As marketing leaders, we are witnessing a fundamental shift in how customers discover and interact with our brands. The traditional metrics of impressions, sessions, and click-through rates, while still relevant, no longer tell the complete story. Generative AI systems like ChatGPT, Gemini, and Perplexity are increasingly becoming the primary interface for the early stages of the customer journey, the top-of-the-funnel (TOFU) research phase that once drove millions of users to our websites. In this new landscape, visibility is not just about who ranks in search results, but who gets referenced and cited within the AI models that guide user decisions. This requires a strategic pivot in our content strategy, one that prioritizes mentions and citations as the new levers of trust and, ultimately, revenue. Recent data from a comprehensive two-year study by Siege Media, which analyzed over 7.2 million sessions across various industry blogs, reveals a significant trend: while top-of-funnel guides and "how-to" posts have seen a sharp decline in traffic, bottom-of-the-funnel (BOFU) content such as pricing pages, calculators, and comparison articles have experienced major growth. This does not mean that TOFU content is no longer important. On the contrary, it suggests that users are now conducting their initial research within generative AI platforms and only visiting websites when they are further down the funnel and closer to a purchasing decision. This is supported by the fact that engagement has increased across all major content categories, indicating that when users do arrive on our sites, they are more motivated and ready to act. The TOFU Paradox: Why Top-of-Funnel Content Matters More Than Ever This shift creates a paradox for content strategists. While TOFU content may be driving less direct traffic to our websites, it is more critical than ever for building brand awareness and trust. Generative engines are now the primary channel for the TOFU and middle-of-the-funnel (MOFU) stages of the customer journey. Users are having conversations with AI, asking questions, and getting recommendations, all without ever visiting a brand's website. When they finally do click through to a site, it is often to make a purchase or get specific, transactional information.

For years, SEO has been a game of keywords. We targeted them, tracked their rankings, and built our content around them. But the ground has shifted. Modern search engines, powered by sophisticated AI, no longer just match strings of text; they understand the real-world concepts—the people, places, and ideas—behind the words. This is the world of entity-based SEO, and for marketing leaders, it represents a fundamental change in how we build authority and win visibility. Understanding and mastering this shift is no longer optional. It’s the foundation of a resilient, future-proof SEO strategy that ensures your brand is not just seen, but understood by both users and the AI systems that guide them. Beyond Keywords: What Is an Entity? In the simplest terms, an entity is a single, well-defined thing or concept. It can be a person (Elon Musk), a place (the Eiffel Tower), an organization (Apple Inc.), or a concept (climate change). Unlike a keyword, which is just a string of text, an entity has attributes and relationships that give it context. Google’s Knowledge Graph, a massive database of billions of entities, knows that the Burj Khalifa is a building, that it’s the world’s tallest, and that it’s located in Dubai. It’s this web of understanding that allows search engines to answer complex questions, not just point to pages with matching words. This distinction is critical. A keyword is the language a user types; an entity is the meaning they intend. By focusing on entities, you align your strategy with how search engines actually think, moving from a purely linguistic game to a conceptual one. Why Entities Are the Bedrock of AI-Powered Search The rise of generative AI and Large Language Models (LLMs) has accelerated the importance of entities exponentially. AI systems like Google’s AI Overviews and ChatGPT don’t just crawl your content for keywords; they ingest it to learn about the world. They build their understanding of your brand, products, and expertise based on the entities you are associated with.

As marketing leaders, we are under constant pressure to adapt our strategies to the ever-evolving digital landscape. The rise of AI assistants has been a particularly disruptive force, and many in our industry have rushed to "optimize for AI" under the assumption that these platforms are simply a new channel for commercial queries. However, a recent analysis by Dan Petrovic, Director of the AI SEO agency Dejan, challenges this assumption and suggests that we may be fundamentally misreading how consumers are using these powerful new tools. Petrovic's research, which analyzed 4.4 billion characters, 613 million words, and 3.9 million conversation turns, reveals that a staggering 65% of AI chats have no commercial intent whatsoever. This finding has profound implications for our content strategies, resource allocation, and our understanding of the customer journey in the age of AI. How Users Actually Engage with AI Assistants The data shows that AI users behave very differently from traditional searchers. While a typical search engine query is a discrete event, an AI chat is often a multi-step task. The median chat is just two turns—a quick question and a quick answer—but this masks a long tail of more complex interactions. Over 80% of chats are under 1,000 words, but a small percentage (4.2%) exceed 2,500 words, representing high-value tasks such as editing, coding, tutoring, and data analysis. The typical user contributes only 16-17% of the conversation, with the AI assistant generating approximately 1.5 times more content than the user inputs. This pattern reveals that users are not simply querying for information; they are engaging in collaborative problem-solving sessions where the AI serves as an active partner rather than a passive search engine. When we examine what users are actually doing in these conversations, the non-commercial nature becomes even more apparent. Petrovic classified 24,259 sessions across 42 intent categories and found that the vast majority of interactions fall into categories such as brainstorming (7.7%), planning (6.5%), conversation and emotional support (6.2%), analysis (5.7%), learning (4.7%), transformation tasks like summaries and translations (4.6%), and creation activities including writing and coding (3.9%). A full 25% of interactions fell into an "other" category that included highly specialized requests, roleplay scenarios, and various experimental uses of the technology.

For multi-location brands, the Google Local Pack represents one of the most critical battlegrounds for customer acquisition. However, the long-held belief that visibility can be achieved through a standardized, checklist-based approach to local SEO is rapidly becoming obsolete. Google's ranking system is not a monolith; it is a dynamic, context-aware engine that rewards businesses based on a principle we call signal-fit. This is the degree to which a business listing aligns with the specific, nuanced expectations of local users within a particular industry. As marketing leaders, it is imperative that we move our organizations beyond generic playbooks and embrace a hyper-localized strategy that treats each location as a unique market. Recent analysis from Yext, covering 8.7 million Google Business Profiles across five major U.S. industries, confirms that brand size and advertising spend are not the primary drivers of Local Pack visibility. Instead, what truly moves the needle is the alignment of a business's digital signals with the specific intent of local consumers. Factors such as review cadence, photo quality, and profile completeness all play a role, but their importance is weighted differently across industries and even geographical regions. This data underscores a fundamental truth: Google plays favorites, but its preferences are rooted in a listing's ability to precisely match local context and a user's immediate needs. Decoding Signal-Fit: Industry-Specific Optimization Strategies The concept of signal-fit is best understood by examining its expression across different industries. A one-size-fits-all approach to Google Business Profile (GBP) optimization will inevitably lead to missed opportunities and wasted resources. For multi-location brands, this means that operational simplicity cannot come at the cost of local relevance. In the hospitality sector, functional information carries significantly more weight than aesthetics. Business hours, detailed descriptions, and complete profiles matter most to travelers, who value reliability over polish. Photo volume beyond a reasonable threshold adds little advantage, as guests care less about another angle of the pool and more about whether there is parking when they arrive at midnight.

Knowledge graphs are the foundation of how AI systems understand relationships between entities. Google's Knowledge Graph, Wikidata, and proprietary AI knowledge bases all work similarly: they map connections between people, places, concepts, and organizations. The Knowledge Integration pillar of the SPARK Framework™ focuses on getting your brand properly represented in these graphs. This means structured data markup, consistent NAP information, authoritative backlinks, and content that clearly defines what you do and who you serve. When AI systems query these graphs, they're looking for entities with strong, verified connections. The more robust your knowledge graph presence, the more likely you are to be cited as an authoritative source. Question for the community: What tools or strategies have you found most effective for improving knowledge graph representation?