Why "Share of Model" Is Becoming a Core Marketing Metric
This article is informed by recent research published in Harvard Business Review examining how agentic AI is reshaping brand discovery, evaluation, and purchase decisions. While the underlying shift is grounded in that research, the analysis, examples, and conclusions here are original and intended as applied interpretation for brand and marketing leaders.
Your brand is being recommended right now by systems you've never spoken to.
They're describing your positioning. Comparing you to competitors. Deciding whether you're relevant to someone's question. And in many cases, they're getting it wrong. This isn't happening in six months. It's happening in every ChatGPT search, every Perplexity query, every AI shopping assistant conversation. Millions of times a day.
Most brands still don't know what AI says about them. They have dashboards tracking every search ranking, every social mention, every review score. But they've never asked Claude or ChatGPT "Who are the top three brands for [our category]?" and analyzed the response.
Welcome to share of model: how often and how accurately your brand appears when AI systems answer questions in your category. If you're not measuring it, the AI is still forming an opinion. It's just doing so without your input.
The Nutrient Company That Disappeared
Here's what this looks like in practice:
A mid-sized nutrition brand spent eighteen months repositioning itself around accessible, science-backed daily supplements. Clear messaging. Clinical studies. Transparent sourcing. Competitive pricing at $40-60 per month. Their content strategy worked. Organic traffic increased. Brand awareness improved among their target demographic.
Then their VP of Marketing asked ChatGPT: "What are the best affordable daily supplements for someone in their 30s?" The brand didn't appear.
When she asked "What are premium, research-backed supplement brands?" they showed up third, described as "high-end" and "specialized," with AI-estimated pricing of $100+ per month. No competitor had outspent them. No campaign had failed. The AI had simply drawn the wrong conclusion from overlapping terminology in their content, categorizing them alongside luxury brands they didn't compete with.
When potential customers asked for affordable options, the brand was invisible. When they asked for premium options, the brand was there but mispriced.
The result: lost demand at both ends of their actual market.
This is the structural problem playing out across categories. Software. Financial services. Travel. Education. If the AI doesn't understand when you're relevant, you're invisible by default. And if it misclassifies you, you're visible in the wrong context.
This is not just a future-state scenario. It's already costing brands revenue.
Why This Changes Everything About Brand Strategy
For decades, brand strategy revolved around visibility and persuasion.
Be seen. Be remembered. Be chosen.
Search engine optimization taught brands to communicate with algorithms, but the output was still a list of links. The user decided what to click. Advertising still worked because persuasion happened before choice. Agentic AI introduces a fundamentally different dynamic. When a consumer asks an AI system to "find the best option" or "handle this for me," the system doesn't present a menu. It presents an answer. Sometimes it completes the transaction itself.
That's the shift from visibility to interpretation.
A brand can be well-known and still misunderstood by AI. It can have strong awareness and still be excluded from recommendations because the model has classified it incorrectly.
Search engines retrieve. AI agents reason.
Search results offer choice. AI recommendations collapse choice.
And as these systems mature, they're increasingly designed to execute multi-step tasks autonomously. Research, comparison, selection, and purchase can all happen inside a single interaction. In that world, marketing doesn't end at persuasion. It must extend into decision logic.
Defining Share of Model
Share of model has three measurable components:
Visibility - How often you appear when category-relevant questions are asked. If someone prompts "best CRM for small teams" fifty different ways, how many times does your brand show up?
Position - Where you rank when multiple brands are presented. First mention matters. So does being the only one the AI elaborates on.
Accuracy - Whether the AI describes your actual positioning, pricing, features, and use cases correctly. Being included but misrepresented can be worse than being excluded.
Unlike share of voice or share of search, this metric reflects meaning, not just presence.
And right now, most brands have no idea what their numbers are.
The Three Modes of AI Intermediation
Brand-consumer relationships are fragmenting into multiple modes, often within the same customer journey. Consider what happens when someone asks an AI to plan a kitchen renovation.
First, they might interact with a cabinet manufacturer's AI tool that helps them configure layouts, estimate costs, and visualize finishes based on room photos. This is Mode 1: Brand agents serving people directly.
Then their personal AI assistant compares that recommendation against alternatives, checking reviews, lead times, and installer availability across multiple brands. This is Mode 2: Consumer agents acting across brands.
Finally, if they've set a budget ceiling and aesthetic preferences, their agent may simply purchase without asking, choosing based on rules and thresholds rather than persuasion. This is Mode 3: Full AI-to-AI intermediation.
Same customer. Same project. Three different points where AI intermediates the relationship. Each mode requires a different strategy. And critically, your ability to succeed in Mode 1 doesn't protect you in Mode 2 or 3.
Stage One: Most Brands Shouldn't Build an AI Agent (And That's Fine)
The instinctive response to AI disruption is often "we need one too." That's a mistake.
Consumer research consistently shows resistance to AI in high-stakes decisions, emotionally meaningful purchases, and contexts where human effort signals care.
Think about categories like estate planning, college admissions consulting, or custom wedding design. Customers want deliberation, reassurance, and human judgment. Automating those moments erodes trust rather than builds it. This is the contrarian truth most brands don't want to hear: AI agents will fail for most companies, and that's fine. The right question isn't "where can we automate," but "where does automation improve the experience without stripping it of meaning?"
For many brands, the answer is nowhere customer-facing. The AI opportunity might be entirely internal: operations, supply chain optimization, customer service routing. But if you do have a legitimate use case, here's what separates success from expensive failure.
If You Build It, Make It a Hybrid
The most effective customer-facing AI strategies blend automation with human judgment. Routine, predictable tasks are handled by AI. Complex, emotional, or ambiguous situations escalate to people.
This model benefits everyone. Customers get speed when they want it and care when they need it. Teams get relief from volume and more time for high-impact work.
A financial services firm tested this approach with retirement planning. Their AI handled contribution calculations, account consolidation logistics, and routine rebalancing. But any conversation touching risk tolerance, family dynamics, or major life transitions immediately escalated to a human advisor.
Customer satisfaction increased. Advisor productivity increased. The AI didn't replace expertise. It freed it for the moments that mattered. Hybrid isn't a compromise. It's an acknowledgment of how trust actually works.
Stage Two: Why Customers Won't Choose Your Agent
Even if you build a capable AI agent, many customers will bypass it.
Independent agents hold two structural advantages that brand-controlled systems can't fully overcome:
Perceived neutrality - They're assumed to act in the user's interest, not the company's.
Broader data scope - They see the customer across brands, contexts, and time. They remember what worked and what didn't.
You can't erase this gap. But you can compete where it matters.
Depth Beats Breadth
Brand agents shouldn't try to replicate general intelligence. They should specialize. Where brands win is in proprietary depth. Real-time inventory. Detailed product taxonomies. Usage history. Operational constraints that only you understand.
An airline's AI can interpret complex fare rules, explain loyalty benefit nuances, and predict delay patterns based on historical data. A general-purpose agent approximates this at best. The challenge is making that value obvious enough that customers choose to engage.
Which brings us to the second critical factor.
Responsible AI Is a Growth Lever, Not Just a Compliance Issue
One of the most underappreciated insights from recent research: responsible AI features influence adoption as much as performance. Sometimes more. Privacy boundaries. Clear disclosures about what the AI can and can't do. Human oversight. Easy escalation when things get complex. These aren't nice-to-haves. They're trust signals.
A healthcare AI that admits uncertainty ("I'm not confident about this, let me connect you with a specialist") earns more trust than one that confidently delivers wrong answers.
Responsible AI is not just a legal concern. It's a competitive advantage. Brands that demonstrate restraint and transparency are more likely to earn trust from both people and the independent agents representing them.
Stage Three: When You Don't Control the Agent
No matter how good your brand agent is, most consumers will rely entirely on independent ones. This creates a new imperative: your brand must be understandable and accessible to external AI systems.
This is less about persuasion and more about integration. If your data is fragmented, inconsistent, or ambiguous, the agent will struggle to interpret it. When that happens, it defaults to brands that are easier to reason about.
Ease of understanding becomes a competitive advantage.
Here's what that means in practice.
Auditing Your AI Reputation
Just as brands track search performance and sentiment, they must now audit how AI systems describe them. This means regularly prompting leading models with category-relevant questions and analyzing the outputs.
The 30-Question Audit:
- Prompt ChatGPT, Claude, Perplexity, and Gemini with 5-10 variations of "best [your category] for [your target customer]"
- Document: Do you appear? What position? How are you described?
- Prompt with comparison questions: "[Your brand] vs [competitor]" - what does the AI emphasize?
- Prompt with attribute questions: "most affordable [category]", "best [category] for enterprises", etc.
- Document every misrepresentation, omission, and inaccuracy
For most teams, doing this manually does not scale. Using a platform like Cited automates this process by continuously monitoring how major AI models represent your brand, surfacing inconsistencies, and showing where your positioning breaks down across prompts and models. You can run one of these audits automatically with any of these described prompts. Track changes. Identify patterns. Implement the solutions through the tool that will improve your AI presence.
When discrepancies appear, the response isn't spin. It's a clarification. Clearer language. Better-structured data. More consistent signals across your owned properties. If you don't correct the model, it will keep guessing. And its guesses will compound.
The New Visibility Formula
Traditional brand visibility was about reach and frequency.
The new formula is: clarity × consistency × machine-readability.
Clarity - Can an AI system understand your positioning from your website, product descriptions, and public content?
Consistency - Do you describe yourself the same way across channels, or does LinkedIn say one thing and your homepage say another?
Machine-readability - Is your information structured in ways that AI can parse, or is it buried in PDFs, images, and unstructured text?
Ambiguity is where AI infers. Inference is where misclassification happens.
Prompt Behavior Is Your New Keyword Research
Small changes in how consumers phrase questions significantly alter AI recommendations. That makes prompt behavior a strategic input. Customer service logs, on-site search queries, and sales call transcripts reveal how people actually ask for help. Those patterns should inform how you structure content and data for AI interpretation.
For example, a B2B software company analyzed six months of chat transcripts and discovered customers asked about "team collaboration" three times more often than "project management," even though the company positioned itself in the PM category. They restructured their content hierarchy to emphasize collaboration use cases. Within two months, their inclusion rate in AI responses to collaboration queries doubled.
Prompt literacy is becoming as important as keyword research once was.
Building for Machine Interpretation
Emerging standards like schema.org markup and structured product data aren't new, but their importance is escalating. Brands that structure product information clearly, consistently, and completely reduce ambiguity.
This means:
- Structured pricing and packaging information
- Clear feature taxonomies
- Explicit use case descriptions
- Machine-readable specifications
A consumer electronics brand added structured data to every product page: compatible devices, technical specs, use case tags, and clear pricing tiers.
Their inclusion rate in AI shopping recommendations increased 40% within three months. Not because they changed their products. Because they made their products easier for AI to understand.
The brands winning at share of model aren't the biggest. They're the clearest.
Why AI Might Actually Make Branding More Important
Here's the second contrarian take: many marketers assume AI commodifies brands by making everything comparable and rational. The opposite may be true. When AI handles functional comparison, what's left is meaning, identity, and values. The things that can't be reduced to a feature matrix. A running shoe brand that stands for sustainability doesn't just compete on cushioning specs. It competes on alignment with the customer's identity. An AI agent might surface that brand because the user has previously expressed environmental values, even if a competitor has slightly better performance metrics.
AI doesn't eliminate brand. It makes brand precision essential.
Vague positioning that tries to be everything to everyone becomes a liability. Clear positioning that owns a specific meaning becomes an asset.
What to Do Tomorrow
Most of this article has been diagnostic. Here's what's actionable immediately.
The Five-Hour Share of Model Assessment
Hour 1: Baseline Audit
- Prompt ChatGPT, Claude, and Perplexity with 10 category-relevant questions
- Document where you appear, how you're described, what's wrong
- Screenshot everything
Hour 2: Competitor Comparison
- Run the same prompts but ask for comparisons: "X vs Y"
- Note what attributes the AI emphasizes for each brand
- Identify gaps in how you're understood
Hour 3: Content Review
- Pull your homepage, product pages, and about page
- Read them as if you were an AI trying to categorize you
- Note every ambiguous claim, inconsistent description, or missing detail
Hour 4: Structure Assessment
- Check if you have schema markup, structured product data, clear pricing
- List every place your information is trapped in images, PDFs, or videos
- Prioritize what to fix first
Hour 5: Stakeholder Briefing
- Document the three biggest misrepresentations you found
- Estimate revenue impact (lost inclusion in relevant searches)
- Propose immediate fixes
This won't solve everything. But it will tell you whether you have a problem and how urgent it is.
Three Questions Every CMO Should Ask This Week
-
If I prompt ChatGPT with "best [our category] for [our target customer]" right now, do we appear? How are we described?
-
What data about our product or service do we have that no general-purpose AI can access or infer? Is that data structured and accessible?
-
If a customer had to choose between our AI agent and an independent one, why would they trust ours more? Can we prove it?
The brands that answer these questions this month will shape how AI represents them.
The ones that wait will inherit whatever the model decided.
The Uncomfortable Reality
AI is not waiting for marketing teams to catch up. It's forming opinions about your brand right now, based on whatever signals it can find. Those opinions influence recommendations. Recommendations influence revenue.
You can ignore this and hope your existing brand equity carries you through. Or you can recognize that brand equity is being reinterpreted through a new lens, and you need to understand what that lens sees. The latter is uncomfortable. It requires new metrics, new audits, and new organizational capabilities.
But the alternative is worse: being misunderstood at scale by systems you can't directly influence. Share of model isn't a nice-to-have metric for the future. It's a current-state diagnosis of how much control you're losing.
The good news: most of your competitors aren't measuring it either.
The question is who moves first.
Sources & Acknowledgments
This article is informed by Harvard Business Review, 
The perspectives, examples, and frameworks presented here are original interpretations intended for applied brand and marketing strategy.