What it takes to be a UX Designer today: A conversation with David Grunewald, AI strategist
Digital Marketing
Why your next chatbot needs a UX brain, not just AI
Back to overview
-
Digital Marketing
For two years at Ariad, after experiences in SMEs, start-ups, marketing teams, and even a year in Canada, UX Writer & Content Designer Uyen Nguyen has consistently embodied one idea: every good digital product begins with a well‑designed conversation.
In our latest training, she pulled back the curtain on what actually makes a chatbot work. And spoiler: it’s not the tech. It’s the thinking.
And to understand why, she brought us back to the basics...
Starting with what most people think a chatbot is. A little window at the bottom of a page. A quick reply. A way to avoid a phone call.
In reality, most companies quickly discover that what looked like a “small project” is actually a strategic engine: complex and sometimes risky. Behind that tiny interface lies a network of decisions that can touch every part of the business: customer experience, brand perception or even revenue generation.
Discover Uyen’s insights below...
Understanding humans before building robots
One of the most common mistakes companies make is trying to build a chatbot that can “do everything.” The expectation is often that the bot should be able to answer any question, on any topic, phrased in any possible way.
In practice, this is unrealistic. Instead...
..start narrow: focus on one clear goal
Instead of building a generalist bot, start with a highly specific and goal-oriented use case, such as:
- Driving sales (product recommendations, checkout assistance)
- Providing structured information (opening hours, pricing, availability)
- Handling a well-defined customer support task (e.g., managing rebookings for delayed or cancelled flights in an airline context)
Once that initial scope performs well and delivers measurable value, you can gradually expand it.
This process is the only way to create a chatbot that truly reduces costs, improves user experience and builds trust.
Where many companies simply add a technical component, she reminded us that every word the bot uses carries a piece of your brand. It’s your brand's ambassador.
Communicate what your bot can (and cannot) do
A similar mistake is failing to clearly communicate the capabilities of your chatbot.
Opening with a vague prompt like:
“How can I help you?”
implicitly suggests that the bot can handle anything. Users will naturally ask broad or unrelated questions. When the bot fails to respond appropriately, the interaction quickly devolves into:
- “I didn’t understand your question.”
- Irrelevant or off-topic answers
- Frustrated users
The scope must be limited and clearly communicated from the very first message.
For example:
“I can help you modify your booking, request a refund, or check your flight status.”
You can also try to use action buttons or predefined options instead of free-text input whenever possible. Structured choices guide users toward supported flows and dramatically reduce ambiguity.
Guide users when errors arise
When a chatbot fails to understand a request, repeating “I didn’t understand your question” is one of the fastest ways to frustrate users. Many projects overlook this and only address errors superficially.
Instead, design error management intentionally:
- Redirect users toward supported topics
- Offer clear, structured options (buttons, suggestions)
- Provide a quick and smooth handoff to a human when needed
What to takeaway for your business?
“Before you make a chatbot speak, give it a reason to exist.”
Don’t create a chatbot just to say you have one.
Build it because it solves something real.
Build it with the right people around the table.
That’s where Ariad comes in. We connect digital and ICT teams with experts who make digital experiences feel human.
If you want your chatbot to actually help your users, reduce costs and elevate your brand, start with the right minds.
We know a few. 😉
