What Small Businesses Can Learn from the Analysis of 1 million AI Conversation?

According to Statcounter, ChatGPT currently dominates the AI chatbot market, which are platforms that allow users to engage with AI through natural, human-like conversations.

Released in September 2025, the study “How People Use ChatGPT” analyzes over 1.1 million conversations to understand how people use ChatGPT for both work and personal use. This is the first large-scale analysis of real AI Chatbot conversations.

The study was conducted by OpenAI researchers in collaboration with academics from Duke University and Harvard University.

In this article, we provide a summary of insights, strategies, and key takeaways from the study, specifically tailored for small businesses and solopreneurs.

ChatGPT Usage Insights for SMEs

Key Findings from the Study

Here are the key data points revealed by the study.

• One in ten adults worldwide now use ChatGPT.
• Thirty percent of all ChatGPT queries are work-related.
• Work usage is higher among educated, high-income professionals over 26.
• Writing makes up 42% of work messages and over half in management and business roles.
• About two-thirds of writing requests involve editing existing text rather than creating new content.

The Three Ways We Talk to AI

The study categorized user interactions with ChatGPT into three groups: asking, doing, and expressing.

Expressing

Expressing refers to interactions where users share personal views or emotions without seeking information or triggering any action. This type of interaction is rare in work-related contexts, representing less than 9% of conversations.

The vast majority of AI conversations fall into the asking or doing categories.

Let’s review these two categories in detail.

Asking

Asking is seeking information or advice. It is similar to using a search engine.

Examples of work-related asking messages:

“How do I create a budget for this quarter?”

“What are the key steps to manage cash flow for a small business?”

“How can I reduce my monthly business expenses without hurting growth?”

“How do I run a low-cost marketing campaign for my startup?”

“What tools can I use to automate invoicing and payments?”

Doing

“Doing messages” are requests for ChatGPT to complete specific tasks for the user. Doing messages represent a unique core value proposition of AI.

By asking AI to perform tasks for us, we turn the tool into a personal assistant.  We can give AI a role by prompting “Write this as if you’re a data scientist.” We can upload files and step-by-step instructions and operating procedures.

Examples of work-related “doing messages”:

“Draft a polite refund policy email for a dissatisfied customer.”

“Summarize this 10-page report into five key bullet points.”

“Create a weekly schedule for my team based on these availability details.”

“Analyze this customer survey and categorize responses by theme.”

“Create a concise company description for my LinkedIn profile.”

The chart below shows the breakdown of all work-related conversations. Over 90% of these interactions fall into four main categories: writing, practical guidance, technical help, and seeking information.

Writing is by far the most common type of request, followed by practical guidance.

In terms of communication types, “doing messages” dominate overall, making up 55.8% of all work conversations, with very stronger presence in two areas:  over 80% in writing and over 90% in multimedia tasks. The majority of writing messages ask ChatGPT to edit or refine existing text, not generate new text from scratch.

“Asking messages” are more common in such categories as practical guidance, seeking information, and technical help.

Source: Figure 11, “How People Use ChatGPT,” OpenAI, September 2025.

AI Trends by Occupation

In the study, occupations were combined into five main categories, listed below in order from highest to lowest ChatGPT usage:

• computer-related,
• management and business,
• engineering and science,
• other professional fields such as law, education, and healthcare, and
• nonprofessional (administrative, clerical, service, blue-collar),

The study revealed a mix of expected trends and surprising insights.

As we expected, writing tasks are less common in engineering, science, and computer-related jobs compared to other professions. Instead of drafting reports or polishing documents, people in these roles rely on ChatGPT for technical help, problem-solving, and decision support.

In contrast, if you work in management or business, you’re likely using AI chatbots primarily for writing tasks.

The most unexpected finding is how consistently professionals use ChatGPT across all occupations. Decision-making and problem-solving consistently rank first and second, documenting information always appears near the top, and creative thinking typically claims third place (see illustration below).

Source: Figure 24, “How People Use ChatGPT,” OpenAI, September 2025.

This table presents the frequency rankings of the seven most commonly requested work-related queries. The ranking uses a scale where 1 (indicated in dark blue) represents the most frequently requested activity, with subsequent numbers indicating decreasing frequency of use.

Actionable Insights for SMEs

From Q&A to Task Execution

When it comes to work, professionals are using chatbots more and more to execute tasks, not just gather information. This is a unique and value add feature of all chatbots, which turns them into a personal assistant.

For example, to forecast demand using AI chatbot, follow this four-step process.

First, gather your historical sales data from the past 2-3 years, including monthly or quarterly sales figures, seasonal patterns, product categories, and any relevant external factors like promotions or market events. Then export this data as CSV files or spreadsheets.

Second, upload your sales records to an AI chatbot tool like Claude or ChatGPT and ask it to analyze patterns, identify seasonal trends, growth rates, and anomalies while creating a tailored demand forecasting model.

Third, have the AI generate specific forecasting prompts and questions it should ask you about upcoming periods, including planned marketing campaigns, new product launches, or market changes that could impact demand.

Finally, test your forecasting system by asking your AI chatbot to predict next quarter’s demand, then continuously refine the model by feeding back actual results and updating the system with new data, improving accuracy over time.

Asking Still Matters

While “Doing” tasks dominate chatbot usage, “Asking messages”, where users seek advice, explanations, or guidance, remain important, representing approximately 30% of all work-related interactions.

Simply asking for information creates only surface-level value. The real value occurs when asking messages are blended with doing requests.

Here is an example of how business owners can improve customer engagement by combining asking and doing requests.  

Ask chatbot to analyze your current customer engagement challenges, identify pain points, and recommend improvement strategies based on industry best practices and your specific context.

Then ask your AI chatbot generate concrete, actionable marketing campaigns, complete with messaging frameworks, channel recommendations, and timeline templates tailored to your business model and target audience.

Finally, request specific deliverables like email sequences, social media content calendars, or customer survey templates.

Writing Drives Chatbots Core Value

Writing is a core part of nearly every business activity:

• Drafting proposals for clients
• Writing website copy or blog articles
• Polishing marketing emails or ad campaigns
• Creating customer service templates
• Preparing internal reports or investor updates

The writing process typically involves several key stages: conducting research, gathering and organizing information, creating an outline, drafting the initial version, completing multiple review cycles, and producing the final document. Different AI tools are best suited for each of these stages.

Here is an example of how business owners can create product descriptions using a combination of AI tools. 

Start by gathering competitor product pages, customer reviews from Amazon and social media, sales team feedback, and technical specifications. Use Evernote‘s web clipper to save competitor descriptions and create a dedicated notebook for your product.

Upload all collected materials to NotebookLM. Query it with: “What pain points do customers mention most frequently?” and “How do competitors position similar products?” NotebookLM is designed to synthesize information across dozens of sources.

Feed NotebookLM’s insights to ChatGPT for positioning strategy. Request: “Create 5 different angles for positioning our product based on this customer feedback analysis.”

Use Claude to write the final product description, incorporating ChatGPT’s strategic framework. Claude’s core strengths are emphasis of creativity in writing, understanding long-form context, and maintaining ethical alignment.

Untapped Potential of Creativity Assistance

Results ofthe study “How People Use ChatGPT” suggest that AI assistance is often overlooked, while it has a huge potential for business owners.

Here are some examples of using AI chatbots for creativity assistance:

When stuck on a marketing campaign, ask AI for “20 unconventional ways to promote eco-friendly products.”

Ask an AI chatbot to generate multiple versions of a title for an article, product name, or brand slogan.

AI chatbots can help you cross-pollinate ideas across different fields and industries. Ask “How do luxury hotels create experiences that SaaS companies could apply?” AI might suggest personalized onboarding sequences mimicking concierge services or creating “VIP customer lounges” in your app.

Transform creative blocks by changing viewpoints. “How would a 70-year-old view this fitness app?” or “What would a minimalist designer do with this cluttered website?“

Take one concept across multiple formats. A blog post about work-life balance can become a social media series, podcast episode, infographic, and video script, all maintaining core creativity while reaching different audiences.

AI can analyze hundreds of examples to reveal patterns. Ask “What do successful viral campaigns have in common that failed ones don’t?”

Occupation Matters, But You’re Multi-Role

As a small business owner or solopreneur, you’re likely wearing multiple hats. One minute you’re in “management mode,” making strategic decisions. The next, you’re deep in technical tasks or drafting client communications.

Here’s where AI becomes your ultimate business partner. Whether you need strategic guidance, technical support, or creative assistance, AI chatbot can match your current role.

The key lies in framing your requests according to your current role. Instead of generic prompts, specify your context: “As a marketing manager, help me…” or “From a customer service perspective, how should I…“

The examples below demonstrate how to leverage AI across your various business roles, turning this versatile tool into a specialized expert for each hat you wear.

Key Takeaways

1. Task Execution is the AI Game Changer

AI’s real value lies in task execution. Instead of asking for information, ask AI create demand forecasting models, generate marketing campaigns, and produce ready-to-implement business solutions.

2. Combine Questions with Action Requests

The most effective AI workflow combines information-seeking with task completion. Start by asking AI to analyze your specific business challenges, then have it create actionable strategies, detailed plans, and implementation resources customized to your needs and industry context.

3. Writing Dominates AI Business Use

Writing forms the core of nearly every business activity from client proposals to marketing campaigns. AI chatbots excel at creating product descriptions, polishing communications, and generating customer service templates. Lear how to write with AI.

4. Untapped Creativity Assistance Opportunity

Most businesses overlook AI’s creative potential. Use AI to generate unconventional marketing approaches, cross-pollinate ideas between industries, shift perspectives on problems, and transform single concepts across multiple content formats for enhanced creative output.

5. Role-Specific Prompting

Frame AI requests according to your current business role for better results. Specify context like “As a marketing manager” or “From a customer service perspective” to transform AI into a specialized expert matching whatever hat you’re wearing.

Disclaimer: We are not affiliated with, endorsed by, or connected to any of the products or services mentioned or linked on this article. Any links provided are for informational purposes only, and we do not receive compensation for them. Please use your discretion when engaging with third-party websites.