When AI Becomes an Advertising Channel

_{Estimated reading time: 8–9 minutes}

By January 2026, the question will no longer be whether conversational AI platforms will integrate advertising models, but how they will manage the critical balance between monetization and user trust.

Just a few days ago, OpenAI officially confirmed what the industry had been anticipating: ChatGPT will begin displaying ads in the coming weeks for users of the free tier and the new Go tier ($8/month). Google has already had ads running in AI Overviews since 2025, reaching 1.5 billion monthly users. Microsoft Copilot incorporates ad formats similar to those used in traditional search. Perplexity launched its advertising program in November 2024, though it paused it ten months later due to scaling and measurement issues.

In short, advertising in conversational systems is not a future possibility—it is a present reality that is redefining the rules of digital marketing.

The traditional advertising model no longer “fits” into new conversational environments

Traditional digital advertising was built on controlled interruption: banners in designated spaces, pre-roll videos of predictable length, and sponsored results clearly separated from organic content. Each format has its own “ad space” where the distinction between content and advertising is clear.

However, in a conversation with AI, those boundaries disappear. There is no scrolling that allows for a clear distinction between content and advertising. There is no visual hierarchy of impact, and no physical separation between the system’s response and the commercial recommendation. Each message is part of a continuous conversational flow where inserting a sponsored message is not the same as displaying a traditional ad: it is shaping a recommendation.

OpenAI has chosen to display ads at the bottom of its responses, clearly labeled as “sponsored.” Google integrates ads directly into AI Overviews when they are relevant to the query. The exact format varies, but the fundamental challenge is the same: how can we preserve the perception of objectivity when there is a commercial transaction behind the recommendation?

From Visibility to Relevance: The New Paradigm for Brands

The key change for brands is that advertising in conversational settings doesn’t follow the rules of traditional marketing. Let me explain why:

• People no longer buy visibility; they seek relevance. Having the biggest budget isn’t enough; the brand must bring real value to that specific conversation.
• The creative message is secondary to contextual utility. A brand may have the best creative campaign of the year, but if it doesn’t address the user’s specific need at that moment, it won’t be seen.
• The brand stops “appearing” and starts being recommended. This nuance changes everything: when an AI system recommends a brand, the user assumes it’s the best option available—not simply the one that paid the most.

OpenAI’s internal projections estimate that “monetization of free users” will generate $1 billion in 2026, rising to nearly $25 billion in 2029. These figures assume that approximately 8.5% of users will convert to paid subscriptions, while the remaining 90%+ will be monetized through advertising and affiliate revenue.

To put this in perspective: only 20 million of ChatGPT’s 800 million weekly active users currently pay for premium tiers, a conversion rate of less than 3%. In other words, advertising isn’t an option for OpenAI; it’s an economic necessity.

The existential risk: undermining the system’s credibility

The most valuable asset of tools like ChatGPT is not their response speed or operational efficiency. It is their perceived neutrality. Users trust that the response they receive is the most appropriate for their question, not the most profitable for the platform; therefore, if that line becomes blurred, the value of the entire system suffers.

Perplexity’s experience is telling. Despite launching its advertising program with high ambitions and prestigious partners (Indeed, Whole Foods, Universal McCann, PMG), the platform paused the acceptance of new advertisers in October 2025. Reasons cited by media buyers include: limited scale, a lack of ROI data, and a lack of measurement tools that meet the standards of established platforms.

A media executive summed it up: “Our hesitation stems from key concerns: limited scale on the platform, a lack of proven ROI, brand safety considerations, and CPM efficiency.”

Any advertising model in conversational settings must meet extremely stringent requirements:

1. Absolute transparency
   Users must know, without any ambiguity, when an answer includes a sponsored recommendation.
2. Real, natural context
   A sales message should only appear if it adds clear and immediate value to the conversation. Forcing irrelevant messages destroys trust.
3. Minimal sales pressure
   A conversation cannot tolerate saturation. The repetition or sales pressure that works in other media is toxic in conversational contexts.
4. Clear distinction between responses and ads
   Internal OpenAI reports suggest that some employees have discussed giving “preferential treatment” to sponsored results over non-sponsored ones. If this happens without full transparency, it will spell the end of user trust.

If these principles are not followed, not only will brands’ reputations be affected, but the adoption of the entire system will be at risk.

Measurement: The Critical Technical Challenge That Will Determine Success

If there is one area where advertising using conversational AI faces its biggest operational challenge, it is measurement. Traditional metrics are not only insufficient; in many cases, they are simply inapplicable.

In traditional display or search advertising, the metrics are clear:

• Impression: The ad was displayed
• Click: The user interacted
• Conversion: The user performed the target action

In a conversation with AI, these definitions break down:

• Is it considered an “impression” when the system mentions a brand in its response? What if it mentions it, but in a negative light?
• Does it count as a “click” when a user asks for more information about the recommended product? What if they ask to compare it with competitors?
• How do we attribute a conversion that occurs days later, following multiple conversations in which the brand was mentioned in different contexts?

The industry needs to develop entirely new measurement frameworks, built specifically for conversational environments. Some emerging components include:

1. Measuring Conversational Influence

Metrics that capture the impact of a mention within the conversational flow:

• Mention Quality Score: assessment of the context and sentiment of the mention (positive, neutral, negative, comparative)
• Conversation Depth: the number of subsequent conversational turns in which the brand remains relevant
• Recommendation Strength: the nature and tone of the recommendation (strong, among options, conditional, not recommended)

2. Conversational Multi-Touch Attribution

The most common attribution models today (first-click, last-click, linear, time-decay) assume a linear customer journey through discrete touchpoints. Conversations with AI break this paradigm:

• A user may have multiple conversations on the same topic over the course of days or weeks
• The same conversation can cover multiple product categories
• “Micro-conversions” (saving a recommendation, asking for specifications) have no direct equivalent in traditional funnels

Emerging solution: attribution models based on “Conversational Sessions” that group interactions by thematic intent, rather than by temporal proximity. Similar to how Marketing Mix Modeling (MMM) models are implemented for offline channels, but with conversational granularity.

3. Longitudinal Metrics vs. Immediate Metrics

Measuring conversational advertising requires:

Real-Time Metrics (In-Conversation):

• Mention Rate: Percentage of relevant responses that include the brand
• Click-Through to Detail: Percentage of users who request more information
• Comparative Position: ranking when presented alongside competitors

Longitudinal Metrics (Post-Conversation):

• Brand Lift in Direct Searches 7–30 Days After Mention
• Conversion Rate on Own Channels (website, app) Attributable to Conversational Sessions
• Incremental approach via A/B tests with control groups (users exposed vs. not exposed to sponsored mentions)

4. Integration with Existing Metering Infrastructure

The real technical challenge is that conversational AI platforms must integrate with advertisers’ existing measurement stack:

Current technical challenges:

• Lack of pixels or equivalent tags: There’s no way to “pixelate” a conversation the way a website is pixelated
• Lack of persistent IDs: Conversations are ephemeral; user IDs are not always traceable across sessions
• Gaps in enterprise measurement platforms: Google Analytics, Adobe Analytics, and other platforms lack native modules for “conversational attribution”

Solutions in development:

• Server-Side Conversational Tracking: Structured events sent via S2S when mentions and follow-ups occur
• Unified Customer Profiles enriched with conversational signals: CDPs (Customer Data Platforms) that ingest conversational events alongside traditional web/app events
• Custom Attribution Models Based on Probabilistic Matching: When there is no deterministic ID, use contextual signals (timing, thematic content, privacy-preserving device fingerprinting) for probabilistic attribution.

5. Reporting and Data Transparency

Currently, data opacity is the biggest obstacle to widespread adoption:

Perplexity:

• It does not provide CTR data
• There is no integration with Google Analytics or Adobe Analytics
• Inability to track cost-per-acquisition
• It does not allow you to create custom audiences or optimize based on performance

OpenAI:

• He indicated that the ads will be subject to “checks” and will allow for user “feedback”
• It has not specified what level of data granularity it will share with advertisers
• The CPV (cost per view) model suggests focusing on awareness rather than direct conversion

Google AI Overviews:

• Native integration with Google Ads
• Specific Placement Reporting (AI Overview vs. Traditional SERP)
• Bidding and optimization capabilities, though still limited

For conversational advertising to be viable at the enterprise level, platforms must:

• Provide granular reporting APIs that allow advertisers to ingest data into their own data warehouses
• Implement verifiable conversion tracking using auditable methodologies
• Provide testing/monitoring frameworks to rigorously measure incremental progress
• Transparency Regarding Ad-to-QueryMatching Algorithms to Validate Relevance

Qualitative Metrics and Their Critical Role

Unlike traditional channels, where everything comes down to numbers, conversational measurement requires qualitative components:

• Sentiment Analysis of Mentions: A mention with a negative context (“Brand X is expensive but functional”) has a different impact than a positive one, even though both count as “mentions.”
• Contextual Relevance Scoring: Was the brand mentioned because it was genuinely the best option, or because it simply appeared on a generic list?
• Brand Safety in Conversational Environments: Monitoring the types of conversations in which the brand appears (to avoid associations with sensitive topics such as politics, mental health, etc.)

Tools needed:

• NLP (Natural Language Processing) for Sentiment Analysis at Scale
• Human review of a statistically significant sample of mentions to calibrate automated models
• Alert systems for brand safety violations

The Future: Hybrid Measurement Models

The reality is that no single metric will be enough. Effectively measuring conversational advertising will require:

1. Extended Marketing Mix Modeling (MMM) models that include “Conversational AI” as an additional channel alongside TV, search, social, etc.
2. Conversational Multi-Touch Attribution (MTA) for brands with complex customer journeys where AI-powered conversations are one of multiple touchpoints.
3. Rigorous incremental testing using geo-tests, user-based holdouts, or synthetic controls to isolate the actual impact of investment in conversational ads
4. Direct business outcome metrics: Since attribution will never be perfect, smart brands will also monitor changes in revenue, market share, brand equity scores, or brand awareness during periods when they scale up or scale back their conversational marketing investment.

In my experience implementing measurement systems for enterprise clients (ranging from retail media to marketing mix modeling for banking and insurance), the key lesson is this: metrics must align with business objectives, not with what is technically easy to track.

The conversational platforms that will succeed are those that offer not only ad inventory but also measurement tools that allow CFOs and CMOs to justify the investment with rigorous data.

Implications for Marketing Teams and Agencies

The structural shift brought about by conversational advertising is redefining the role of marketing teams and agencies:

Before: Buying ad space, optimizing formats, maximizing impact
Now: Designing brand presence in intelligent systems

New critical skills:

• Content Architecture for AI: Preparing data so that AI systems can process, cite, and recommend it. Having a solid semantic layer is essential here.
• Understanding Language Models: Learning how LLMs work to anticipate what kind of information they prioritize
• Data Strategy for Conversational Measurement: Implement a tracking infrastructure that captures conversational events and integrates them with customer data platforms
• Narrative Design: Building Coherent Brand Narratives That Work in Fragmented and Conversational Contexts

Competitive advantage will not lie in who gets the most impressions, but in who adds the most value when the conversation calls for it.

Conclusion: Trust as a Currency of Exchange

The integration of advertising models into conversational platforms is now a reality. ChatGPT officially confirmed this in January 2026. Google has had this feature up and running since 2025. Microsoft integrated it into Copilot. The debate is no longer about whether there will be ads; the real debate is how to preserve trust in an environment where every response carries strategic weight.

Because in a conversation, advertising stops being just an impression and becomes a recommendation. And when that happens, marketing is no longer competing for attention—it’s competing for credibility.

The brands that will thrive in this new paradigm will be those that:

• Invest in creating genuinely useful and expert content
• Implement rigorous measurement systems that demonstrate conversational ROI
• Prioritize contextual relevance over mass visibility
• Ensure consistency between what you promise in conversations and what you deliver in real-life experiences

For platforms (OpenAI, Google, Perplexity, Microsoft), the challenge is even greater: they must monetize without compromising the asset that gave them their reason for being. If they fail to strike this balance, they will not only lose users; they will lose the momentum of a technological revolution.

The future of advertising will not be determined by whoever has the most advanced AI models. It will be determined by whoever can integrate commerce and conversation without breaking trust.

Alex Masip
, Director of Data & Martech – MIO Group

Alex Masip Data & Martech Director – MIO Group