3 Ideas to Help Online Chatbots Deliver Results for Your Business

Chatbots are now finding use cases in numerous industries, from flight bookings to counseling and even online dating. Artificial intelligence has made chatbots more skillful than before, equipping them to answer questions about the business and resolve queries and grievances.

The widespread adoption of chatbots has opened new ways for businesses to expand their reach and improve efficiencies. At the same time, not using this technology with caution can prove counterproductive. A recent Forrester report finds that while 71 percent of companies have incurred expenses toward getting a chatbot, less than 20 percent of adults actually use it for help. 

Here are three guidelines that can help business owners implement chatbots more powerfully to witness concrete results.

1. More Conversational, Less Commercial

On the surface, it feels as if conversational AI is delivering wins—more people now chat with online therapists, shopping assistants, and even romantic partners. 

These entities are not human but AI-generated, part of an illusion that is making parallel versions of reality for many of us. The New Yorker puts it well: real people cannot “flatter or comfort” with the smoothness of an optimized AI chatbot. Commercial organizations have internalized this knowledge, using chatbots to convince people to try new clothing lines or book a seaside resort.

Creating a balance between conversational and commercial intent will be mandatory for businesses to witness positive results from chatbot integration. Too much of either can sway a consumer toward distraction or skepticism, both of which are suboptimal. 

Businesses can build this balance by researching the intent of users interacting with their chatbots. Accordingly, they can invest in an AI chat application whose personality matches their brand’s identity.

2. More Stringent Fact Checking

One vexing problem of AI-based chatbots is their tendency to answer every question, regardless of whether the facts are verified.  

A Columbia Journalism Review study found that many AI tools deliver misleading information over 60 percent of the time. A Nature study also had similar observations, revealing that many chatbots don’t want to admit ignorance. This false overconfidence can lead to factually incorrect answers. 

In 2024, Air Canada lost a small claims case that started from its chatbot’s mistake. The chatbot had told a passenger that they could claim bereavement fares retroactively. Air Canada claimed it did not accept such applications and tried to defend itself by insisting the chatbot had made an error.

What makes such problems more stressful is that people don’t always (or easily) spot wrong answers. The conversational tone of AI chatbots mimics a human, which can inspire confidence in the users. So, their likelihood of believing false information may also be higher. 

For example, some businesses may underplay the risk of exposure to chemicals in their online publications and posts. A survey by RepRisk, a data science company, found that 25 percent of recent corporate ESG claims were merely greenwashing. 

In reality, many business operations and products continue to pose climatic and health harm, from nonstick pans to aqueous firefighting foam (AFFF) with forever chemicals. In fact, the ongoing AFFF lawsuit holds manufacturers accountable for increasing vulnerability to health and immunity problems. 

According to TorHoerman Law, businesses must take accountability for warning users of associated product risks. Revealing the complete picture is essential, even if some stakeholders feel it will make consumers skeptical. 

Accordingly, business owners must analyze the communication patterns and content of the chatbots they use. Downplaying legitimate risks and sharing inadequate or misrepresented information can cause a reputational crisis for the company. 

3. Knowing When to Direct to Human Input

Another reason some chatbots don’t deliver results as expected is their overconfidence in solving their abilities—not roping in human inputs at crucial stages. This gap can annoy some potential customers who eventually do not convert. 

According to Harvard Business School, chatbots may offer over-simplified advice for complex questions. For example, in response to niche business queries, AI applications may respond with well-meaning but generic advice that cannot effectively solve the problem.

Businesses should endeavor to complement AI chatbots with human-driven customer service. For tech companies, this will involve working on the underlying language models to let AI assess when to withdraw and transfer to an assisting human. IBM research in this direction considers a “Human in the Loop” model, which uses incremental learning and continuous improvement.

On another note, this problem is more likely for businesses whose chatbots do not switch on and off easily. The constant pings can get frustrating when you’re in the middle of something! While working on improving your chatbot, it is best to reexamine the UX design.

AI chatbots are excellent on paper—an automated assistant that attends to leads and customers, saving time for your business. In reality, they require monitoring and upgrades to remain relevant and encourage revenue growth. It requires more effort than just the initial investment and checklist-ticking of having a chatbot on your website.