There are good chatbot experiences, and there are bad chatbot experiences.
Bad chatbot experiences serve up irrelevant content, ignore (or are incapable of parsing) context, and leave customers frustrated. Good chatbot experiences, on the other hand, help customers succeed while leading to gains in efficiency, cost, and customer experience.
Typically, these “good” experiences deliver relevant and timely content, often powered by sound knowledge management strategies.
Optimizing content for chatbots
How, then, do organizations create the kind of content chatbots actually like? What are some things an organization can do, from a knowledge management and content creation standpoint, to make it easier for chatbots to deliver better outcomes for the end user?
Here are a few things to consider.
1. Create content based on user needs
Creating relevant content means looking at the data. What are the top issues that customers need help with? Which tasks along the customer journey lead them to initiate customer service interactions?
Where are the obvious gaps in self-service content?
For the answers, dig into search query reporting, top case drivers, and website analytics. This data can come straight from Google Analytics, third-party tools, or the reporting function within a standalone knowledge management solution. An exercise in customer journey mapping can also reveal the key parts of the customer lifecycle that are missing corresponding support content.
For more sophisticated chatbot deployments, analytics from the conversational platform itself can provide insights into what queries customers commonly use, or what kind of issues they are commonly trying to solve.
All of this information should help guide efforts around creating and maintaining the kind of knowledge that fuel more seamless chatbot experiences.
2. Move beyond PDFs and simple FAQs
It’s broken record time. PDFs do very little (if anything) for chatbots, which means they do very little for the customers interacting with those chatbots. The traditional FAQ is only marginally better. And content silos full of disorganized information are similarly limited.
While these content types (and the silos they’re kept in) might still have their place, they need to be broken out into structurally rich microcontent if chatbots have any hope of finding them. Think of it as repurposing those long-form guides, tutorials, and manuals into discrete pieces of information “chunked” for better discoverability.
3. Get (structurally) rich quick
Optimizing content for chatbots goes beyond breaking things down into bite-sized pieces. To extend that microcontent across channels, it needs to be well structured and in the right format.
In this regard, a knowledge management system can be a tremendous asset. The MindTouch knowledge management platform, for example, employs a Guided Content Framework that helps maintain consistent site hierarchy, document structure, and search visibility.
While this framework lends to improved customer self-service experiences in a variety of channels, it also helps teams publish the structurally rich content that chatbots need to do their jobs.
4. Metadata matters
Metadata—the information about how content it is categorized, tagged, and classified—helps chatbots understand the audience and context that a given piece of content is appropriate for. It allows the chatbot to quickly find content that matches the detail it has extracted from a user’s inquiry, question, or request, and with greater accuracy.
Here are some key metadata to optimize for each piece of knowledge content:
- Categories and tags
- Classifications (article or content type, for example)
- Image titles, alt tags, and captions
(Keep in mind, this is not an exhaustive list.)
Extending content to all channels
In many ways, optimizing content for conversational interfaces means creating a “single source of truth” that can be extended not only to chatbots, but all customer support channels. It extends beyond the types of content chatbots like to the very ways that knowledge is organized and delivered.
More often than not, the better the underlying knowledge management strategy and information architecture, the more personalized and contextually sensitive the chatbot experience is overall.
Which is exactly what customers expect.