In recent months, businesses are asking more and more questions around the ranking factors of ChatGPT and other AI Search Engines – and with good reason. The technology is advancing at an alarming pace, and may soon become a key platform for search. Understanding how it recommends products, services, and information may become a must-have for future strategies from your SEO and link building agency, and for businesses wanting to dominate the AI search landscape.

Naturally, this prompted some digging for answers, and we’re pleased to report that we’ve found some insights worth sharing.

Before diving in, a quick disclaimer: While this article focuses primarily on ChatGPT as the market leader, many of the points explored here may also apply, in one way or another, to AI search engines like Claude, Perplexity, DeepSeek, Gemini, or even Midjourney.

With that in mind, let’s examine how AI search looks at the start of 2025.

Search Data

In December 2024, OpenAI made ChatGPT’s search function available to non-paid user accounts for the first time, resulting in increased referral traffic for many of our clients and surpassing referrals from other AI search engines. While this move may have been purely strategic on OpenAI’s part to stay ahead of the competition, it also provided valuable data to analyse.

Currently, there’s no way to extract keyword data from AI search engines. This may change in the future, but for now, businesses can still track landing page data, which we know offers various insights. By understanding user journeys – what they search for, why they search for it and which pages they would land on – it’s possible to at least begin the process of optimising websites for a future of AI Search.

Thankfully, this process isn’t entirely dissimilar from ranking in Google, making the concept less “alien” overall. That said, businesses tend to focus on writing and ranking service pages on their websites, rather than supporting ‘top of funnel’ content…which means they run the risk of missing out on AI Search traffic.

This is due to the fact that most AI interactions are conversational, rather than the 2-3 word searches we’re used to entering in the likes of Google. We could search for “financial advice” in Google, but we’re more likely to ask an AI assistant a detailed question, such as: “This is my financial situation. I want to achieve X. How do I go about doing this?” These queries often come with the expectation for the AI to engage in further conversation and follow-up questions.

From our research, it seems this conversational approach to AI search is why we’re seeing its referral traffic landing on very specific product or information pages.

Conversational Search

As touched on above, the biggest advantage of ChatGPT (and indeed, other AI platforms) is the ability to ask detailed questions and receive nuanced, conversational responses. Unlike traditional search engines, which serve static results, ChatGPT’s responses evolve based on user input, creating a more personalised and interactive experience.

That said, we’ve noticed that when topics are raised for which more information is readily available online, ChatGPT becomes more helpful, offering more enriched, accurate and up-to-date responses. Conversely, when less information is available online, or when that information conflicts, responses become more limited and less accurate.

For example, ChatGPT is able to deliver a wealth of expertise regarding consumer rights, employee rights and personal finance, thanks to extremely well-established documentation on websites like moneysavingexpert.com, which it will often link to in responses.

On the other hand, when asked to explain the rules of a tabletop game (as requested by two of our staff members, both of whom are enthusiasts), responses were inconsistent and sometimes incorrect. This is likely because the information ChatGPT draws from comes from a smaller pool of sources – YouTube influencers and casual bloggers – each of whom may be detailing slightly different interpretations of the same rules.

Naturally, it would be beneficial to see how much AI referral traffic ChatGPT and other platforms are being sent to landing pages as a result of this.

Our Research

In addition to our own day-to-day use of ChatGPT and other AI tools, our insights also come from an early study that was commissioned by one of our clients. As part of this research, we asked ChatGPT a series of questions reflecting different stages of a typical user journey. Rather than using keyword-based queries, we framed our prompts conversationally, mimicking how real users interact with AI – explaining a situation, objectives, and challenges.

ChatGPT provided step-by-step solutions to each scenario, often recommending online resources and suppliers. While not the case in every test, it would typically do this without being prompted.

When asked why it had chosen to reference or suggest specific websites, ChatGPT openly disclosed its ranking process. It had searched Google’s SERPs, weighed the value of results based on a number of ‘trust’ factors across the internet as a whole, and then made recommendations accordingly.

ChatGPT Search Ranking Factors

Google Rankings

Our study suggests that ChatGPT considers Google search engine rankings when generating recommendations, typically selecting 3–5 results based on additional factors outlined below.

Brand Mentions

There appears to be a preference toward brands and products that are frequently mentioned across the web in relevant contexts. This aligns with recent trends in Google’s algorithm, where brand mentions are gaining importance in ranking for generic, high-volume search terms.

Recommendations & Reviews

The number of positive reviews and overall sentiment around a brand play a role in AI search rankings. ChatGPT draws from Trustpilot, Amazon, company websites, and blog reviews, but it doesn’t just count the results; it analyzes their content to extract key selling points and USPs.

Resources & Supporting Content

Websites with in-depth guides, tools, and helpful resources tend to be recommended more frequently. This includes free trials, demos, and sample offerings.

Domain

e SEO professionals speculate that domain authority and age influence ChatGPT rankings, but our findings suggest these factors are correlations rather than direct ranking signals. Household names like Ikea naturally score higher due to their extensive brand mentions and high volume of user-generated content rather than simply because of their domain authority.

Conclusion

Beyond boosting your Google rankings, it seems that having a great product or service, supporting resources, positive customer reviews and a balance of recommendations on niche-relevant and authoritative websites makes you more likely to be recommended by ChatGPT.

Due to the “conversations” users are having with ChatGPT and other AI tools, we expect to see greater value in supporting users at every step of their journey, not just the step before they come to your website to convert.

For years, Google has encouraged companies to engage in holistic online marketing, but never really seemed to reward it. Now, it seems ChatGPT may be moving toward rewarding this behaviour and, in return, rewarding companies who follow a new form of “best practise”.

Error Correction: The Key to Quantum Stability

One of the most significant challenges in quantum computing has been managing errors caused by the fragile nature of quantum states. Willow addresses this issue head-on by demonstrating an exponential reduction in errors as the number of qubits increases. This is a pivotal achievement in quantum error correction, marking a turning point in the quest for stable quantum systems. The chip’s ability to maintain coherence over extended computations has been lauded as a “30-year challenge” overcome in the field.

Unprecedented Computational Power

The power of Willow’s 105 qubits was highlighted in a recent demonstration where it solved a problem in under five minutes that would take the fastest classical supercomputers an estimated 10 septillion years to complete. To put this into perspective, 10 septillion years is many times the age of the universe. This feat underscores the transformative potential of quantum computing in addressing problems previously deemed intractable.

Real-World Applications

Google’s CEO, Sundar Pichai, has emphasized the wide-ranging implications of this breakthrough. From accelerating drug discovery and enhancing battery design to advancing fusion energy research, Willow opens doors to solving some of the most complex challenges facing humanity. “Quantum computing has the potential to revolutionize entire industries,” Pichai noted, highlighting Google’s vision for harnessing this technology to create tangible global benefits.

Market Impact and Future Prospects

Following the announcement, Alphabet’s stock saw a significant rise of over 5%, reflecting investor confidence in the potential of quantum computing. While Willow’s advancements are impressive, experts caution that practical, large-scale applications of quantum computing remain years away. Significant engineering challenges, such as scaling qubit counts and integrating quantum systems into existing infrastructures, must still be addressed.

A Quantum Future

Developed at Google’s quantum AI lab in Santa Barbara, California, Willow is a testament to the progress being made in this nascent yet rapidly advancing field. The chip’s success represents not just a technical milestone but a beacon of what the future holds. As quantum computing continues to evolve, the possibilities—from accelerating scientific discovery to reshaping industries—seem limitless.

Google’s Willow has set a new benchmark in quantum computing, showcasing the profound potential of this technology to solve problems on a scale unimaginable with classical computers. While challenges remain, the path forward looks brighter than ever, with Willow as a cornerstone for what promises to be a quantum-powered future.