AI Search Is Rewriting the Rules of Visibility. Here’s the Playbook.
Why traditional SEO isn’t enough and what teams should be doing now
Thanks for reading AlphaEngage issue #106. Read past issues.
Inside: How people search for and discover your products and services is changing. Your digital strategy must adapt.
Editor’s Note: This week, I’m trying something a bit different by focusing on a single topic. Primarily because it’s a huge topic that affects nearly every company.
I’ve been working alongside or managing search engine optimization (SEO) teams for decades as part of a broader digital strategy career. I’ve worked on SEO projects for employers, agencies, clients, large companies, small businesses, and startups.
Through it all, one thing has remained constant: someone, somewhere, with some level of influence, is claiming that SEO is dead. Every. Single. Day. For at least 20 years.
But today, their claims hit differently than before. Today, we are witnessing an undeniable shift in user behavior and search experiences that are challenging even the most seasoned and experienced SEOs.
The Rapid Shift Towards AI Search
AI and LLMs (large language models, such as ChatGPT) are evolving at a rate faster than anything we’ve ever seen, including the Internet itself, social media, or even smartphone technology. Increasingly, people are turning to these tools to find the answers they need—answers that are usually much more useful.
As of mid-2025, Google remains the dominant player in search queries by a wide margin. But consider how rapidly LLMs like ChatGPT, Claude, Gemini, Grok, Meta, Mistral, and others have evolved on their own just within the past 12 months. It’s nuts.
And with super-engaging all-in-one tools such as Perplexity, YouChat, and Poe, traditional search engine usage will fade. Not as fast as AI has evolved, but the transition is very real.
You can see this as Google works to push its users to adopt a new “AI Mode” search experience, which utilizes Google’s own Gemini LLM to transform searching for answers into more of a conversation, just like ChatGPT or any other LLM.
While the experience is relatively poor right now—especially for local searches of products and services, which account for a significant portion of Google’s traffic—AI Mode will eventually become the primary interface for Google. That’s almost a guarantee, given the obvious competitive threat, the likes of which Google has never faced before.
Why This Matters for Your Business
Why is all of this important for executives like yourself hoping to read more about leveraging AI across your organization? Because most of your indirect website traffic is almost certainly coming from Google. And it’s about to tank, if it hasn’t already.
Recent studies indicate that nearly 60% of Google searches now result in zero clicks, while websites have experienced a decline of up to 65% in organic search traffic since the introduction of public LLMs, like ChatGPT. Those numbers are terrifying for companies that rely heavily on organic traffic conversions.
This decline can be attributed to several factors, including Google’s massive content deal with Reddit, “AI Overviews” (powered by Gemini) at the top of most Google search results, and, of course, the all-new AI Mode, along with other AI search tools that are quickly growing in popularity.
How it Works Today and Will Tomorrow
Now, before you march down to the marketing team to ask them why your product or service isn’t cited in ChatGPT for a specific prompt, you must first understand these two things very clearly:
LLMs are trained on massive amounts of data and have a knowledge cutoff date. No amount of SEO is going to get you “to the top” of an LLM that lives on top of static data.
Some LLMs can access the web in real time during a session, depending on the platform and tools enabled. They may use search engines, partner data, or proprietary indexed content to provide current information. Retrieval-Augmented Generation (RAG) systems, in particular, use external sources to fetch relevant data during inference, combining live or indexed content with the model’s responses. However, anything retrieved during a session is temporary and does not modify the model’s underlying data or training.
Once you understand that the web search functionality of an LLM sometimes uses live search and other times pulls from pre-indexed content collections, you’ll realize that only the live-search-enabled platforms can be influenced in “real-time” through targeted content initiatives. The others are locked to whatever snapshot they’ve already captured. Either way, the right content in the right places still matters, just like it does in SEO.
Unlike today’s Google, which ranks web pages based on a complex mix of factors—relevancy, authority, backlinks, structured data, freshness, page speed, engagement metrics, and even their own business relationships—LLMs with search care almost entirely about semantic similarity. They’re matching content to the intent and language of the prompt.
While some LLMs use search engines like Bing or Google to fetch links, they don’t rely on those engines' rankings to decide what to cite. Authority or popularity may influence which pages are fetched, but what gets surfaced is based on relevance to the user’s query. That is why LLMs frequently surface obscure blog posts or technical documents that barely crack Google’s top 50, a depth that few Google users ever reach anymore.
You could argue that when LLMs are using Google or Bing to fetch links 50 deep, then optimizing for those two search engines is still relevant as long as you’re listed in the first 50 results. And you would be mostly correct, although it still doesn’t influence whether or not the LLM will surface your site. It does give you a significantly better chance, but the LLM will ultimately decide based on the similarity of the prompt.
Additionally, most major LLMs are actively developing their own proprietary web retrieval systems, which will presumably compete directly with Google and Bing, rendering the above considerations largely irrelevant.
Share of Voice is Back
With LLM web search, we’re returning to an old advertising metric called “Share of Voice,” which measures how often a company, brand, or its products or services are mentioned, in this case, across various LLMs’ core data and indexed or live web searches.
It is important to note that not all LLMs consistently cite their sources. Platforms like Perplexity and ChatGPT (with browsing) often do. Others, such as Claude or Gemini, may reference data without clear attribution, making the measurement of share of voice challenging at best.
There are a few relatively new SaaS offerings that attempt to measure share of voice, and their methodologies are similar since the LLMs themselves do not report this metric (or any, really). The tools essentially automate exact searches across multiple LLMs to determine who appears more frequently than the other.
What Companies Should Be Doing Now
Brands currently attracting significant organic traffic should absolutely continue to execute on proven SEO strategies and tactics. Nobody really knows how long the “blue links” of Google today will last.
Parallel to that, those same companies should begin to follow the Generative Engine Optimization (GEO) path outlined below. And if you’re a startup, we know that LLMs with search capabilities don’t prioritize website age or authority. That means you can flank both traditional SEO and GEO, with a strong lean towards the latter.
Content with Purpose, at Scale
Start by increasing your surface area. That means publishing highly specific, semantically rich content across multiple platforms, not just your site. Think Quora, Reddit, Medium, product review sites, and niche forums.
Mention your brand and product by name. Use the exact phrases people are likely to use in LLM prompts. Don’t chase high-volume keywords; chase clarity and coverage. RAG systems and LLMs don’t care about search volume; they care about relevance.
Focus on extreme specificity. People use AI search differently than they do Google. Get deep into the weeds with your individual pieces of content. High-level, broad content is unlikely to be effective (nor was it with traditional SEO).
This means don’t just write about the product and all of the problems it solves. Write about a specific use case and mention the relevant product. The link back to the product won’t matter to the LLM, either.
Instead, the LLM will surface the use case because that will be what the person searching with the LLM used in their prompt. They need to learn that your product can be used in the XYZ scenario they outline in their prompt. Instead of writing “Our sensors help monitor air quality,” write “A school in Phoenix used the Model X sensor to detect VOC spikes after HVAC failures.”
Just as with SEO, content remains king. But content is also the entire kingdom now. Since you need to focus on so much specificity, it means that one product doesn’t just provide for a few great pieces of long-form, pillar content; dozens or even hundreds of content opportunities can arise from one product or service. And the content doesn’t have to be pillar content—it just needs to answer the prompt with extreme specificity and quotable bits.
Most companies won’t have the resources to scale content like this. Even with AI helping to write content for itself, it will still be difficult. The key to picking and choosing what content to publish is to always be listening. Your teams should constantly be pinging various LLMs to see what surfaces. Just like with SEO, whoever appears first is a target. Do it better, or choose an alternative route.
PR is Back!
Another old-school tactic that will need to make a drastic digital comeback in the age of LLM search is public relations. Obviously, PR is still alive and well, but increasing your surface area also means getting mentions and citations in as many places as possible. Not just brand mentions, but specific products, services, and solutions to problems that are being used in LLM prompts.
It’s not that LLMs are looking to see how often a particular product is mentioned across the web (like SEO backlinks), but that it is mentioned where the LLM is looking at that very moment, and that the mention is surrounded by semantically relevant content as it relates to the prompt.
Operationalizing GEO Across the Organization
Treating Generative Engine Optimization as a new marketing discipline means incorporating it into the operating rhythm, not just a content task. Leadership should establish a structured process for cross-functional visibility, testing, and accountability.
Start by assigning internal ownership. Someone needs to lead GEO as a defined initiative, ideally from marketing or product marketing, but with representation from product, support, and comms. This isn’t just about creating content—it’s about representing the company accurately across emerging AI interfaces. That responsibility doesn’t live with one person or one team alone.
Set up a recurring GEO review cadence—monthly or quarterly, depending on resources. The purpose is to test high-value prompts across multiple LLMs (e.g., “best software for X,” “how do I solve Y”), identify when and where your brand surfaces, and determine which competitors are winning. Think of this like modern keyword rank tracking, adapted for generative AI.
These sessions should be used to prioritize content creation or rework efforts, define which platforms to focus on next, and adjust surface area strategy based on where your products are being mentioned or not. This is where teams can coordinate across silos. For example, product marketing may uncover a gap in messaging, while support teams might already have the answers buried in documentation or help desk logs.
Finally, make sure GEO performance is being tracked and reviewed at the leadership level. Don’t just look at traffic—focus on visibility, brand mentions, and competitive comparisons. If your company isn’t showing up for the kinds of prompts your customers are inputting into LLMs, that’s not just a missed opportunity; it’s a strategic blind spot.
Where This Goes From Here
The shift from traditional search engines to AI-powered search tools won’t fully replace Google overnight. Still, companies that take generative visibility seriously now by scaling purposeful content, expanding their surface area, and embedding prompt-based testing into their processes will gain the same kind of early mover advantage that SEO and social media once provided.
This isn’t about chasing algorithms. It’s about answering real questions in real places. And that’s something leadership teams can and should own from the top down.