In this Post:
- Why ChatGPT Product Recommendation (ChatGPT AIPR) Matters
- How ChatGPT AIPR Works (Short Version)
- Visual Explanation (ChatGPT AIPR Atlas)
Why ChatGPT Product Recommendation (ChatGPT AIPR) Matters
People are increasingly using ChatGPT to ask questions.
This ChatGPT adoption is changing how people find answers and solutions.
This is a big market landscape change for all industries.
ChatGPT changes how products are discovered – in a way that fits humans’ most natural behaviors to ask questions.
It presents the need for companies to invest in a long-term strategy.
First to understand how ChatGPT perceives question narratives, then understand how ChatGPT recommends solutions and products that fit the narrative.
In this process, ChatGPT exhibits a pattern of behavior to “find the right answer”
How does a product fit into a solution narrative?
How can you see evidence in real life that prove this?
How can you make this easy for ChatGPT to discover this? Remember this?
This is Narr Theory’s ChatGPT AIPR Strategy in the broader domain of the Voice of AI.
The Key Behavioral Shift – Narratives
The main difference between ChatGPT and Google search – is the user behavior.
On Google, people enter keywords.
In ChatGPT, we have conversations.
Instead of putting in keywords, we can ask “what product should I buy?” or “what solution is best?”
People can ask for comparisons, follow up questions, technical details, and even social validations.
We can leave the conversation for a moment and come back to it later.
We can even use our voice and “talk to AI” in the process.
This stresses the importance of how solutions appear in narratives.
Here is an example.
Keyword Search:
“best car for day-to-day”
Narrative Search:
“I need to buy a new car. I will be using it for day-to-day commute so I want it to be fuel-efficient and easy to maintain. I want a car that won’t break down easily. Preferably a red car but white and silver are also ok. I prefer SUVs but sedans are ok if it fits my budget better. I am looking for values around $55-80K – what do you recommend?”
Notice…
- There are technical constraints (fuel efficiency, maintenance, color, price)
- Tradeoffs (fits my budget better)
- The Narrative (how it all fits into a conversation)
So what happens when ChatGPT is in the position to recommend a product?
ChatGPT Product Recommendation (ChatGPT AIPR) Explained
Understanding how ChatGPT recommends products is similar to how humans recommend – they pass on the narrative of what they know.
Since ChatGPT does not have hands-on experience with products like humans do, AI reads “what others say” at supercomputer scale and synthesize the “Best Recommendation” from the corroboration of evidences.
How ChatGPT AIPR Works (Short Version)
Below is the high level representation of the step-by-step behavior that ChatGPT tend to exhibit when recommending a product. This is not, however, the “exactly how every single AI system thinks” or “every how AI behaves exactly, 100% of the time.”
While Narr Theory strives to obtain such accuracy in understanding ChatGPT’s thinking and behavior patterns, similar to how ChatGPT can tell if a product is the “best in this narrative”, we strive to get closer and closer to the 99% confidence in the overall approach to how this can be examined, understood, and proven.
The below framework is derived from repeated observations, testing, and carefully curated corroboration with first-party authority explanations where available (e.g. OpenAI), rather than having direct access to proprietary secrets that only the publishers of truth like OpenAI or Alphabet can access.
If you want a more detailed breakdown, visit Narr Theory’s ChatGPT AIPR Atlas here.
- User Asks a Question (What product do you recommend for XYZ?)
- This creates a narrative by the user including things like:
- What they need
- Why they need it
- Where
- How
- Constraints, deciding factors
- Tradeoffs
- This creates a narrative by the user including things like:
- ChatGPT “Understands” the Question
- ChatGPT remembers contextual information of the user (if available)
- Their occupation
- Their experience with XYZ
- Their preferences
- ChatGPT also puts it into context of the “real human world”
- The corresponding answer should resemble practical human application and limitations.
- E.g. “something about size..? like drill for food truck”
- ChatGPT remembers contextual information of the user (if available)
- ChatGPT Identifies and Breaks Down Constraints & Tradeoffs
- ChatGPT is great at putting user’s questions into real human context.
- A user asks “car for commute”
- ChatGPT may break it down to “fuel efficient”, “easy to maintain”
- “easy to maintain” – parts network
- ChatGPT may break it down to “fuel efficient”, “easy to maintain”
- User mentions “best for my budget”
- ex. Tradeoff between fuel efficiency vs. price
- A user asks “car for commute”
- ChatGPT has built in heuristics that model the human world. Many times, ChatGPT may suggest “best value for budget” options, or the extreme-end option like “the most fuel efficient car”.
- Often, ChatGPT will consider factors not mentioned specifically in the prompt
- If the user has accessibility issues, ChatGPT may recommend products that fit that criteria.
- Often, ChatGPT will consider factors not mentioned specifically in the prompt
- ChatGPT is great at putting user’s questions into real human context.
- ChatGPT Looks Within Its Internal Model Memory First
- Every time a prompt is entered, ChatGPT is using its latest training module to action on the prompt.
- ChatGPT in fact have some core internal memory for patterns of questions like “how many fingers do humans have?” and even some brand-related heuristics like “what is a good shoe brand?” (Nike, Adidas)
- ChatGPT will first see if they know the answer to the question from “what I already know”.
- If the “knowledge” is not within this internal model memory, ChatGPT performs a web search.
- The more long-tail or niche a narrative, the more likely the “knowledge” will need to searched.
- ChatGPT in fact have some core internal memory for patterns of questions like “how many fingers do humans have?” and even some brand-related heuristics like “what is a good shoe brand?” (Nike, Adidas)
- Every time a prompt is entered, ChatGPT is using its latest training module to action on the prompt.
- External Knowledge Required – ChatGPT Performs Web Search
- ChatGPT has uses Bing and Shopify for their web search, as specified by OpenAI.
- While this does not mean that Bing is used 100% of the time, it is their primary source.
- ChatGPT will turn the user’s “decision criteria” into a differently worded format that will give them the best chance of retrieving the best results in web search.
- Sometimes, ChatGPT may perform this multiple times simultaneously depending on what system you use and what membership you have (e.g. ChatGPT Pro), and the level of truth-confidence that ChatGPT targets to achieve (e.g. 80% vs. 95% confidence).
- While it is uncertain to know exactly how web page results show up on ChatGPT’s web search, we can presume that it resembles how Bing displays results.
- Sometimes, ChatGPT may perform this multiple times simultaneously depending on what system you use and what membership you have (e.g. ChatGPT Pro), and the level of truth-confidence that ChatGPT targets to achieve (e.g. 80% vs. 95% confidence).
- ChatGPT has uses Bing and Shopify for their web search, as specified by OpenAI.
- Corroboration Synthesis
- This is particularly important for the non-technical descriptions of products – the narrative.
- “This product was… most useful, practical, beginner-friendly, easy to use, visually appealing, children loved it, popular, has good reputation, recommended by, best performance… etc.”
- This is the part that especially set AI apart from humans.
- ChatGPT will scan through multiple web pages and look for clues that corroborate or contradicts a certain claim.
- ChatGPT will then synthesize its findings to form its recommendation(s)
- This is particularly important for the non-technical descriptions of products – the narrative.
- User Feedback, Recalibration, and Ongoing Conversation
- The special thing about ChatGPT Product Recommendation (ChatGPT AIPR) is not just that it can find information at superscale. It’s also its conversational relationship with humans.
- ChatGPT can incorporate the user’s personal context from previous conversations into the decision narrative, although not explicitly mentioned.
- ChatGPT is a great solution for people to discuss their workflow and leisure activities. People have conversations with ChatGPT on “how to” conversations, “what is”, “help me solve this problem”, and other types of conversations that may have to do with a certain topic.
- This ongoing conversational nature presents more than one opportunity for the user to consider a product purchase, while their decision criteria potentially gets narrower and more specific.
- ChatGPT Product Recommendation
- Even after a long conversation with the user, the decision-conversation reaches a diminishing point, where any new input by the user, unless the new information presents a new variable, has diminishing impact on the decision criteria.
- At this point, ChatGPT’s position on its recommendation may not change much.
- ChatGPT Product Recommendation is then ultimately formed from either the initial, short-conversation exchange or from a much longer and personalized conversation with the user.
- Even after a long conversation with the user, the decision-conversation reaches a diminishing point, where any new input by the user, unless the new information presents a new variable, has diminishing impact on the decision criteria.
ChatGPT AIPR Atlas
Narr Theory’s ChatGPT AIPR Atlas is an interactive visualization of how ChatGPT recommends a product from the user question, ChatGPT understanding user’s intent, question interpretation, narrative breakdown, knowledge retrieval, corroboration synthesis, and to the final ChatGPT product recommendation.
This can help visualize what gaps may currently be existing for your product’s likelihood of being recommended by ChatGPT, and why competitors are being recommended instead.
Visit the ChatGPT AIPR Atlas here.
Key Takeaway
Whether a user discuss a product recommendation with ChatGPT for 10 seconds, 10 minutes, or 10 months – the main advantage point for ChatGPT is personalization.
As humans’ adoption of ChatGPT increases, ChatGPT’s overall usefulness for real-human conversation and everyday application is also likely to improve.
This presents moments in conversations between humans and ChatGPT, where ChatGPT is in the position as the influencer of knowledge, perception, and decisions.
Having a product recommended by ChatGPT is not as simple as optimizing your website’s GEO (this has minimal impact, in fact, as it only controls the websites you own) or paying web pages to talk about your positively.
It is important to understand and leverage both the tactics that will help with ChatGPT AIPR in short term, and invest in acquiring knowledge, skills, and experience narrative intelligence tools that can help you understand your products’ and brands’ position in the every-day narratives, especially at the points of authority and influence, long term.
ChatGPT may show exponential improvements in the next 3-5 years in its ability to find more accurate, truthful, and credible information for questions.
For this reason, Narr Theory hypothesizes that companies that leverage ChatGPT’s ability to show how people are asking questions and searching for solutions will have be advantageous over others.
It may become essential to develop initiatives to keep a pulse on strategic analysis and discussions around:
- ChatGPT AIPR Gap Analysis
- Narrative Product Position Strategy
- Unfulfilled Narrative Whitespace
- Voice of Customer (Narrative Focus)
- Voice of ChatGPT
Narr Theory, the publisher of this post (not AI generated) and the creator of the ChatGPT AIPR Atlas, specializes in helping companies understand their product and brand’s position in ChatGPT AIPR.
Schedule an introductory call to see where you can start.
Into the Future
In the long-term, it is important to be aware of the the motif of ChatGPT developers to help predict how the ChatGPT AIPR methodology may evolve in the future. At the rate of technology advancement, “tomorrow” may come sooner than later.
Around February and June of 2026, ChatGPT officially released its first advertisements on the platform (US and Canada). This signals OpenAI taking action on its motif of monetization, and willingness to leverage its position in the market.
Narr Theory’s AIPR Participant Motifs (post coming soon) explores these ideas that may influence the landscape both in immediate and distant future.
ChatGPT is intelligent, abstract, and dynamic in nature while its creator, OpenAI, have the motif not to reveal their key secrets to the public.
Narr Theory is dedicated to researching the latest and most proven methodologies in AIPR that is congruent with the first-party publishers of truth (OpenAI, Google) at the highest level of accuracy, and other sources of repeatable and reliable sources of corroborated truth deriving from our epistemic core.
To stay posted on our thought leadership, follow us on LinkedIn or subscribe to our mail list here.
Leave a Reply