you want to gradually expand its tasks or perhaps have several bots working side by side in the future? You can view the optimisation of your chatbot as an ongoing project that requires a dedicated team within your organisation.

#46 Happy? Share your success with others!

Congratulations! You’ve successfully launched a chatbot after an exciting period full of new learning experiences. Now it’s time to share your experience with others and help them get the best out of their chatbot. Consider whether there are events in your area where you might be able to give a presentation or a workshop. Maybe you’ll be the next keynote speaker at a leading event!

Build your own chatbot

We hope that with these handy tips you’ve gained enough knowledge and inspiration to get started with your own chatbot! Want to start immediately with automating the service processes within your organisation? Register now for the OBI Bots Platform and build your own chatbot, even without any programming knowledge. The platform is free for the first 14 days, so you have time to get to know your way around it.

Inspired? Share this message!
OBI4wan
At OBI4wan, we have a genuine passion for customer contact, media monitoring and reputation management. We continuously feed ourselves with new knowledge about groundbreaking technologies in these industries. And of course, we love to share this knowledge! Dive into our whitepapers, ebooks, blogs, industry reports and success stories. Are you already hanging on our every word? Hopefully we can share our enthusiasm with you and start a conversation!

Related items

• 14 minute read

From data-driven chatbots to future proof customer contact

What do you need to consider when developing a chatbot and what trends and developments can we expect in the future? Rea...
Read more Arrow right
• 10 minute read

Chatbots: a game changer for customer contact

The combination of machine learning and human insight ensures better service and increased customer satisfaction. Develo...
Read more Arrow right
• 7 minute read

How to develop a chatbot in 5 steps

Chatbots ensure greater efficiency, cost reduction & increased satisfaction among customers and employees. But how do yo...
Read more Arrow right