The place Are We & What Can We Do About It?

The world of search is present process its most vital transformation for the reason that creation of Google itself. At a latest HiveMCR occasion in Manchester, I shared crucial insights into this shift, exploring the place we stand with AI search and what search engine optimisation professionals can do about it.

We’re at the moment in a pivotal time for search. Generative AI providers are beginning to influence search functions, with Apple even noting a fall in Google searches in Safari as a consequence of this shift (although Google has reportedly denied this).

Past this, the search engine optimisation panorama is seeing visitors decreases attributed to Google’s AI Overviews (AIOs) and different rising SERP Options. Google appears to be reacting, for instance, by together with hyperlinks in AI Overviews, which seems to be rising pages per go to for them.

Whereas this would possibly profit Google, it presents a problem for web site visitors, making it a broader internet concern, not only one for Google or search engine optimisation. This uncertainty is resulting in points with buy-in and sources, making it tougher than ever to realize objectives and promote search engine optimisation providers.

How can we transfer ahead? Let’s reply key AI search questions with knowledge:

How LLMs & AI search are actually affecting Google visitors?
What are the alternatives for SEOs inside AI search?
How can we take advantage of out of those alternatives?

The Present Actuality: AI Visitors is Nonetheless Small

Regardless of the excitement, knowledge reveals that AI visitors continues to be tiny for many websites in comparison with conventional natural search. It at the moment represents solely 1-2% of referral visitors for almost all of internet sites. The income potential from AI search can also be presently trivial in comparison with conventional search.

Moreover, knowledge means that customers usually are not ditching Google completely for brand spanking new AI platforms like ChatGPT or Perplexity. As an alternative, most customers are using them as extra platforms.

A big 98.1% of ChatGPT desktop guests are additionally shared with Google, with only one.9% being unique. Equally, 98.2% of Perplexity guests are shared with Google, leaving only one.8% unique.

Whereas ChatGPT visitors is rising, it’s nonetheless minimal in comparison with Google (5.14B vs 81.31B visits in April 2025). Google has skilled a slight yearly visitors lower (-0.89% YOY), nevertheless it’s Bing that has seen a a lot bigger drop (-21.76%) within the final 12 months.

Google holds important benefits with its product, enterprise, person knowledge, and loyalty. They’re actively accelerating releases to meet up with the “first mover” benefit held by OpenAI/ChatGPT. Even Sam Altman from OpenAI reportedly doesn’t imagine ChatGPT will substitute Google.

Let’s summarize:

LLMs Visitors and Referral Visitors from LLM’s to websites are rising however nonetheless tiny compared to conventional search (1-2% referral).
Persons are not ditching Google for ChatGPT or Perplexity however utilizing them as extra platforms.
Google had only a -0.89% yearly visitors lower whereas Bing had a -21.76% one.

Google received’t let ChatGPT win the AI search race with out leveraging its person and product knowledge benefit, they’re catching up with OpenAI/ChatGPT “first mover” benefit, accelerating releases, and even Sam Altman from OpenAI doesn’t assume ChatGPT will substitute Google.

However I additionally thought we really needed to have a extra numerous search panorama too and disliked Google Monopoly proper? We’re obscure, I do know. Let’s think about we’re  already in 2030…

A Glimpse into 2030 (Speculative)

Think about looking forward to 2030. A future the place all search platforms have transitioned to an “AI Agent First Expertise”. This expertise is envisioned as Predictive, Conversational, Personalised, and Multimodal, importantly linking prominently to sources, improved from the preliminary AI Overviews. By this time, Google would possibly nonetheless maintain nearly all of the search market share (speculated +50%), with ChatGPT (20%), Bing (15%), Perplexity (10%), and others (5%) additionally having notable shares.

On this potential future, youthful customers would possibly proceed to desire social platforms like TikTok for discovery, empowered by AI search options. Most customers would nonetheless make the most of engines like google like Google or Perplexity with AI enhancements that personalize the interface based mostly on the question. Others would possibly desire LLMs like ChatGPT or Gemini as their main search and assistant platforms for advanced informational duties.

By 2030, SEOs ought to have first-party entry to LLM question and efficiency knowledge, understanding their influence and conduct to develop differentiated methods for every platform.

The main focus would shift to primarily monitoring model visibility, citations, and income throughout numerous LLMs and search platforms. SEOs would notice they’re findability specialists able to connecting companies with their audience on no matter platform they use, celebrating that not all their efforts are focused on a single platform anymore.

Good! Nothing to fret about. Let’s return to 2025 to know tips on how to get ready and take advantage of out of the brand new AI search actuality.

Again to 2025: Understanding AI Overviews

Returning to the current, Google has spent the final 12 months rolling out and testing AI Overviews (AIOs). Research signifies that AIOs can hurt Click on-By means of Fee (CTR) and visitors, even for websites featured inside them.

Nonetheless, it’s essential to notice that AIOs are primarily proven for informational queries, the place a abstract is extra useful than for business or transactional ones. AIO inclusion varies considerably by business too, being excessive in Well being however minimal in e-commerce.

Additionally, whereas AIOs are showing extra typically for business queries, within the latest analysis carried out by Kevin Indig, they discovered that 80% of customers looking for business/transactional phrases nonetheless click on on non-AIO outcomes.

So, for many companies (except they’re publications monetizing via adverts or associates), the queries most impacted are broad informational and navigational phrases, not usually the money-making ones. Branded and navigational queries usually tend to nonetheless drive visitors.

As you possibly can see, context performs a significant position however for many SEOs their very own AIO’s influence continues to be unclear since evaluation is just not straight-forward with out a Google Search Console filter. Nonetheless, it’s doable and demanding:

This can enable you resolve that are the AIO’s price pursuing for inclusion & what to focus on as an alternative:

How related & common are these queries in your clients shopping for journey?
What’s the influence in your model visibility and authority recognition towards a related viewers?
What’s the influence on potential gross sales? Income?
What’s the influence on clients help? Operational Value?

It is best to use these not just for AIO’s potential evaluation however any LLMs and search platforms. Keep in mind that it’s additionally about model recognition, authority and belief constructing too.

Optimizing for the AI Search Period

So, what if you happen to resolve to focus on and optimize for AIOs and different LLMs? AIOs differ from conventional LLMs as they’re designed to be grounded in data from the online, powered by Gemini 2.0, and built-in with Google’s rating programs and information graph. Google’s official stance is that no extra optimization is required past common search tips for AIOs. Nonetheless, this isn’t completely useful recommendation, after all.

Knowledge reveals that 52% of AIO citations usually are not even inside the high 50 conventional outcomes, and 29% hyperlink to extra Google properties.

Analysis from Profound has additionally proven that there’s additionally solely a 12% overlap between Google’s conventional search outcomes and the sources utilized in ChatGPT solutions. Totally different LLMs present distinct preferences in sourcing data.

Though the pillars of conventional search and LLMs optimization are shared, optimizing for search-integrated LLMs has some key variations in comparison with conventional search:

Some LLMs, like ChatGPT and Claude, don’t execute JavaScript (in contrast to Gemini), making Server-Facet Rendering (SSR) crucial for essential content material to look of their solutions.
Web page expertise can also be not a direct rating issue for LLMs in contrast to conventional search outcomes that take them under consideration.
Citations play a a lot larger position for LLMs than for conventional search outcomes.

Past technical issues, enhancing content material semantic relevance and entity recognition is much more essential than earlier than with LLMs, though additionally essential to optimize content material for contemporary engines like google. Figuring out subject gaps, bettering topical authority, and refining content material construction and relationships are key. Establishing your model visibility and recognition is important too.

The largest modifications between LLMs vs conventional search are noticed in person conduct: the size, complexity, nature of queries, subjects searched, and interplay kinds. Because of this it’s essential to start out monitoring LLM prompts and the content material they set off, along with AIOs, which is now doable because of Similarweb new AI Chatbot visitors report.

Key Efficiency Indicators (KPIs) additionally change. Conventional rankings usually are not related in LLMs. As an alternative, success ought to be measured by model citations in sources or model mentions and hyperlinks inside the LLM solutions.

It’s really helpful to start out monitoring your model’s inclusions, hyperlinks, and citations versus rivals in every goal LLM, and to watch the related gamers and common prompts per subject on every platform, which is feasible because of instruments like Waikay, Rankscale, or Peec.ai, and Dejan AI Rank.

That is essential as a result of completely different fashions cite completely different sources for a similar questions, knowledge from Profound reveals a significant distinction in AIO’s & ChatGPT sources: Solely 21% domains & 9.9% URLs overlap. Every LLM reveals completely different choice patterns when sourcing data, which can also be really helpful to consider.

For instance, based mostly on James Allen analysis we will see ChatGPT closely skews towards established, authoritative, and factual sources, not that includes a lot UGC. Nonetheless, Gemini blends authoritative sources (blogs, information) with group enter.

The clearest takeaway is that our present search engine optimisation abilities are transferable, and the rise of AI search merely means there’s now a broader pool of platforms via which to attach with our viewers. Even Google reportedly agrees with this.

This shift raises an attention-grabbing query: are we nonetheless doing “search engine optimisation” or one thing else? Those that optimize for YouTube or TikTok search are arguably doing a type of search engine optimisation; optimizing for LLMs and AI search options matches into this evolving definition, and the best way we name it must do extra with market calls for and the way we bundle the providers.

The decision to motion is evident: begin monitoring, analyzing, and adapting. By understanding the nuances of AI search options like AIOs and the conduct of various LLMs, SEOs can put together and take advantage of this new actuality. We’re evolving into findability specialists, able to navigate the more and more numerous search panorama.