October 3, 2018
Structured Content: The Key to Successful Chatbots and AI

Chatbots are a popular topic right now, and for a good reason.

At an Intelligent Content Conference in 2017, someone pointed out that chatbots might be just the thing needed to break the “copy-paste” habit.

Many companies are evolving their digital experiences by incorporating some type of chatbot – for customer support, for website support and more. But as important as chatbots are becoming to the digital experience, if they aren’t implemented right, all they will end up causing is frustration. 

Another Content Silo 

Here’s the biggest concern about chatbots – they are quickly becoming another silo of content in the organization. Organizations want them for different reasons. A chatbot can help a customer or prospect get an answer to a question quickly, any time of the day. It can gather basic information for a support call before connecting in a live agent. It can also provide personalized information if it leverages AI and machine learning.

But to build the best chatbot, you need to consider its implementation as part of your wider digital experience strategy – including your content strategy and content model.

Leveraging Intelligent Content for Chatbots

Not all chatbots use AI, some basic chatbots work from a set of content that follows a decision-tree, but to deliver personalized experiences with a chatbot, you will need one that leverages AI and machine learning. In both instances, structured content is critical for the chatbot to work properly.

Ann Rockley said it in her webinar with us.

“For chatbots to work well, you need structured content. There are three main elements to chatbot interactions: context, intent, entity.

  • Context - the reason for the interaction
  • Intent - the purpose or goal of the interaction
  • Entity - delivering the content that is required as a result of the interaction (this is where structured content comes into play)

Any entity can be a granular component of information, or if it's a larger piece of content, it can be semantically tagged. The chatbot can then interface with it and pull out a smaller piece of information it requires.”

But Rockley isn’t the only person espousing the importance of structured (intelligent) content for a successful chatbot. Cruce Saunders, founder, and principal content engineer at [A] offers the resource guide, “Engineering Content for Bots, AI and Marketing Automation.” In the guide, he said,

Bots and intelligent assistants, such as Google, Siri, and Alexa actively seek to understand our content so they can bring answers and ideas to our customers. We can gain market advantages by engineering content to be easily understood, and used and amplified by the AIs, external and internal cognitive platforms, BI, and marketing automation.”

He also said that we shouldn’t be thinking about where the content comes from for a chatbot. Instead, we should be figuring out how to “prepare, organize and structure that content so the chatbots can use it.”

Chatbot Content and Your CMS

You could store your chatbot content directly in your chatbot software, but it makes more sense to manage it within your CMS, as part of your existing content. To do that, you need a CMS that provides an open API the chatbot can connect with to pull content. It also requires your CMS to support a structured content model.

Saunders talks about the process of building a content model that includes the elements required for AI and chatbots. These elements include entities, intents, and responses.

chat-bots-ai-diagram

(Image from Engineering Content for Bots, AI, and Marketing Automation)

When you build your content model in this way, you can leverage all the content in your CMS in a chatbot.

To build this out further, content that a chatbot may need may be located in a number of different repositories. But a chatbot cannot connect to each repository; it requires a central location with a defined (structured) content set to work properly. A CMS that can import or integrate content from multiple content repositories, like a component content management system used for help documentation, or a series of Word documents used for training, can build out the content model for that additional content and use it to support chatbots.

The Bottom Line: Chatbots Need the Right Content and Content Structure

You could copy and paste the content you want for your chatbot into the chatbot application. But consider the nightmare of keeping it up to date and synchronized with similar content on other digital properties. Sounds painful.

By defining a complete structured (intelligent) content model, used across the organization, that incorporates the elements a chatbot requires, you make it easier to feed the best content to your chatbot, and in turn, to your customers and website visitors.

Posted by David Hillis
Subscribe to Blog

Subscribe to the Ingeniux blog for the latest in web experience management. 

Subscribe to Blog