Would it surprise you to learn that people often feel more comfortable sharing information with machines than with other humans? This appears to be the case. Therefore, what people share in customer service interactions with chatbots can provide unrivaled customer truths.
These deep insights can be easily analyzed and collated using AI and turned into an even more advanced level of data-driven decision-making. Companies that leverage this phenomenon can improve customer service (again) but also accelerate outbound marketing efforts and more strongly iterate product and service development.
Trust and AI – Research Shows Humans Have Honest Conversations with Chatbots
In part one of our hidden truths, we examined what people talked to chatbots about. We touched on the popularity of the Psychologist bot on Chatacter.ai and the use of AI to triage patients in the UK. We also delved into how much people talked to entertainment-type bots and some of the dangers of oversharing information and opinions.
A study published in August 2023, titled “It’s only a computer: Virtual humans increase willingness to disclose,” found that in a health and medical setting, chatbot users were more willing to disclose information than in conversations with a human. The research discovered participants who believed they were talking to a computer “reported lower fear of self-disclosure, lower impression management, displayed their sadness more intensely, and were rated by observers as more willing to disclose.” The study’s results suggest that using “virtual humans” (VHs) in clinical interviews may help overcome a barrier to obtaining truthful patient information.
Why the Increased Openness?
The research is important because medical patients sometimes hold back information due to a fear of self-disclosure. Patients can feel they are viewed negatively by a healthcare professional. They can be afraid to reveal “personal, sensitive or stigmatizing information.” Patients can also convey “impression management,” where they reveal information that leads healthcare providers to view them positively. The crux is that a less accurate picture of medical history can have health consequences.
But why the increased openness with chatbots? Or, in this case, virtual humans? Chatbot users can gain some sense of anonymity and feel their identity is protected, so they may reveal more. There’s also the idea that people will share more information with an “impartial machine” that won’t judge. In healthcare, computer-administered assessments are shown to achieve greater disclosure (Weisband & Kiesler, 1996). The researchers say these formats give a “sense of invulnerability to criticism, an illusion of privacy, the impression that responses ‘disappear’ into the computer.”
What are the Implications of Customer Openness for Gleaning Insights from Chatbots?
Personal insights and information must only be gleaned for diagnosis in a medical setting. However, outside of healthcare and where data privacy, regulatory compliance, and data security are respected and customer permission gained, the potential to gain and leverage customer insights from chatbots is immense.
Generative AI and natural language processing (NLP) provide chatbots with almost human-like conversational abilities. This may also encourage people to share more information or be more honest about their preferences and concerns.
Once you deploy your own custom GPT bot, available no-code and off-the-shelf from CustomGPT, you can use it on your help desk or website to answer customer inquiries and increase customer satisfaction. But these bots have a much more significant benefit to your business. CustomGPT.ai CEO Alden De Rosario, in AI Chat Logs — The Hidden Goldmine Your Company Hasn’t Discovered Yet, says:
“Once deployed, your employees and prospective customers are telling you exactly what they want, what their roadblocks are, and other customer insights. This information is invaluable to literally every stakeholder in your company.”
Chat logs are the log entries of your customer’s interactions with your custom GPT chatbot and your bot’s responses. The information contained in them is of interest to every department in your organization, from customer support to product development, R&D, leadership, and legal and compliance teams.
Because of AI’s advances, people have longer conversations with ChatGPT-style bots, and many see AI as a judgment-free zone. De Rosario adds,”
“The chat logs are showing raw, unbiased, and unfiltered interactions of employees and customers with the AI.
These raw and unfiltered interactions usually bring out the truth — a perfect scenario for true unbiased feedback. Customer support managers and product managers love this!”
A prime example of chatbot log use is that of Snapchat. The social media company rolled out its MyAI, discovered 10 million people sent 10 billion messages to the bot, and then analyzed the conversations. It was able to pinpoint the products and services consumers were asking questions about to improve its advertising revenue opportunity. Snapchat says:
“We are able to use conversations with My AI to enhance a key optimization that we offer advertisers, called Snapchat Lifestyle Categories (SLCs). These categories are historically based on factors like which content a Snapchatter watches. New signals from conversations with My AI will now help us enhance the quality and depth of SLCs, ultimately boosting the relevance of the ads Snapchatters see on the app.”
Snapchat is using these insights to develop ways to use “commercially-oriented conversations” to improve the relevance of its content and, thus, its advertising experiences for users. It’s also investigating the use of sponsored links within MyAI conversations.
Custom GPT logs provide:
Accurate and comprehensive data collection
Today’s chatbot logs can contain an unparalleled level of customer data that’s entirely accurate (the customer said it) and comprehensive (you have the entire log). Unless you’ve been conducting detailed in-person surveys or recording sales agent calls, there’s every chance your business hasn’t seen this level of insight. Even you have, remember, customers appear to tell the “truth” to chatbots.
Better understanding of customer needs
Authentic and detailed customer information will provide the most in-depth understanding of your customer’s needs yet. And it’s easy to gain this understanding by analyzing chatbot logs and conducting sentiment analysis with your AI.
Image Credit: Upcoming design of Insights dashboard from CustomGPT.ai
Data-driven strategic decision making
Using your chatbot log data, turned into amalgamated insights, you can now respond by:
- Exponentially improving your customer service by re-training your chatbot
- Teaching your bot to sell or upsell by leveraging AI’s advantages for sales
- Answering customer points of frustration by sharing guides and how-to’s with your bot, on social media, or by email marketing
- Using the insights to improve your outgoing marketing strategy entirely
- Enhancing your product and letting your customers know
- Quickly launching new products or product features in response to demand
A Unique Opportunity for Customer Understanding
Generative AI and custom GPT chatbots vastly improve customer engagement by interacting with customers and giving human agents and managers more time to focus on critical customer problems and sales. But the benefits go far beyond when you start to look at the more profound insights into customer preferences and experiences that are gained from chatbot logs. This is just another way that AI is revolutionizing how companies serve their customers: by perhaps truly understanding these customers for the first time.
Frequently Asked Questions
Why do people often share more detail with chatbots than with human agents?
People often disclose more to chatbots because they feel less judged and more private. The August 2023 study “It’s only a computer: Virtual humans increase willingness to disclose” found that participants who believed they were speaking to a computer reported lower fear of self-disclosure, lower impression management, showed sadness more intensely, and were rated as more willing to disclose than people speaking with a human. For businesses, that can turn chatbot conversations into a richer source of customer pain points, objections, and intent when the data is handled responsibly.
How can you tell whether chatbot-generated customer insight is accurate enough to trust?
Chatbot insight is trustworthy when answers are grounded in approved content, supported by citations, and escalated to humans when the system lacks evidence. Whether a team uses CustomGPT.ai or another retrieval-grounded assistant, that is safer than relying on a generic chatbot that can guess. In the provided benchmark, CustomGPT.ai outperformed OpenAI in RAG accuracy, which matters because fewer hallucinations mean fewer false signals in customer insight. A practical rule is to treat unsupported questions as gaps to review, not as facts.
Can chatbot conversations improve marketing or sales decisions?
Yes. Repeated chatbot questions show what prospects actually care about, and the wording in those chats can improve proposals, FAQ copy, email messaging, and sales follow-ups. Stephanie Warlick, Business Consultant, described the workflow this way: “Check out CustomGPT.ai where you can dump all your knowledge to automate proposals, customer inquiries and the knowledge base that exists in your head so your team can execute without you.” When teams review conversation themes instead of isolated tickets, they get real objections and phrasing they can use directly in revenue work.
Can a domain-specific chatbot reveal unmet customer needs in complex fields?
Yes. In complex fields, a domain-specific chatbot can expose unmet needs by showing which specialized questions keep repeating and where users still need clarification. Barry Barresi, Social Impact Consultant, described that kind of specialized workflow as “Powered by my custom-built Theory of Change AIM GPT agent on the CustomGPT.ai platform. Rapidly Develop a Credible Theory of Change with AI-Augmented Collaboration.” The insight value comes from the pattern: repeated expert-level questions often signal demand for clearer content, better onboarding, or new services.
How quickly can a company start collecting useful customer insight from a chatbot?
Teams can start learning quickly once a chatbot is connected to approved knowledge and deployed through an embed widget, live chat, search bar, or API. A no-code builder, multi-source ingestion, and conversation analytics reduce setup friction, so the first useful signals usually appear as repeated questions, failed retrievals, and common follow-up prompts. Bill French, Technology Strategist, highlighted the user-experience side of this with: “They’ve officially cracked the sub-second barrier, a breakthrough that fundamentally changes the user experience from merely ‘interactive’ to ‘instantaneous’.” Faster responses can keep people engaged long enough to reveal more context and intent.
How should companies protect sensitive feedback when customers are more open with chatbots?
Protecting sensitive chatbot feedback starts with governance: GDPR compliance, data not used for model training, and independently audited SOC 2 Type 2 controls. Teams should separate approved knowledge sources from raw chat logs, limit access to conversation data, and route high-risk or sensitive issues to humans. Those safeguards matter because users may be more candid with a chatbot than with a person, so the value of richer insight depends on handling that data responsibly.