The development of chatbot technology is at a critical juncture, with AI advancements propelling these services to new heights. The Chatbot Rank methodology revealed a modest 9-15% increase in successfully processed chatbot requests for each block of user scenarios over the 2022-23 year; but for the period from 2023-24 a more significant increase in successfully processed chatbot requests is indicative - in the range of 20-30%. Despite these gains, chatbots still carry the reputation of being tools for simple tasks. When faced with complex queries, users often bypass the chatbot and directly request a human operator, even though the bot is capable of handling the task.
This perception is partly due to product teams focusing on their development priorities rather than on clear metrics for customer experience in text channels. As a result, advanced chatbots that can accurately recognize natural language might fail to maintain effective communication, diminishing the team's efforts and reducing the ROI from chatbot integration in digital services.
Markswebb aimed to address this issue by developing the principles of an effective conversational user interface (CUI). This comprehensive set of rules is designed to manage communication quality with chatbots. In this article, we will discuss these principles and demonstrate how they can enhance service quality.
Contents
Markswebb's methodology for evaluating chatbots focuses on two major factors affecting user experience:
While the functionality is straightforward - either a feature exists or it doesn't - the interaction quality is assessed using the principles of conversational user interface (CUI), developed in Markswebb in 2022.
These CUI principles, encompassing over thirty qualities, ensure that chatbots communicate effectively and human-like. The principles are used to create a checklist that measures the "humanity" of a chatbot, ensuring it saves the user's time and communicates politely. This structured approach allows us to objectively compare chatbots, highlighting both strengths and areas for improvement.
The effectiveness of this principle is measured by the success rate - whether the user gets their issue resolved. The bot must recognize if it can answer the client's question and switch to a human operator if necessary. This principle includes five rules, such as ensuring the bot does not give the same answer twice in a single dialogue.
This principle encompasses a range of rules concerning the bot's language and feedback. Adhering to these rules makes the bot's reactions and behavior akin to those of a live consultant. One rule is to proactively inform the client about the status of their request or its completion.
Clients often get frustrated when they need to rewrite requests to fit the bot's "mechanical" language. Bots that adapt to the client's language make the interaction more comfortable. This principle is outlined in five rules, such as the bot recognizing queries with typos and allowing users to return to any stage of the task.
This principle involves using accessible language without unnecessary terms or jargon, making responses as straightforward as possible. It includes four rules, such as keeping welcome messages, operator handover notifications, and clarifying questions concise.
Useful responses should not contain superfluous information, and users should not need to refer to other sources. The principle consists of three rules, such as the bot answering each question separately if the client's request contains multiple queries.
Some services hide the fact that a bot is responding to save client service resources. Users may feel frustrated if they realize later that they were talking to a bot. This principle includes two rules, such as the bot clearly indicating when a human operator joins the conversation.
This principle pertains to the context in which the bot operates, focusing on UI settings such as font customization, message storage, and the ability to share chat elements. The principle is detailed in three rules, such as making phone numbers clickable and allowing attachments in the chat.
This principle addresses the seamlessness of the client's experience when switching between communication channels. It includes two rules, such as ensuring that users can confirm actions without leaving the chat.
Acknowledging the client's emotional state is crucial for building a connection. The bot should recognize negative sentiments and show understanding, which is vital in critical situations. This principle includes one rule: the bot should not ignore negative statements but rather demonstrate that it understands and is ready to take appropriate action.
We at Markswebb developed the CUI principles based on Paul Grice's cooperative dialogue rules and Geoffrey Leech's politeness principles. These guidelines create a communicative code that views chatbot behavior as deliberate and thoughtful speech acts.
Seven of the nine principles focus directly on the chatbot's dialogue, while the remaining two pertain to the context and conditions in which the bot operates, such as the chat's overall functionality and its role in the digital service.
The principles have been validated through user surveys, resulting in a comprehensive list of 47 rules that describe the capabilities or functions needed for a chatbot to meet the CUI principles.
The rules describe features or functions that help a chatbot to comply with the CUI principle - in other words, to communicate as comfortably as possible from the human point of view.
Explore Markswebb’s 9 CUI principles - a proven framework for designing human-like interactions and enhancing user trust. Learn from real-world implementations and start transforming your chatbot experience today!
Principle 1: The bot solves the user's problem
A key measure of effectiveness is the success rate—whether the user gets their issue resolved. For example, if a bot provides the same answer multiple times, users might feel ignored. A well-designed bot, however, acknowledges previous responses and encourages users to rephrase their queries if necessary.
Principle 2: The bot communicates politely
Polite communication involves a range of behaviors, from using appropriate language to proactively updating users on their request status. For instance, confirming the completion of a task helps users feel in control and reassured.
Principle 3: The bot adapts to the request
Bots should understand user inputs, even if they contain typos or informal language. Additionally, users should be able to navigate back to any step in the process easily. For example, a bot that can handle misspelled words demonstrates a higher level of understanding and keeps the user engaged.
Principle 4: The bot provides simple answers
Bots should avoid jargon and unnecessary details, offering clear and concise responses. Visual aids, like screenshots, can be useful for guiding users through complex tasks. For example, a concise welcome message and direct questions can help users quickly move forward with their inquiries.
Principle 5: The bot provides the optimal amount of information
Bots should deliver just enough information to resolve the user's issue without overloading them. For instance, if a user asks multiple questions, the bot should address each one individually, ensuring comprehensive support.
Principle 6: The bot informs users they are interacting with a machine
Transparency about the bot's nature is crucial for managing user expectations. Bots should clearly indicate when a human operator takes over the conversation, reducing the risk of user frustration.
Principle 7: The chat has a functional interface
A user-friendly chat interface includes features like clickable phone numbers and the ability to send attachments. Such functionalities align the chatbot experience with familiar messaging platforms, enhancing user comfort.
Principle 8: The bot integrates with other channels
Seamless integration across communication channels ensures that users can complete tasks within the chat environment without switching platforms. For instance, confirming a service activation directly within the chat window streamlines the user experience.
Principle 9: The bot responds to the user's emotional state
Recognizing and responding to user emotions, especially negative ones, builds a stronger connection. A bot that can address frustration or urgent issues effectively demonstrates empathy and competence.
Implementing these principles does not necessarily require advanced technology or significant investment in AI. Focusing on improving the digital service quality from a user scenario perspective can yield substantial benefits. Effective chatbots can enhance product metrics, improve customer satisfaction (CSI), and build overall service loyalty (NPS). Ultimately, a well-designed chatbot reduces the burden on customer service teams and fosters trust and reliability in the digital service.
We respond to all messages as soon as possible.
We’ve evolved dozens of successful financial services and are eager to prove that our expertise can be implemented in other industries and around the world. Have a look at our success stories!
