Conversational AI Advertising by the Numbers: Every Public Data Point from H1 2026
The first half of 2026 produced more public performance data on conversational and agentic AI advertising than the previous three years combined. Ad spend forecasts, chatbot CPM pricing, click-through benchmarks, and the first audited agentic campaign results are now all on the record. This report aggregates every significant public data point from the past six months in one place, with sources and dates, so the numbers can be compared side by side.
A note on method: every figure below is attributed to its original source and publication window. Where sources conflict, both numbers appear. Adgentek's own figures are limited to the metrics we have published previously.
How big is the AI advertising market in 2026?
US AI ad spending will reach $32.03 billion in 2026, nearly tripling year over year, and will grow to $68.25 billion by 2030, according to eMarketer's June 2026 US AI Advertising Forecast. The structural finding inside that number matters more than the headline: more than 80% of 2026 AI ad spend runs through traditional search listings appearing alongside AI-generated results, such as ads next to Google AI Overviews, rather than inside chatbot conversations. eMarketer sizes ads inside standalone AI chatbots at under $1 billion for 2026, roughly 3% of total AI ad spending, while OpenAI itself projects $2.5 billion in 2026 ad revenue. The gap between those two numbers is the central forecasting dispute of the year: OpenAI has told investors ChatGPT will generate $100 billion in global ad revenue by 2030, while eMarketer puts the entire US chatbot ad market at just over $5 billion that year.
What do ads inside AI chatbots cost?
Chatbot ad pricing compressed roughly tenfold in ten weeks. ChatGPT ads launched on February 9, 2026 at approximately a $60 CPM with a $200,000 minimum commitment. By the May 5 self-serve launch of OpenAI's Ads Manager, the minimum was gone, observed CPMs had fallen to around $25, and cost-per-click bidding went live with $3 to $5 bid floors, rising to a reported $8 to $15 per click in B2B categories. Trade reporting in June placed the working CPM range at $25 to $45, positioning the channel above social CPMs and in line with competitive search. eMarketer projects chatbot CPMs compress further to roughly $15 by 2030. For comparison across the independent AI ad networks, publicly listed pricing sits near a $3 CPM at the low end and around a $10 eCPM for native text placements.
How fast is chatbot advertising scaling?
ChatGPT's ad platform went from invite-only pilot to open US self-serve in 86 days, and the adoption data followed. OpenAI Ads crossed $100 million in annualized revenue within six weeks of self-serve opening. Third-party tracking from AdClarity by BIScience measured roughly $109 million in total ChatGPT ad spend in April 2026 alone, with a projection of $500 million as advertisers move off test budgets. Independent rollout tracking on May 26, 2026 measured sponsored placements in 26.5% of ChatGPT responses globally and 49% in the US, against a user base reported at roughly 900 million weekly active users. OpenAI states that fewer than 20% of eligible users see an ad on any given day. Measurement infrastructure arrived with the May 5 release: a browser-side conversion pixel and a server-side Conversions API, plus integrations with commerce media platforms.
How do conversational ad formats perform?
The performance data that exists points one direction: conversational placements outperform display, and interactivity multiplies the effect. The published benchmarks from the past six months:
| Metric | Figure | Source and date |
|---|---|---|
| Average CTR, native ads in AI apps | 3% at scale, framed as 4x display | Koah published platform data, June 2026 |
| CTR range, GenAI app placements | 3-5% across advertiser categories | Kontext published platform data, 2026 |
| CPA/CPI vs. traditional channels | 2-4x better | Kontext published platform data, 2026 |
| LLM referral conversion rate | 1.5x other digital channels | Criteo, reported May 2026 |
| ChatGPT referral time on site | 60-80% longer than social traffic | Trade reporting, May 2026 |
| Conversational inventory ROAS | 70% above target | PubMatic pilot data |
| Click-to-convert, conversational vs. traditional | 1.5x stronger | PubMatic pilot data |
| Interactive format engagement vs. display | 3-8x higher | Adgentek, Spark format |
| Interactions per session, interactive format | 3.2 average | Adgentek, Spark format |
The pattern across sources is consistent. Static text placements in AI apps already clear 3% CTR, several multiples above display norms, because the ad arrives inside a declared-intent moment. Interactive formats compound that: when users can ask questions inside the unit, as they do with Spark, engagement runs 3 to 8x above display and each session averages 3.2 interactions, with every interaction generating structured intent data. The mechanics behind that difference are covered in how conversational ads work.
What results are agentic campaigns producing?
Agent-executed media buying moved from proof-of-concept to repeat business in H1 2026, and the published results are specific. The first fully autonomous end-to-end campaign, run by Butler/Till for Geloso Beverage Group on PubMatic's AgenticOS, delivered 82% lower buy-side supply chain costs (characterized as a 5.5x reduction), 40% more impressions than planned, 30% lower CPMs, a 98% video completion rate, under 1% inventory failing DoubleVerify standards, and zero made-for-advertising inventory. Across early AgenticOS campaigns broadly, PubMatic reports 40-50% supply chain cost reductions, 87% faster campaign setup, and 70% faster troubleshooting. In a June 2026 head-to-head, Level Agency ran AgenticOS against its incumbent DSP with identical budgets and reported more than 2x reach per dollar on qualified audiences. In France, Amnet's agentic campaign for Interbev cut setup time by 80%. IAB research from the same period found two-thirds of digital video buyers are using or planning agentic AI campaigns in 2026. The transaction model behind these results is defined in our glossary entry on agent-to-agent media buying.
What do consumers think about ads in AI?
Consumer sentiment is the most cautionary dataset of the six months. In a January 2026 Ipsos survey, 63% of US adults said ads in AI search results would make them trust the output less, and 57% said ads in AI chatbots would reduce trust in the advertised brand, per a Partnercentric survey reported alongside it. OpenAI's own pilot readout counters that ads have not measurably dented consumer trust in ChatGPT and that users dismiss ads at low rates. Both can be true: the surveys measure anticipated attitudes, the pilot measures observed behavior. The design implication is the same either way. Formats that add utility to the conversation, and privacy architectures that keep raw conversations away from advertisers, are what separate sustainable monetization from trust erosion. That is the standard we hold across AI surfaces running Adgentek inventory.
What the H1 2026 data adds up to
Three conclusions hold across every independent source. First, the money is real but concentrated: $32 billion flows through AI-adjacent advertising this year, while the conversational core remains under $1 billion and is growing into its forecasts. Second, static chatbot placements are commoditizing fast, with CPMs compressing from $60 toward $15 while engagement-priced interactive formats hold their value, because interaction produces intent data that impressions cannot. Third, the supply chain economics of agentic transactions are no longer hypothetical: 40-80% cost reductions are showing up in audited campaigns. The channel is being built in public, quarter by quarter, and the operators reading the data early are the ones positioned for it. For the landscape view, start with how to advertise in AI chatbots, and definitions for every term in this report are in the AI advertising glossary.
This report will be updated as new public data is released. Sources: eMarketer, OpenAI, AdClarity by BIScience, PubMatic, Butler/Till, Level Agency, Criteo, Koah, Kontext, Ipsos, Partnercentric, IAB, and contemporaneous trade reporting, January through July 2026.
Put this data to work
The 40-80% cost reductions in this report did not come from better dashboards. They came from agent-executed buys. If you run media for brands or performance clients and want to test outcome-declared buying on the highest-intent channel in market, Adgentek ORCA is in pilot with select agency partners now: declare the CPA or ROAS target, and agents handle everything between the declaration and the verified result. See how advertisers work with Adgentek, or reach us directly at hello@adgentek.ai.
