Key takeaways
- Ads now show up in roughly 20-32% of ChatGPT web-search responses (90-day average vs. the past 7 days) and in about 29% of Google AI Mode commercial queries, per SE Ranking's August 2026 study.
- Standard PPC and competitive-intel tools (SpyFu, Semrush ads reports, Adthena) can't see these placements because they're conversational objects, not crawlable SERP elements. You need purpose-built scrapers or APIs.
- Winning an ad slot and earning an organic citation are two separate games. SE Ranking found the advertiser's domain shows up organically in only 3.63% of ChatGPT ad placements, and AI Mode cites the advertiser in just 12% of cases.
- Your account type changes what you see. Paid ChatGPT tiers (Plus, Pro, Business, Enterprise, Edu) show zero ads. If you're testing from a work seat, you'll wrongly conclude nobody advertises in your category.
- For brands that want to go beyond ad-spotting and actually fix their AI visibility problem, a full GEO platform like Promptwatch is worth a look alongside the dedicated ad trackers below.
Why tracking ads in AI search is suddenly a thing
A year ago this guide wouldn't have existed. OpenAI announced plans to test ads in January 2026, the US pilot launched in February, and by June the company was calling itself "an advertising business" at Cannes Lions, projecting $2.5 billion in 2026 ad revenue. Google moved in parallel, rolling shopping ads with "Direct Offers" into AI Mode in February.
The speed of the rollout is the part that catches people off guard. Promptwatch's own tracking of ChatGPT ad frequency shows the feature went from zero on May 26 to appearing in 1.69% of responses the next day, then climbing in bursts past 30% by July 1. The most recent 7-day average sits around 32.4%, up more than 12% from the week before. You cannot check this once and call it done. The rate moves week to week, sometimes day to day.
Google's side of the story is messier, honestly. SE Ranking's 50,032-keyword study found AI Mode serves a text ad on about 29.45% of commercial queries, but called that figure a floor, not a ceiling, because the same prompt can return an ad in one session and nothing in the next. Other vendors disagree wildly on how often AI Overviews even trigger ads at all, with estimates ranging from 15% to near 50% depending on methodology. If a vendor hands you a single clean percentage without a date range attached, be suspicious of it.
What makes AI search ads hard to track
There's no Auction Insights page for ChatGPT. Google's own Ads Transparency Center doesn't cover AI Mode's conversational ad units either. So the entire category of tools here works by repeatedly sampling fixed prompts, rendering the actual chat or SERP, and parsing the structured ad object that comes back. A few things trip people up constantly:
- Account type matters. Free and Go-tier ChatGPT accounts see ads. Paid Plus, Pro, Business, Enterprise and Edu accounts never do. Teams running their monitoring from a company Plus seat will see nothing and wrongly conclude the category is ad-free.
- Single snapshots lie. Both Promptwatch and SE Ranking explicitly warn against drawing conclusions from one run. You need days, ideally weeks, of repeated sampling at a fixed country and session type before reporting a share number.
- CPC and niche swing results hugely. SE Ranking found ad presence in AI Mode jumps from 24.33% on sub-$2 CPC keywords to 53.56% on $10+ keywords, and from 2.64% in healthcare to 72.38% in pets. A single "average ad rate" for your industry is close to meaningless without knowing which niche and CPC band you're in.
- Paying for an ad doesn't buy you a citation. In roughly 88% of AI Mode ad cases studied by SE Ranking, the advertiser's domain wasn't among the cited sources at all. These are two separate budgets, not one lever.
The tools that actually catch inline AI ads
Dedicated ad trackers
These are built specifically to find and log the "Sponsored" card that shows up under ChatGPT answers or the ad units inside Google AI Mode.
SE Ranking's AI Search ad tracker lives inside its Competitive Research module, no separate subscription required. It shows the advertiser's identity, destination link, and a month-over-month presence trend across the whole market, not just your own campaigns.
AIRIX started as a free, no-signup scanner (three free scans, roughly 30 seconds each) that captures the advertiser, title, snippet, image and landing page for a given brand or industry prompt. Its paid Growth tier at $39.99/month adds continuous monitoring, email alerts when a new advertiser enters your category, creative history, and visibility tracking across 16 AI platforms running 32 live queries in under a minute. It's one of the few tools that explicitly states what it's not: a replacement for your SEO suite's ad spy tool, because those tools simply can't see conversational ad objects.
cloro takes a developer-first approach. It's an API that renders the live Google SERP and returns structured JSON, including an aioverview.ads field with sponsored shopping ads, price, and store domain, alongside citation and source data. The same API key covers ChatGPT, Perplexity, Gemini, AI Mode and Copilot. Free tier gives you 500 credits a month; the Lite tier at $30/month bumps that to 37,500 credits. If your team can write code against a JSON response, this is the cheapest way to build your own tracking dashboard.
Profound's Ads Studio, launched July 7, 2026, is the most ambitious entry so far. Built on Profound's dataset of 1.9 billion+ real-user prompts, it introduces two metrics worth stealing even if you don't buy the tool: Paid Share of Voice (how often your ad appears versus competitors in AI conversations with sponsored results) and Ads Relevance (whether the ad actually fits the conversation it's showing up in). It also pulls impression, CTR, CPC and CPM data through "OpenAI Ads nodes" for ongoing benchmarking.
Full visibility platforms with an ads layer
If tracking who's advertising is only half your problem and the other half is "why isn't my brand showing up at all," a standalone ad tracker won't get you there. This is where broader AI visibility platforms earn their keep.
Promptwatch's own data on this exact question, the ChatGPT Ads Over Time report, tracks the daily ad rate across a 90-day window and breaks down ad-bearing responses by prompt type: 73.3% land on organic/informational prompts, 18% on brand-specific prompts, and 8.6% on competitor-comparison prompts. That last bucket matters if you're worried about competitors buying ads against your brand name. Promptwatch also tracks ChatGPT Shopping Feature Usage, the product-card and price-comparison surfaces that sit next to text ads and concentrate heavily on transactional prompts.
Unlike pure ad spotters, Promptwatch connects what it finds to action: content gap analysis, automated content generation, and Unified Actions that prioritize what to fix. If an ad is winning a slot where your organic citation should be, that's a content and crawler-log problem, not just an ad-budget problem, and the platform treats it as both at once.
Other names in this space worth knowing: Evertune positions itself around combining GEO monitoring with AI advertising in one dashboard, and Scrunch AI and Profound AI both track monitoring at enterprise scale, though neither matches Promptwatch's crawler-log depth or Reddit/YouTube citation tracking.
Comparison table
| Tool | What it tracks | Starting price | Best for |
|---|---|---|---|
| SE Ranking AI Search | ChatGPT ad presence, advertiser, trend over time | Bundled into existing plans | Teams already on SE Ranking |
| AIRIX | ChatGPT ads + 16-platform visibility | Free (3 scans) / $39.99/mo | Quick competitive checks |
| cloro | AI Overview, AI Mode, ChatGPT ads via API | Free (500 credits) / $30/mo | Developers building custom dashboards |
| Profound Ads Studio | Paid Share of Voice, Ads Relevance, CPC/CPM | Enterprise pricing | Large advertisers benchmarking spend |
| Promptwatch | Ad rate trends, shopping features, plus full GEO stack | $95/mo | Brands that want tracking and a fix, not just a number |
How to actually run this yourself
If you don't want to buy a tool yet, you can get a rough read manually, but do it properly:
- Pick 15-20 prompts your actual buyers would type, split evenly between informational, brand-name, and competitor-comparison phrasing.
- Run each prompt from a logged-out or free-tier ChatGPT account and from a fresh AI Mode session, same country setting each time.
- Log every result, including the empty ones. Your denominator is every run, not just the runs where an ad appeared.
- Repeat daily for at least two weeks before you report a percentage to anyone. A single day's run can land on an outlier, as both Promptwatch and SE Ranking note in their own methodology write-ups.
- Note which advertiser landing pages are winning the slot and whether that domain also shows up as a citation. If it doesn't, that's two separate opportunities, not one.
That's basically what the dedicated tools automate for you, minus the tedium of refreshing a chat window forty times a day.
Where this is headed
OpenAI named its first third-party measurement partners (Fospha, Measured, INCRMNTAL for full-funnel attribution; Haus and WorkMagic for incrementality; DoubleVerify and IAS running brand suitability pilots) on October 5, 2026. That's a strong signal this isn't a side experiment. By the time this becomes a mature ad channel with its own auction-insights equivalent, the brands that were already tracking placement and citation overlap will have a real head start on everyone else.
If your bigger concern is less "who's buying ads against me" and more "why does ChatGPT never mention my brand at all," that's a different and frankly bigger problem, and one worth solving with crawler logs and content gap analysis rather than an ad tracker. The GEO software directory at bestgeosoftware.com is a reasonable place to compare the broader category if you're shopping around.