The Post-SEO Advertising Landscape: A Scenario Planning Analysis

The Current Inflection Point

The old model was straightforward: Google ranked pages, brands optimised for keywords (SEO), and paid for premium placement (SEM). The user clicked links. That entire loop is breaking. AI search traffic is already converting at 4.4x the rate of traditional organic search, and 58% of consumers now rely on AI for product recommendations, more than double from two years ago.

According to Gartner, traditional search engine usage is expected to decline 25% by 2026, and organic traffic could drop up to 50% by 2028.

The fundamental shift: in traditional search, brands competed for placement around content. In LLMs, ads won't be placed around content, they'll be woven into the reasoning layer itself. Influence moves from the page to the model.

Three Scenarios for "Pay to Play Inside LLMs"

SCENARIO 1: The Auction Model (Most Likely, 2025–2028)

This is the Google Ads model adapted for conversational AI, and it's already materialising. At CES 2026, agency leaders from WPP, Horizon Media, and Omnicom agreed that auction-style dynamics are likely to carry over, with advertisers paying for top billing within specific results.

What it looks like in practice:

1. Sponsored answers. A user asks "what's the best project management tool for a remote team?" and a paying brand gets mentioned first or most prominently in the response. Perplexity has already launched this with "sponsored related questions", a non-intrusive format where paid content appears as suggested follow-up prompts rather than embedded in the answer itself.
2. Contextual intent matching. Instead of bidding on keywords, brands bid on intent categories. LLMs have an advantage over search here: a user's conversation history contains specific dimensions, brand preferences, past purchases, and style preferences, far richer than any keyword or cookie. An LLM that knows you're 6'7" and have struggled with clothing fit can serve a genuinely useful sponsored recommendation that a search ad never could.
3. Tiered visibility. Free-tier users see more sponsored content; subscribers see less or none. This mirrors the emerging hybrid model across platforms.

The economics: OpenAI loses money on its $200/month subscription tier because users consume more compute than expected. About 75% of revenue comes from paid subscriptions, but the majority of users remain on free tiers. Ads become the mechanism to subsidise the enormous inference costs for free users.

SCENARIO 2: The Data Licensing & Embedded Commerce Model (Emerging, 2026–2030)

This is the more radical structural shift, where brands don't just buy ad slots, they feed their product catalogs, inventory data, and pricing directly into the LLM's retrieval layer so it can make real-time commerce recommendations.

Criteo is already building this, positioning its commerce dataset (720 million daily active users, over a trillion dollars in observed transactions, billions of SKUs) as the "missing ingredient" that makes product discovery inside tools like ChatGPT actually work. The model isn't traditional advertising, it's closer to data licensing combined with affiliate commissions.

What this looks like:

1. Pay-per-query models. Instead of CPM or CPC, brands pay when their product data is retrieved and used in a response. Criteo is exploring data licensing, pay-per-query, and native ad formats inside LLMs, the commercial models are still fluid.
2. Affiliate-style commissions. The LLM recommends a product, the user buys it through a tracked link or in-chat checkout, and the LLM platform takes a cut. This makes the LLM function like a concierge earning commission rather than a billboard owner selling space.
3. Branded environments. Entire GPTs, plugins, or agents operate as branded experiences within the larger LLM ecosystem, a Nike shopping agent, a Fidelity financial planning agent, a Mayo Clinic health agent.

SCENARIO 3: The "Trust Tax" Model (Aspirational, 2027+)

This is the scenario where the tension between monetisation and user trust produces something genuinely new. Conversational AI platforms are built on an implicit promise: "I am neutral. I am objective. You can trust my answers." When ads arrive, these platforms must pay a "Trust Tax", the cost of preserving user belief in the system's integrity.

This pressure could push toward a model where:

1. Transparency is mandatory. Every sponsored recommendation is clearly disclosed, with the reasoning for why it was included. Unlike Google, where the "Ad" label became practically invisible, LLMs may be forced (by regulation and competitive pressure) to say something like: "I'm recommending Brand X here, they're a sponsor, but they also genuinely match your criteria based on these independent reviews."
2. Quality gates replace pure auction. Paid placements could actually improve recommendations by introducing real-world validation, when a brand pays for placement in a high-trust environment, it implicitly vouches for its product, raising the bar. LLM platforms would reject sponsors whose products don't meet quality thresholds, because a bad recommendation degrades trust in the entire system.
3. Dual-currency bidding. Brands compete on both price and quality score (reviews, customer satisfaction, return rates), creating a system where paying more doesn't guarantee placement if your product isn't good.

The Parallel "Organic" Game: LLMO

Running alongside all three paid scenarios is the organic equivalent of SEO for LLMs, already being called LLMO (Large Language Model Optimisation) or GEO (Generative Engine Optimisation). Adobe launched its LLM Optimiser in October 2025, observing a 1,100% year-over-year increase in AI traffic to U.S. retail sites, with AI-sourced visitors 12% more engaged and 5% more likely to convert than traffic from paid search, email, or social media.

The tactics are familiar but different in emphasis: getting mentioned on high-authority sites that LLMs cite (Wikipedia, Reddit, review platforms), structuring content so it's easily retrievable by RAG systems, building entity-level brand recognition rather than keyword rankings. Brands in the top 25% for web mentions already get 10x more AI visibility than others.

What Brands Should Prepare For

The structural reality across all three scenarios:

1. The funnel collapses. The traditional marketing funnel falls apart when the AI assistant handles awareness, evaluation, and action in a single prompt. A user can go from "I need running shoes" to "ordered" without ever visiting a brand's website.
2. Measurement becomes harder, then different. Click-through rates are meaningless when the user never leaves the chat. New metrics will centre on mention frequency, sentiment, position within the response, and conversion attribution through tracked links or in-chat actions.
3. Brand narrative matters more than ever. When an LLM is synthesising information about your brand from thousands of sources, the coherence and clarity of your story across the internet becomes your most important asset. Contradictory messaging, poor reviews on influential platforms, and weak third-party coverage will all directly impact how (and whether) an LLM recommends you.
4. The first-mover window is now. OpenAI has confirmed internal tests of conversational native advertising, expected to launch within the next year. ChatGPT has surpassed 800 million weekly active users. Every significant public LLM is expected to offer an ad network within the next 18 months.

The brands that will win in this environment are those treating LLM visibility as a core strategic priority today, not waiting for the ad platforms to be fully built before engaging.