I spent fifteen years obsessing over rankings. Position one was the holy grail. Position two felt like a consolation prize. Anything below the fold might as well have been invisible.
And then, sometime around mid-2025, I started noticing something uncomfortable. We had clients ranking first for their primary keywords. Beautiful SERP real estate. And their traffic kept declining anyway.
The culprit was obvious once I stopped ignoring it. Zero-click searches now account for 60% of Google searches in the US and EU. Sixty percent. More than half of everyone searching never clicks anything because they get their answer right there on the results page, courtesy of AI Overviews and featured snippets.
Gartner's prediction that traditional search volume would drop 25% by 2026 stopped sounding alarmist and started sounding conservative. We needed a new way to think about visibility, and we needed it fast.
Enter Share of Model.
What Share of Model Actually Means
The term came from Jack Smyth at Jellyfish, and it captured something that had been bouncing around in conversations without a proper name. Share of Model, or SoM, measures your brand's presence within AI data sets as a proportion of total mentions in your category.
Think of it like market share, but for AI's understanding of your industry.
When someone asks ChatGPT or Perplexity about the best project management software, does your tool come up? When Gemini discusses enterprise security solutions, is your company part of that conversation? When Claude synthesizes information about sustainable fashion brands, do you exist in its mental model of that space?
This is fundamentally different from ranking. A ranking tells you where you appear on a list that humans scroll through. Share of Model tells you whether AI systems even know you exist when they're formulating answers about your category. It measures whether these models correctly understand your brand's unique proposition, not just whether they've heard of you.
The distinction matters more every day. When 60% of searches result in no clicks, being position one means less. What matters is whether AI systems incorporate your brand into the answers they generate directly on the search results page.
Why Traditional Metrics Are Failing Us
I was on a call last week with a CMO who was genuinely confused. Their keyword rankings had improved across the board. Their domain authority had ticked up. By every traditional SEO metric, they were winning.
But their organic traffic was flat. Lead generation from search was actually down 15% year over year. Something in their measurement framework was fundamentally broken.
Here's the uncomfortable truth: we've been measuring inputs when we should have been measuring outputs. Ranking position was always a proxy for visibility, not visibility itself. For two decades, that proxy worked well enough because ranking correlated strongly with traffic and conversions. The proxy and the reality moved together.
That correlation is breaking down. When AI systems answer questions directly, when users get information without visiting websites, when the search experience itself has fundamentally changed, our old proxies stop telling us what we need to know.
Share of Model is an attempt to measure something closer to the actual outcome we care about: whether potential customers encounter our brand when they're looking for solutions we provide. In a world where that encounter increasingly happens through AI intermediaries rather than blue links, measuring our presence in those AI systems becomes essential.
The Polling-Based Measurement Approach
Actually measuring Share of Model requires a different methodology than traditional rank tracking. You can't just check whether you appear in position one for a keyword. You need to understand how AI systems represent your brand across a range of relevant contexts.
The approach that's gaining traction involves polling-based models. The idea is to run somewhere between 250 and 500 high-intent queries against various AI platforms on a daily or weekly basis. These queries represent the kinds of questions your potential customers might ask when they're in a buying mindset or researching solutions.
For each query, you track whether your brand is mentioned, how it's mentioned, whether the information is accurate, and what context surrounds the mention. You compare this to how competitors are mentioned in similar queries. Over time, you build a picture of your brand's presence in the collective understanding of these AI systems.
This is more complex than checking a SERP, obviously. But it measures something more meaningful. You're not measuring where a search engine ranks you. You're measuring whether AI systems have internalized your brand's existence and value proposition accurately.
Several tools have emerged to help with this. Otterly.ai has built infrastructure specifically around measuring AI visibility. Semrush has added AI toolkit features to their platform. Conductor has incorporated AI measurement capabilities. Seer Interactive has developed frameworks and templates for tracking AI presence. The tooling is still maturing, but the fundamental measurement problem is becoming solvable.
The Metrics That Actually Matter Now
Share of Model is the headline metric, but it sits within a family of measurements that together give you a picture of your AI visibility health.
Citation Frequency tracks how often AI systems reference your content as a source when generating answers. This is the AI equivalent of earning a backlink, except more valuable in some ways because it directly influences the answers users see.
Brand Visibility Score measures the overall presence of your brand across AI-generated responses in your category. It's a broader measure than citation frequency because it includes mentions even when you're not cited as a source.
AI Share of Voice compares your brand mentions to competitor mentions across AI platforms. This is where Share of Model becomes most tangible. If ChatGPT mentions your competitor twice as often as it mentions you when answering questions about your category, you have a visibility gap that no amount of traditional SEO will fix.
Sentiment analysis in the AI context measures whether mentions of your brand are positive, negative, or neutral. Being mentioned isn't enough if AI systems associate your brand with problems or negative experiences.
And perhaps most interesting is LLM Conversion Rate, which tracks the percentage of AI referrals that convert into customers or leads. Early data on this metric is striking. AI-referred visitors convert at 14.2% compared to Google's 2.8% conversion rate. That's a 5x difference. When someone comes to your site after an AI recommended you in response to their question, they're far more likely to take action.
Why AI Referral Traffic Converts Better
That 14.2% versus 2.8% conversion gap deserves more attention because it suggests something important about the future of acquisition.
When Google sends you traffic, you're getting people who clicked one of several results. They might be casually browsing. They might click the back button in three seconds and try the next result. They arrived because of your position on a list, not because an intelligent system specifically recommended you.
When an AI system sends you traffic, it's because the AI synthesized available information and concluded that your brand was relevant to what the user was asking about. That's a different kind of recommendation. The user didn't pick you from a list; an intelligent intermediary suggested you as a specific solution to their specific question.
This creates a fundamentally different relationship from the first moment. The visitor arrives with context. They know why they're there. They've already received some form of implicit endorsement. Of course they convert better.
This also means that optimizing for AI visibility isn't just about maintaining traffic as search behavior changes. It's potentially about accessing a higher-quality traffic stream than traditional search ever provided.
What AI Systems Actually Care About
Understanding what influences Share of Model requires thinking about how large language models form their understanding of brands and topics.
These systems are trained on massive amounts of text data. Your brand's presence in that training data matters enormously. If your brand appears frequently in high-quality contexts discussing your industry, the model develops a stronger representation of you. If you're rarely mentioned, or mentioned primarily in low-quality or negative contexts, your representation suffers accordingly.
This is why brand mentions have become so important, perhaps more important than backlinks in the traditional sense. A backlink passes authority from one page to another in search engine calculations. A brand mention contributes to how AI systems understand and represent your brand. Both matter, but the mechanisms are different.
What seems to help build Share of Model includes getting mentioned in authoritative publications in your space, appearing in industry discussions and expert roundups, having your brand associated with your key category terms across multiple contexts, maintaining consistency in how your brand is described and positioned, and ensuring that when you are mentioned, the information is accurate and positive.
What hurts Share of Model is perhaps more instructive. Being ignored hurts. If competitors are discussed extensively and you're not mentioned, AI systems will learn that you're not a significant player in your space. Inconsistent positioning hurts. If different sources describe your brand in conflicting ways, AI systems may struggle to form a coherent understanding. Negative associations hurt. If your brand frequently appears in contexts discussing problems, complaints, or failures, those associations get encoded.
The Measurement Infrastructure You Need
Actually tracking Share of Model requires infrastructure that most organizations don't have yet. You need systematic ways to query AI systems with relevant prompts, capture and analyze responses, track changes over time, and compare your presence to competitors.
The manual approach involves literally asking ChatGPT, Perplexity, Claude, and other relevant AI systems questions that your customers might ask. You record the responses. You note whether and how your brand is mentioned. You repeat this regularly to track changes.
This is tedious but enlightening. I'd encourage any marketing team to spend an afternoon doing exactly this as a starting point. Ask fifty questions about your industry, category, and the problems you solve. See what comes up. See who gets mentioned and who doesn't. See whether the information about your brand is accurate.
For ongoing measurement at scale, you'll want to use or build more automated approaches. The tools I mentioned earlier provide different levels of automation and sophistication. Your choice depends on your scale, budget, and how central AI visibility is to your overall strategy.
The key metrics to track include mention frequency across different AI platforms, accuracy of information when you are mentioned, sentiment of surrounding context, comparison to competitors for the same queries, and changes over time as AI models are updated and retrained.
What This Means for Content Strategy
If Share of Model matters, and I believe it does, then content strategy needs to evolve. Creating content that ranks isn't sufficient if that content doesn't also contribute to your brand's presence in AI training data.
This sounds daunting, but the good news is that the changes required are mostly about doing things you should probably be doing anyway, just with more intentionality and scale.
First, brand mentions matter more than ever. Getting discussed in relevant publications, podcasts, industry reports, and expert forums contributes to how AI systems understand your brand. This isn't new advice, but the mechanism through which it matters has changed. These mentions aren't primarily about driving referral traffic or passing link authority. They're about existing in the corpus that AI systems learn from.
Second, clarity of positioning matters more than ever. AI systems struggle with ambiguity. If your brand positioning is muddled, if different sources describe you in different ways, if it's unclear what category you compete in, AI systems will have a harder time forming a coherent representation of who you are and what you do. Clear, consistent messaging across all touchpoints helps AI systems understand and accurately represent you.
Third, being part of comparative discussions helps. When industry publications or analysts compare solutions in your category, being included in that comparison contributes to your Share of Model. Being excluded means AI systems may not learn to associate you with your category at all.
Fourth, factual accuracy about your brand across the web matters more than ever. AI systems synthesize information from many sources. If there's outdated or incorrect information about your brand scattered around the internet, that misinformation may get incorporated into AI models. Audit your brand presence regularly and work to correct inaccuracies.
The Relationship Between GEO and Share of Model
Generative Engine Optimization and Share of Model are related but distinct concepts. GEO is about optimizing your content to be cited by AI systems. Share of Model is about the overall space your brand occupies in AI understanding of your category.
Good GEO contributes to Share of Model. When your content is cited, your brand gets mentioned, and over time those mentions contribute to your presence in AI representations. But Share of Model is broader. It includes brand awareness activities that might never result in direct citations but still influence how AI systems understand and discuss your brand.
Think of GEO as the tactical layer and Share of Model as the strategic measurement. You do GEO to improve your Share of Model, and you measure Share of Model to understand whether your overall strategy is working.
Companies that focus only on GEO tactics without measuring Share of Model may win citations for specific content without building overall brand presence. Companies that track Share of Model but don't do GEO work may understand their position without having tactics to improve it. You need both.
Practical Steps to Start Taking
If you're convinced that Share of Model matters but unsure where to start, here's a practical sequence.
Start with manual research. Spend time asking AI systems questions about your industry. Document what comes up. Note gaps and inaccuracies. This baseline understanding is irreplaceable.
Audit your brand presence more broadly. Where does your brand appear online? In what contexts? Is the information accurate and current? Are you mentioned in the publications and discussions that matter in your space?
Develop a measurement approach. Whether you use existing tools or build something yourself, establish a regular cadence of tracking your AI visibility. Monthly is probably the minimum useful frequency; weekly is better if AI visibility is a strategic priority.
Align your content strategy. Make sure your content efforts are contributing to Share of Model, not just chasing traditional rankings. This doesn't mean abandoning ranking-focused content. It means ensuring that your overall content mix includes brand-building activities that influence AI understanding.
Track the emerging tools. This space is developing quickly. New measurement approaches and tools appear regularly. Stay current with what's available and be willing to evolve your approach as better options emerge.
The Future Is Already Here
I started this piece talking about clients with great rankings and declining traffic. That pattern is becoming more common, not less. The shift toward zero-click search, toward AI-mediated information retrieval, toward conversations with intelligent systems rather than browsing lists of links, is accelerating.
Share of Model is our best current attempt to measure visibility in this new reality. It's not a perfect metric. The measurement approaches are still maturing. The tools are still developing. But the fundamental insight is right: what matters is whether AI systems know about you, understand what you offer, and mention you when relevant questions are asked.
For those of us who built careers on ranking first, this is a significant conceptual shift. But it's also an opportunity. The 14.2% conversion rate on AI referrals suggests that success in this new game might be more valuable than success in the old one. Higher-quality traffic, better-qualified visitors, more efficient conversion.
The brands that figure out Share of Model first will have an advantage that compounds. As more search behavior moves through AI intermediaries, as the 60% zero-click rate creeps toward 70% or 80%, as AI becomes the default way people find information about products and services, the brands with strong Share of Model will be the ones that exist in the answers.
The brands that don't will be invisible, no matter how well they rank.