By Charles Lee Mathews.The empty chat box is a design failure.
Conversational AI has created a new design question for brands: if the customer simply wants an answer or an action, why make them have a conversation to get it?
The problem with the chat box
“Stop treating the open-ended text box as the pinnacle of digital experience design,” says Rafiq Phillips, chief technology officer at Naritive. “More often than not, an empty chat bubble is just a lazy UI cop-out used when a team doesn’t want to do the hard work of designing an intuitive, structured user journey,”
Naritive’s analysis of close to a billion interactions across the features it has built shows feature bloat kills intent — and the same rule applies to conversational agents. “If a single tap delivers resolution, forcing an open-ended conversation is just bad engineering,” Phillips says.
Niel Mouton, managing director of WPP Enterprise Solutions SA, wants the scoreboard changed. “Stop designing for engagement. Design for outcomes and let the customer decide whether they want a conversation at all. In service, success is someone leaving quickly with their problem solved or their question answered. Measure time to resolution, not containment,” he says. On escape routes: “It should never be the mandatory first step, and the route to a person should be visible at first contact, not unlocked after three failed loops,” says Mouton.
Pivot from conversation to execution
In the mid-2000s, Phillips built the first external MXit chatbot to pull location data into chat when mobile browsers were clunky and data was expensive. He argues the industry has inverted a formula that already worked. “Twenty years ago, we used text interfaces to eliminate friction; today, brands are deploying conversational AI to introduce it,” he says.
His prescription is agentic rather than conversational. “Conversational bots just talk, and sometimes they hallucinate polite apologies about why they can’t access your profile or complete your request. Agentic systems actually do the work when they connect directly into backend APIs, inventory systems, loyalty engines and payment gateways,” says Phillips.
Mouton’s strategic point follows directly from the Gartner split. “Plan for customers bringing their own assistants. If people are three times more likely to ask a general assistant than yours, the real question is whether your product data, policies, and service actions are accurate and reachable by the assistant they already trust,” he says.
Charlie Stewart, chief executive of Rogerwilco, frames it simplest. “Rather than being ‘bot-first’ or ‘human-first’, the best way forward is ‘outcome-first’.”
Paula Hulley, managing director of Digitas SA and lead of Publicis Commerce SA, says brands assume customers want to chat far more than they do, and miss how much those customers value getting things done fast, without complication and on their own terms. “In many situations, customers do not want a conversation. They want an answer,” she says. “The future of customer experience is not about having more conversations. It is about ensuring every interaction earns the right to exist,” she says.
Customers want tools that work, not synthetic intimacy. Until brands accept that, their chatbots will continue doing the one thing they were never intended to do: talk to themselves.
Charles Lee Mathews is a senior editor to MarkLives MEDIA and a senior writer to MarkLives.com, as well as co-founder of The Writers, a writing consultancy.
👉 Read more: Customers are turning to AI — just not your chatbot (October 2026)
