AI Advertising Has a Visibility Problem. Similarweb Wants to Fix It

Advertisers have spent years navigating the complexities of social media algorithms and search engine optimization, but the emergence of AI-powered conversational platforms has introduced an entirely new challenge: understanding where their ads appear when users are chatting with machines rather than scrolling through feeds.

Similarweb's launch of its AI Ads dataset marks a significant shift in how advertisers can monitor and optimize their presence across ChatGPT, Google AI Mode and Google AI Overviews. Until now, brands operating in these spaces have been flying blind, unable to see their own ad placements or gauge how they stack up against competitors. The new tool provides something that hasn't existed before: a window into what's actually happening in conversational AI advertising.

The timing matters. According to Similarweb's data, 26% of ChatGPT responses already include sponsored ads for users on Free and Go tiers. Nearly 30% of ad-eligible queries in Google AI Mode show ads as well. More than 40% of Google searches now trigger an AI Overview, and those overviews often carry ads. These aren't test placements or limited experiments anymore. They're part of how millions of people encounter information every day.

Most major advertising platforms have built transparency tools over the years. Meta offers its Ad Library. Google maintains an Ads Transparency Center. TikTok has its Creative Center. These tools allow advertisers to see what competitors are running, when campaigns launched and how creative strategies evolve. ChatGPT and Google's AI features offer none of that.

Harel Amir, general manager and head of product for Similarweb Ad Intelligence, described the situation as a rare opportunity. "This is the rare moment when a major ad channel is still wide open," Amir said. "The advertisers who can see what's happening with AI ads now will have a real head start. Until today, nobody could see it at all."

That gap has created problems for advertisers at multiple levels. Teams managing Google's broader digital advertising programs often don't know whether their own ads are running in AI Mode or AI Overview contexts. Google's existing visibility tools don't answer that question. For ChatGPT, ads only appear when the platform's algorithms detect intent from the conversation, and placement decisions come down to the combination of ad creative, copy and landing page quality. Without visibility, advertisers have been left guessing.

Jonathan Bar Vardi, head of strategy at Natural Intelligence, a marketing firm that has become a major buyer of ChatGPT ads, described the challenge. "When we started advertising on ChatGPT, we were flying blind, no visibility into who else was in the auction or what was working," Bar Vardi said. "Similarweb changed that. It shows what other advertisers don't know: who's spending, where they're appearing and what's driving performance."

Advertising in conversational AI platforms operates differently than traditional search or social media. In search, advertisers bid on keywords and their ads appear based on query matching. In social media, targeting relies on user demographics, interests and behaviors. In AI platforms, the context of the entire conversation determines ad placement.

ChatGPT ads are selected based primarily on the conversation's context. When someone enters a prompt, OpenAI's ad system analyzes the topic, runs brand safety checks, identifies relevant ads based on the landing page, headline and copy, then uses a relevance model to determine the best match. Advertisers can provide "context hints" describing topics or keywords where their product might be useful, but these don't function like exact-match search keywords. OpenAI controls final ad delivery based on the full conversation's context.

Google AI Mode and AI Overviews follow similar principles. Ads appear within AI-generated responses when the system determines they're contextually relevant. But advertisers participating in Google's broader advertising programs often can't see whether their ads are appearing in these AI contexts or how they perform differently than traditional search placements.

The format itself is different too. ChatGPT ads appear below the end of a response, clearly labeled as sponsored and visually separated from the answer. Each ad includes the brand name, logo, headline (up to 30 characters), description (up to 60 characters) and a square image. The limited space means messaging needs to be more concise than display ads or even search ads typically require.

Similarweb's AI Ads dataset draws insights from real user panel conversations rather than synthetic prompts. That distinction matters because it reflects what actual users see rather than what advertisers might expect based on controlled testing. The company's approach provides an Ads Gallery showing what ads are running and in what context.

Future releases of the dataset are expected to add advertisers' true share of voice, ad categories and conversational intent analysis. Those additions would give brands the ability to understand not just what ads are appearing, but how dominant specific advertisers are in particular conversation types and what kinds of user questions trigger ads most reliably.

The data already shows some patterns. Travel-related questions seem to trigger ads particularly often. Queries about specific products or services frequently surface ads from competitors rather than the named brand, a tactic known as "poaching" that's been common in search advertising for years. Questions involving decision-making or comparisons tend to generate ad placements more consistently than purely informational queries.

OpenAI introduced ads in ChatGPT in February 2026, starting with a test in the United States. The rollout has since expanded to Canada, Australia, New Zealand and the United Kingdom, with future launches planned for Mexico, Brazil, Japan and South Korea. Ads appear for logged-in users on Free and Go plans but not for Plus, Pro, Business, Enterprise or Education accounts. OpenAI also doesn't show ads to users who say they're under 18 or who the company predicts are under 18.

ChatGPT has more than 900 million weekly active users worldwide. A third of the US population is expected to use it this year. Those numbers represent a substantial shift in how people find information. According to McKinsey, half of consumers use AI-powered search to research and discover products and services, with 44% saying it's their primary method for deciding what to buy.

That behavioral change has implications beyond advertising mechanics. Consumers used to scan through multiple search results, review sites and online forums to make decisions. Now they're increasingly using AI tools to ask questions, research products, compare options and get recommendations within a single conversation. Brands that aren't visible in those conversations risk being left out of consideration entirely.

The introduction of ads in conversational AI platforms has raised questions about user trust and platform transparency. OpenAI has stated that ads don't influence ChatGPT's answers and that the ad system is separate from the chat model. Advertisers can't shape, rank or alter responses. Conversations remain private from advertisers, who only receive aggregate performance data like views or clicks.

But the experience of seeing ads within what feels like a private conversation creates what some researchers describe as "an aura of surveillance." One journalist who asked ChatGPT 500 questions to test the ad experience reported feeling hyperaware of the personal data being shared with the bot, even knowing advertisers couldn't see the conversations. The ads made interactions feel less private.

OpenAI has attempted to address these concerns through user controls. People can dismiss ads, share feedback, learn why they're seeing particular ads, delete ad data and manage ad personalization settings. The company claims it's seeing no impact on consumer trust metrics, low dismissal rates of ads and ongoing improvements in relevance based on feedback.

Research on AI-generated ads adds another layer to the conversation. A study from Syracuse University's Newhouse School found that AI-generated ads closely resemble human-made work but don't perform as well in driving short-term sales. Human-created ads outperformed AI ads by an average of 11 points, particularly in conveying storytelling and emotional resonance. Consumers often can't tell the difference between the two types, but the effectiveness gap remains significant.

Advertisers can access ChatGPT ads either through OpenAI's self-serve Ads Manager Beta or through technology partners like StackAdapt. Partners allow brands to manage AI ads alongside other programmatic channels through familiar tools and processes.

The ad format requires different creative approaches than other channels. With only 30 characters for headlines and 60 characters for body copy, messaging needs to be exceptionally concise. Visuals use a 1:1 aspect ratio at 256x256 pixels. The compact format means creative needs to be clear, specific and closely aligned with the conversation context where it might appear.

Best practices are still emerging, but early guidance suggests focusing on user needs rather than keywords alone, submitting multiple headline and copy variations to improve matching, and aligning creative closely with the topics where ads will appear. Context hints should describe questions or situations someone might ask rather than listing broad standalone keywords.

Measurement currently includes core metrics like impressions, clicks, click-through rate, average CPC, average CPM, spend and conversions. Advertisers can add UTM parameters to track traffic in existing analytics tools. OpenAI supports pixel-based measurement and Conversions API for tracking post-click actions.

ChatGPT's main competitors, Google's Gemini and Anthropic's Claude, don't currently feature sponsored ad buttons in outputs, though Google has said it's not ruling them out. That creates a temporary competitive advantage for platforms willing to go without ad revenue, but also raises questions about long-term sustainability.

OpenAI has been explicit about why it's introducing ads. CEO Sam Altman previously expressed distaste for ads in chatbots, calling the mixture of ads and AI "uniquely unsettling" during a 2024 discussion at Harvard Business School. He described ads as "a last resort" for the business model but said he'd do it if that was the only way to provide universal access to great services.

The company's position has clearly shifted. OpenAI discontinued Sora, its social media app for AI videos, and scrapped plans for an erotic version of ChatGPT. Leaders are attempting to streamline operations and focus on core products. Ads represent a way to monetize the massive user base while keeping free and low-cost tiers available.

The billions currently spent on search ads are likely to shift toward AI-powered search and conversational platforms. Olivier Toubia, a marketing professor at Columbia Business School who focuses on AI, described it as "a huge multibillion-dollar market that is emerging." How that market develops will depend partly on whether advertisers can get the visibility they need to optimize campaigns effectively.

Similarweb's AI Ads dataset provides that visibility at a moment when the channel is still relatively open. Advertisers who can see what's happening now, who's bidding on what conversations and what creative approaches are working have an opportunity to establish positions before the space becomes as competitive and expensive as traditional search advertising. Whether that advantage translates into long-term success will depend on how quickly the market matures and how platforms balance advertiser needs with user trust.

AI Ads fill a transparency gap
Prior to Similarweb's launch, advertisers could not see how ads appeared or performed in conversational AI platforms.

Conversational AI advertising is context-driven
Ads are displayed based on user conversation context and intent, not just keywords.

User privacy and ad relevance are prioritized
Ads do not influence AI responses and are limited for certain user groups.

Human creativity remains crucial
AI-generated ads are less emotionally effective than human-created ads.

Early insights provide competitive advantage
Advertisers using Similarweb AI Ads gain valuable visibility into an emerging ad ecosystem.

Advertising formats and measurement are evolving
AI platforms require new strategies for effective ad design and performance tracking.