The CPC Bubble

In July 2026 we pulled the Google Ads auction data for the keyword families that define the marketing-intelligence market: AI visibility, competitive intelligence, brand monitoring, market intelligence, and marketing audits. The numbers describe a market that has quietly repriced itself. A click on "llm visibility tool" now averages $82. A click on "brand monitoring software" averages $202 — with top-of-page bids reaching $143. These are business-software auction prices that would have been unremarkable for insurance or legal keywords a decade ago, attached to a category that barely existed three years ago.

The same dataset shows something stranger: directly adjacent to these auctions, high-intent buyer queries carry no advertisers at all. The bubble and the vacuum coexist, often one keyword apart. This paper publishes the measured prices, explains who is setting them and why, and describes the arbitrage the pricing leaves open.

What the Auction Says

The measured averages, US market, July 2026 — each figure is Google Ads auction data (average cost-per-click, with the top-of-page bid range where relevant): brand monitoring software $202 (90 searches/mo). Brand monitoring tool $113 (210/mo). Competitor monitoring tool $82 (260/mo). LLM visibility tool $82 (720/mo). Share of voice tool $72 (70/mo). Market intelligence tools $62 (320/mo). AI search optimization $61 (1,300/mo). AI SEO tool $56 (2,400/mo). Marketing analytics tools $55 (880/mo). AI visibility tool $49 (1,600/mo). PPC competitor analysis $48 (170/mo). Market intelligence $41 (1,300/mo). AEO tool $37 (720/mo). Competitive intelligence software $37 (320/mo). Generative engine optimization $29 (4,400/mo).

Fifteen commercial terms in one category family averaging above $35 per click. For calibration: the broad head term "competitor analysis" — 8,100 monthly searches, the largest query in the set — averages $19, and "marketing audit service," a directly transactional query, averages $6. The premium sits precisely on the tool-shaped, category-defining terms: the words a buyer types when they have decided a product should exist and want to know who makes it.

Who Sets a $202 Click

Auction prices are not opinions; they are the market-clearing price of a buyer. A rational bidder pays $202 for a click only when the economics behind it support that price — enterprise contracts, five-figure annual deals, sales-led funnels where one closed account repays thousands of clicks. The categories in this dataset are dominated by exactly that economics: enterprise brand-monitoring suites, demo-gated competitive-intelligence platforms, and a new generation of venture-funded AI-visibility vendors converting the AEO wave into enterprise pipeline.

This is worth stating precisely, because it is the most useful market signal in the dataset: the auction price of a category tells you what the incumbents believe a customer is worth. When a category sustains $50–200 clicks on triple-digit monthly volumes, the sellers in it are not competing for self-serve credit cards. They are competing for contracts — and every self-serve or mid-market buyer typing the same words is priced out of their own search results, served ads for products they will never buy.

The Vacuum Next Door

The second finding is the inverse of the first. In the same measurement pass we ran live search-results probes on longer, more specific buyer queries in the same families — the comparison questions, the "best X for Y" phrasings, the category-recommendation searches that sit one step closer to a decision. On query after query, the paid shelf was empty. No advertisers. Not cheaper clicks — no clicks being bought at all.

The mechanics are unglamorous: enterprise bidders concentrate budgets on the head terms their agencies can report on, and the long tail — individually small, collectively substantial — goes unbought because no one owns a spreadsheet row for it. The result is a market where the head is bid to $200 and the tail is priced near Google’s auction floor. For any vendor whose economics do not require enterprise contracts, the tail is where the buyers are and the head is where the money burns.

What the Prices Mean for the Buyers Doing the Searching

A buyer searching these terms should understand what the results page they see actually is: a list of companies whose unit economics can absorb a $50–200 introduction fee. That filters for a specific kind of vendor — enterprise pricing, sales-led motion, procurement-shaped contracts. The products best suited to a mid-market team are systematically underrepresented in the paid results for the category’s own name, not because they are worse, but because their price points cannot fund the auction.

The practical consequence: in categories with this auction structure, the organic results, the answer engines, and the community threads carry more information about the full market than the ads do. The paid shelf shows you who has raised the most money for distribution. It does not show you the market.

The Arbitrage

For operators, the dataset describes an arbitrage with three legs. First: the demand that costs $49–202 per click to buy is, in most of these categories, still winnable organically — the SERPs for the AI-visibility family in particular are young, thin, and moving, because the category is newer than the content that should own it. A page that ranks captures the same buyer the $82 click buys, at the marginal cost of having written it.

Second: an increasing share of these buyers never reach the results page at all. They ask an answer engine, receive a synthesized shortlist of three to five names, and search for one of those names directly. Presence in AI answers is not purchasable at any CPC — it is earned through third-party corroboration: reviews, comparison pages, community presence, the sources the engines actually read. The most expensive auction in this dataset, brand monitoring, is literally a market for watching a conversation that the answer engines now mediate for free.

Third: the auction data itself is targeting intelligence. A rival paying to appear on a query is publishing, at their own expense, a verified statement that the query converts. The queries a category’s incumbents bid hardest on constitute a map of where the money is — a map that is free to read and expensive to ignore.

Method and Limitations

Source: Google Ads auction data (average CPC, top-of-page bid ranges, and average monthly search volumes) retrieved July 29, 2026, for the United States, English. Average CPC is the historical average paid across advertisers and positions; top-of-page bids describe the range advertisers pay for prominent placement. Search volumes are trailing twelve-month averages. The empty-paid-shelf observations come from live search-results captures on specific buyer queries in the same families, same date.

Limitations, stated plainly: auction prices move, and these are a snapshot, not a constant. Averages compress a wide range — a well-run account with strong relevance scores pays below these figures; a new account pays above them. Volume figures are averages of a spiky reality. And one measurement pass is one measurement pass: we re-run this dataset quarterly, and the movement between editions — which terms inflate, which deflate, where new auctions form — is the part worth watching. Every figure in this paper is a real retrieved value; nothing is modeled or extrapolated.