Chatbots - Design for Success - Insight into working life at Stream
Stream’s unique software solutions, combined with a clear understanding of the significance of data, enables us to work with clients to help build brand loyalty, increase sales, drive customer satisfaction and drive employee engagement. Stream helps build trust that delivers brand loyalty and commercial rewards and, through this, can prove that the cost of inertia far outweighs the price of change.
Stream have just won a contract to develop an online coaching and professional development tool for a major healthcare company. Watch this space for more information.

We have recently completed 5 incentives for Anglian Home Improvements taking guests to London, Loutraki, Dubrovnik, Dubai and Rio de Janeiro.
Chatbots - Design for Success
17 August 2017

Chatbots - Design for Success

Chatbots are increasingly becoming a “must have” for almost any size of business. Chatbots can sit on your own website as well as being deployed across multiple social channels, and messaging apps, to provide a consistent customer experience.

Leading busy lives, customers expect to be able to interact with brands at a time to suit them, whether that’s at 4pm or 4am.

Creating consistency in messaging, as well as enabling customers to find information easily and quickly, helps create valuable contacts and leads for your customer service team to follow up.

A chatbot can be deployed to interact with customers online and through voice-activated and messaging apps and devices. Chatbots can have very human qualities: they are immediately available, they can engage in natural conversation and they require little or no effort. Messaging apps rule the world, with 4bn daily users, so now is the right time to get onboard. Voice and conversation is the new UI (User Interface) and chatbots enable you to take full advantage of the potential value.

Chatbots are simple and task orientated. A chatbot could be developed initially to engage in conversations that start with a specific question. It could then later be developed to be able to handle more complex conversations. As more data is stored and analysed, the chatbot continually learns to ensure that the flow is consistently delivering against customer expectations.


With all chatbots, we would recommend starting small with a specific user case. If you are already using live chat, then you should be able to analyse your chat logs to see which questions appear most often and which could easily be handled by a chatbot rather than a member of your customer support team. This could be anything from booking an appointment to requesting information on a product or range of products to finding information on local events.

Once you have defined your scenario, you can then use your existing chat-logs to understand the questions customers are currently asking and the steps they go through to get the required information. If you don’t have chat logs to analyse, then why not role-play the scenario by using members of your own team or even your customers? This will help with the next three stages of designing your bot.


The personality of your chatbot must reflect your brand, as well as engage with the customer, and this is one of the most important aspects if the chatbot is to succeed. If you are using an external agency to develop your bot, make sure they have a really clear vision of your goals, brand, values, external messaging and your tone of voice.

To create the chatbot’s personality, you effectively need to create one that has all the characteristics which you would look for in the perfect customer service agent. The personality must also be linked to its purpose. As you have previously defined the purpose of the chatbot, how it will add value and its characteristics, you are now able to determine how the chatbot would respond to specific conversations with customers.

If you are already using live chat, you have probably already had customers ask if they are speaking to a bot. Whilst this isn’t great when it’s a human, it is worth saying that it is OK to make customers aware that they are dealing with a bot. You don’t necessarily need to pretend your bot is human; you just need to give the customer the ability to be transferred to a customer service agent whenever they want to. If your chatbot is engaging, and providing customers with value, then they will happily interact.


It is key that you define the “must have” functionality for the chatbot. It is advisable to create a list of requirements segmented into how desirable they are from “essential” to “desirable”. Always refer this list back to the purpose of the bot. Don’t try to over develop the list of requirements initially. Once you have built the framework and tested its success against your initial purpose and scope, then you can add and adapt functionality for each new purpose.


This is the most complex aspect of designing the chatbot and requires all the other steps to be researched, defined and confirmed before starting. The flow allows you to think about how the chatbot will perform in every scenario and what value they will provide the customer. Using buttons to direct the user to specific information based on your analysis of conversations is advisable.

Good design and UI is essential in making the chatbot engaging and interactive. Here are some practical suggestions:

  • Don’t allow conversations to lead to a dead end; ask questions that prompt the customer and make them aware of what they can ask
  • Don’t make messages too long; think about how much you would write to a friend. Customers don’t have time for long winded messages
  • Make sure you are using the data provided; throughout the interaction with the customer, if they provide information that the chatbot doesn’t need at that point, store it for use later on

Using Artificial Intelligence and machine learning tools the chatbot can be re-trained to ensure that the flows are kept up to date. The more data you have available to analyse, the better the chatbot can interact with customers. Re-training is a vital element of good chatbot design. Think of your chatbot as a new customer service agent that is able to talk to multiple customers at once. That person still needs development and training to deliver the best experience to the customer.

One last design tip: You will get many customers asking the question “are you a robot?” Don’t try and pretend the chatbot is a human support agent. Why not come up with a fun response to this question that engages the user – for example: “Yes, but I’m learning to be more human.”

Chatbots are a relatively new tool in the marketer’s toolkit. We at Stream love to kick ideas around on this and other new marketing opportunities and we’d love to talk it through with you over a coffee.