How to use AI chatbots to enhance your content, not replace it
Our evolving approach to designing conversational interfaces that will actually deliver user value
Lucy Pickering is a user experience and content strategist at William Joseph
Conversational interfaces use natural text or voice to let people interact with digital services. This marks a major UX shift, moving away from rigid, organisation-led menus toward intuitive dialogues that adapt to a user’s context and intent.
In other words, navigation shifts away from users exploring information in the way that organisations have structured and grouped it, to instead being free to access or engage in the way most relevant for them.
The two most common conversational interfaces are:
- AI chatbots: text-based tools, ranging from simple rule-based support to generative AI collaborators
- voice user interfaces: speech-driven tools like smart speakers and voice assistants (such as Siri or Alexa)
In this post, I’ll focus on AI chatbots and explore the key principles we apply when incorporating them into digital content and design.
Communicate real value and benefits
Offer a clear value proposition
Any conversational feature should have a clear promise to the user of the value they can expect to get from using the feature. This should be understood before any prototyping starts.
For example, if you’re designing a chatbot for your website, clearly define why a user would interact with this rather than using cross-site search or their preferred LLM tool, which would draw on a much broader dataset.
Lead with these user benefits
Users are keen to know what the benefits are of interacting with a chat feature, rather than the technology it uses. In fact, an obviously AI chatbot experience or labelling something as ‘generated by AI’ is likely to degrade trust.
Highlight to users what a chat feature helps them do and why they should engage with it. For example, does it help them compare support options available to them based on their individual circumstances?
Don’t use chat as a sticking plaster
Solve real user problems
If the primary problem is that users can’t find the information they’re looking for, this is likely to be due to an organisational challenge, rather than something technology can solve.
If this is the case, it would be more efficient to invest time into building understanding of users or alignment between teams, rather than into a new product or feature. Focus on people and process before tools.
Connect to existing user journeys
Any conversational interface should feel connected to the user journey rather than an additional ‘layer’ over the top. It should be informed by data from users’ previous interactions.
And crucially, a chat feature can’t fix disorganised content
While a conversational interface may help to surface answers to specific questions, it will not fix thin, inaccurate or poorly structured content. Like users, AI tools require websites and pages to be easy to crawl to find the most relevant information.
This type of feature should only be considered if you’re confident that users can already reliably find your organisation’s content from:
- a traditional search engine
- your homepage
- your navigation
For example, at Citizens Advice, the human-centred design team have been exploring what AI tools mean for their advice content.
In the future my team could be designing conversations instead of content, and personalised journeys instead of website navigation.
However, Katie also highlights that:
The next generation of digital services won’t just be built by emerging technology, they’ll be built on better stewardship of knowledge.
And so the team has recently completed a content audit that covered more than 1,600 pages and 3 million words. They asked questions like:
- is this still accurate and accessible?
- who actually needs it?
- do we still have the expertise to maintain it?
- could someone else do this better than us?
The outcomes of this audit are sure to be much more impactful for Citizens Advice’s users than investing in an AI chatbot feature.
Keep people in control and build trust in a ‘collaborator’
Keep people in control
Revisiting the point about trust, conversational agents should be presented as a supplementary resource or tool, rather than implying AI is sentient or works in a vacuum. For example, say: ‘Researched with AI’ not ‘Generated by AI’.
Build user trust through transparency
Be specific about technology’s contributions and communicate what is happening during the process. Use microcopy that explains stages, such as ‘refining options’.
Don’t exclude anyone
Consider how accessible a conversational interface might be. Will it support speech-to-text inputs and audio outputs, or will it be limited to text?
The team at Content Design London identified that chatbots can cause problems for people who use:
- high zoom, and may not know they are in a chat window or may lose other functions they need, like ‘find’
- screen readers, which may be interrupted when new content arrives in the chat window
Give users the choice to use the feature
An interface (of any kind) shouldn’t automatically pop-up and obscure content or interrupt users.
Offer routes for errors and escalations
If a chatbot conversation doesn’t work for someone, ensure they have the opportunity to report errors and the agency to choose an alternative interaction, like moving from chatting with a bot to a human.
Get in touch
These are the content and design principles that we’re applying at William Joseph as we navigate these technologies with our clients. The principles will continue to evolve as we design, test and build and discover other examples.
If conversational interfaces or AI chatbots are something you are also exploring, we’d love to chat about it. You can find me on LinkedIn or at info@williamjoseph.co.uk