What AI Agents See When They Look…

1. Key Themes

Theme: AI engines, not vendors, now control the price buyers see

The pricing page is no longer the source of truth

Even when AI engines cite a company's own pricing page, it is rarely the lead source.

"Although the results revealed that the companies' own pricing pages appeared in 46% of the answers, they appeared first in only 12%."

Because answers vary by assistant, two buyers researching the same product can leave with different numbers.

"This creates an unpredictable mechanism where customers researching the same thing may come to completely different conclusions."

The price depends on which assistant the buyer opens

Engines differ in how much they favor vendor-owned sources, and the buyer is never told where the figure came from.

"The price a buyer reads therefore depends largely on which assistant they happened to open. The buyer is never told whether the figure came from the vendor or from someone estimating on its behalf."

ChatGPT leans toward company pages while Google buries them.

"ChatGPT has the most influence and leads the company's own page by 38% and that's because OpenAI's model defaults to company-owned pricing pages, as opposed to Google that - according to the author - buries those."

Agent adoption makes this urgent

"research from G2 shows that more than 70% of businesses now rely on AI agents for software research."

Theme: Machine readability is a prerequisite for being cited

Many pricing pages are invisible to crawlers

A page that looks fine to humans can be an empty shell to a machine.

"Ultimately, a web page cannot be cited if the machine reading it encounters an empty shell, which happens to be the condition a meaningful share of the Cloud 100 is in."

Of the pricing pages studied, a minority fail the test.

"out of the 77 public pricing pages in the sample, 57 were fully readable to crawlers, while 10 hid at least 40% of their body content."

The causes are technical and mundane.

"information can load through JavaScript or hide behind restricted paths in robots.txt, rendering it invisible to raw HTML crawlers."

The fix is an engineering deployment, not a copy rewrite

"Profound solved a similar issue by moving the content to server-side rendering which helped the pricing page being more visible to raw HTML crawler and rose citation bot traffic rose 13% week over week."

Readability and transparency are separate levers

Publishing exact numbers is not what earns citations.

"Companies such as Personio, Pendo, Carta and Gong display plans without a single price point, even though all 4 sit above the median concluded by the study." "readability determines how many gaps the AI agent is compelled to fill, whereas transparency determines how much detail a company is willing to reveal."

Theme: Third-party sites fill the vacuum and vendors can't control them

Three external domains dominate pricing answers

"the same 3 external sites kept surfacing across the full dataset, and none of them belong to the companies being priced." "Vendr, which sells procurement and negotiation data, appeared in 18.7% of runs. Reddit appeared in 18.6% and G2 in 15.9%."

Vendors have uneven control over these sources

"While a company can edit its own G2 profile, it has zero control over anonymous Reddit threads or procurement logs, severely limiting vendor influence over how pricing is presented."

Absent pricing produces wild variance

"One vertical AI unicorn in the sample publishes no pricing whatsoever. Across 76 responses pulled mainly from Reddit, engines quoted a monthly range from $100 to $2,400, representing a 24x difference for the exact same product."

Data licensing deals amplify Reddit's weight

"The two biggest answer engines pay for structured access to Reddit threads, so a 2-year-old complaint about a renewal quote can carry more weight than the pricing page it complains about."

Theme: Documentation outperforms marketing pages in AI citations

Plaid's billing docs beat its pricing page

"Plaid had its own page cited first 42% of the time. Its billing documentation was cited in 70% of answers about its pricing, ahead of the pricing page itself at 64% and FAQs at 50%."

Why docs win

Docs are static and server-rendered, so every crawler can read them, and they are precise because implementers get blamed for billing errors.

"Developer docs are usually static, server-rendered HTML, which every crawler can read." "Docs are also written for someone who has to implement the billing and will get blamed when the invoice is wrong, so they define edge cases and units that a pricing page glosses over."

Theme: Agent-to-agent commerce is emerging, but checkout lags

Seller-side agent strength matters far more than buyer-side

"Swapping in a stronger buyer moved the final price by 2.6%. Swapping in a stronger seller moved it by 14.9%."

Buyer agents are unreliable negotiators

"Buyer agents also broke hard budget limits and gave away their own ceilings despite being told not to."

Agentic checkout hasn't taken off yet

"OpenAI shut down Instant Checkout in March 2026 after about 6 months, with Walmart reporting in-chat conversion 3x lower than click-out. Buyers researched inside the assistant and bought somewhere else."

2. Contrarian Perspectives

Publishing exact prices is not what earns AI citations The conventional advice is that price transparency wins in an AI-search world. The study complicates that: several above-median companies show plans with no price points. The variable that matters is how readable the page is, not how much it discloses.

"What complicates the obvious conclusion is that publishing exact numbers is not what earns the citation. Companies such as Personio, Pendo, Carta and Gong display plans without a single price point, even though all 4 sit above the median concluded by the study."

Your pricing page may be the least influential pricing asset you own Companies have spent decades optimizing the pricing page, yet developer documentation can out-cite it.

"Its billing documentation was cited in 70% of answers about its pricing, ahead of the pricing page itself at 64% and FAQs at 50%."

Seller-side AI matters far more than buyer-side AI in negotiation Most attention goes to empowering buyers with agents. The negotiation study suggests the advantage accrues to sellers, with a 14.9% price movement versus 2.6% for buyers.

"Swapping in a stronger buyer moved the final price by 2.6%. Swapping in a stronger seller moved it by 14.9%."

3. Companies Identified

Growth Unhinged — Newsletter and research outfit run by Kyle Poyar and Nikolas Laskaris. Mentioned as the source of the 7,675-response pricing study.

"Kyle Poyar and Nikolas Laskaris at Growth Unhinged ran a prompt against the Cloud 100... They've done this 7675 times."

Profound — AI visibility/citation tooling company and co-publisher of the study. Cited as a case study for fixing crawler visibility.

"Profound solved a similar issue by moving the content to server-side rendering... citation bot traffic rose 13% week over week."

Plaid — Fintech API company and the best performer in the sample. Its billing docs are the model for AI-readable pricing.

"Plaid had its own page cited first 42% of the time. Its billing documentation was cited in 70% of answers about its pricing."

Fireworks AI — AI inference platform. Praised for LLM-friendly documentation practices.

"Fireworks AI's docs expose an llms.txt index and a copy-for-LLM button. It also publishes scheduled price changes side by side with their dates, so an engine knows which number is current."

Vendr — Procurement and negotiation data provider, and the top third-party source in AI pricing answers.

"Vendr, which sells procurement negotiation data, appeared in 18.7% of runs."

Reddit — Community platform with data licensing deals with Google and OpenAI. Heavily weighted by engines for pricing answers.

"Reddit appeared in 18.6%... they did sign a data licensing deal with Google in early 2024, and OpenAI followed a few months later."

G2 — Software review marketplace. Source of the 70% adoption stat and a top-three cited domain.

"research from G2 shows that more than 70% of businesses now rely on AI agents for software research... G2 in 15.9%."

OpenAI / ChatGPT — Answer engine that most favors vendor-owned pricing pages. Also shut down Instant Checkout.

"ChatGPT has the most influence and leads the company's own page by 38%... OpenAI shut down Instant Checkout in March 2026 after about 6 months."

Google (Gemini / AI surfaces) — Search and answer engines that downweight vendor pricing pages and licensed Reddit data.

"as opposed to Google that - according to the author - buries those."

Walmart — Retailer cited for the failure of in-chat checkout.

"Walmart reporting in-chat conversion 3x lower than click-out."

Personio, Pendo, Carta, Gong — Software companies that show plans without prices yet sit above the median in the study.

"Companies such as Personio, Pendo, Carta and Gong display plans without a single price point, even though all 4 sit above the median concluded by the study."

Hard Skill Exchange (sponsor) — Hosting a free "Recursive GTM" session on Oct 15 with leaders from Dialpad, ZoomInfo, 1mind, Vivun and Aligned.

"Live demos and 30-minute practice sessions with leaders from Dialpad, ZoomInfo, 1mind, Vivun and Aligned."

Neil Patel (source of image) — Credited for the web crawler illustration; minor mention.

"Image source: Neil Patel"

4. People Identified

Kyle Poyar — Author at Growth Unhinged and a pricing/growth researcher. Co-led the 7,675-response study that anchors the article.

"Kyle Poyar and Nikolas Laskaris at Growth Unhinged ran a prompt against the Cloud 100."

Nikolas Laskaris — Co-author at Growth Unhinged. Co-led the same study.

"Kyle Poyar and Nikolas Laskaris at Growth Unhinged ran a prompt against the Cloud 100... They've done this 7675 times."

Ruben Dominguez — Author of the newsletter article; also wrote related pieces on agent spend and agent deployment.

"Your AI Agent Might Be Paying $11,000 a Month to Answer Yes or No — Ruben Dominguez"

5. Operating Insights

Audit "price variance" between your published price and what engines report Run the same prompt across several engines, log the answers, and compare them with your page. The article argues this is a metric nobody tracks.

"Companies measure citation share and sentiment, but not the gap between the price they publish and the price the engines report.... It takes one prompt, a handful of engines and a week of logging."

Make pricing content server-rendered and crawler-readable Check whether JavaScript, tabs, or robots.txt restrictions hide content from raw HTML crawlers. Profound's fix was a deployment, not a rewrite.

"information can load through JavaScript or hide behind restricted paths in robots.txt, rendering it invisible to raw HTML crawlers."

Treat documentation as a pricing surface, with LLM-friendly formatting and dated price changes Define units and edge cases in billing docs, expose an llms.txt index, and publish scheduled price changes with dates so engines know which number is current.

"Fireworks AI's docs expose an llms.txt index and a copy-for-LLM button. It also publishes scheduled price changes side by side with their dates, so an engine knows which number is current."

6. Overlooked Insights

Stale community posts can outweigh current vendor pricing Because answer engines pay for structured Reddit access, old threads can carry disproportionate weight in how your price is represented, creating a lingering reputational and pricing liability that is hard to remove.

"a 2-year-old complaint about a renewal quote can carry more weight than the pricing page it complains about."

Agentic checkout has already had a visible failure Despite hype around agent-led commerce, a major retailer's data showed buyers research in-assistant but purchase elsewhere, suggesting the research-to-purchase handoff remains a durable human or click-out step.

"OpenAI shut down Instant Checkout in March 2026 after about 6 months, with Walmart reporting in-chat conversion 3x lower than click-out. Buyers researched inside the assistant and bought somewhere else."