Adobe closes $1.9bn Semrush deal as Pew data shows AI summaries cut clicks

The one measured number in the record was collected on Google's results page. Pew Research Center's percentages sit either side of a summary: a standard result was clicked in 8 percent of visits when an AI summary appeared and 15 percent when none did, a fall of 46.7 percent of the clicks. Another 1 percent clicked a link inside the summary itself ^[7][8][3]. So no more than 9 percent of summary visits produced a click of either kind, against 15 ^[4]. Sessions ended on the results page 26 percent of the time with a summary and 16 percent without, 62.5 percent more often ^[9][7].

Repeated sampling is the tractable half of the problem. Output is stochastic, so the same prompt returns a different set of brands in a different order, and a vendor with enough capacity can report the whole distribution. techfundingnews counts that among four properties which make a single reading close to worthless ^[10]. There is no data for the other half. When ChatGPT recommends a brand without linking to it there is no referrer and no session, so the company recommended never learns it happened, and neither does the one left out ^[11]. A customer cannot reconcile a score built on those events against its own analytics.

Opting a site out of OAI-SearchBot keeps it out of ChatGPT's search answers, per OpenAI ^[13]. On the commerce side, OpenAI's help documentation says product results are selected independently, are not ads, and are not influenced by OpenAI partnerships, with ads running as a separate system ^[14]. Merchants apply to a product feed programme. OpenAI notes that a product appears when ChatGPT judges it relevant to the user's intent, that the price shown may not be the lowest available, and that merchant updates can take time to appear ^[16].

Semrush closed at $6.76 the day before Adobe's November 2025 announcement, and Adobe offered $12.00 a share in cash, 77.5 percent more, or $5.24 a share ^[4][3][1][2]. Control sits inside that premium alongside Semrush's existing business, and Adobe described the asset as a brand visibility platform ^[5].

In my view the defensible asset is the collection apparatus: panels across regions, languages and account states, sampled often enough to produce a distribution. Profound's $96 million is about 62 percent of the more than $155 million it has raised in under eighteen months ^[1][2][6]. That is a lot of money for infrastructure and not much for a dashboard. The counter-case is that buyers never wanted an auditable level, only a direction of travel and a list of pages to fix. If that is right, the sampling noise is irrelevant and the panel is a commodity.

What would make the level checkable is reproducibility data from the vendors, or attribution data from OpenAI. The second has moved once already. Instant Checkout launched on 29 September 2025, and in early March 2026 OpenAI refocused on discovery, with purchases completing on the merchant's own site or through dedicated apps ^[17][18]. techfundingnews attributes the retreat to conversion ^[19]. Where the purchase completes on the merchant's site, the merchant's own funnel records it.