In this week’s digital news, Similarweb data shows 95% of ChatGPT’s outbound clicks land on just a handful of sites, reshaping how publishers should value AI referral traffic. Google Ads has added two new product reporting metrics to some reports: product clicks and non-product clicks.
GSC’s AI Overviews impressions show no clicks, rank every AIO link at position one, and distort average position, masking real traffic and revenue declines. YouTube has just launched a UK version of its “Shopping Affiliate Programme.
A comprehensive Similarweb report reveals that AI chatbots are not cannibalising traditional search, but rather stacking on top of it. SparkToro CEO Rand Fishkin noted that the data “probably infuriate[s]” both AI zealots and sceptics, highlighting that 461 million of ChatGPT’s 494 million users (95%) actively used Google during the same window. Globally, traditional search still pulls 3.3 billion average monthly unique visitors compared to 655 million for AI chatbots, making search roughly five times the size of the entire chatbot category. Furthermore, the report shows that only 6.8% of ChatGPT answers in the U.S. include an external link, while SEO consultant Aleyda Solis pointed out that 65% of URLs cited by ChatGPT sit deep within site architecture, yet 58.8% of the resulting referral traffic actually lands on homepages.
Because cited pages and clicked landing pages are largely separate populations, SEO teams can no longer rely on simplistic correlation metrics between AI citations and down-funnel conversions. Brands must split their reporting frameworks, recognising that while search remains massively dominant, capturing traffic from generative platforms requires distinct attribution models that account for homepage navigation and platform-specific audience behaviours.
Google Ads has introduced two new product reporting metrics: product clicks, where a user clicked directly on a specific product within an ad, and non-product clicks, where a user clicked another element such as a headline, store name or CTA. The split allows advertisers to better understand whether individual products are driving direct purchase intent or simply contributing to broader brand engagement. The metrics have been spotted appearing in some reports, though no formal announcement from Google has accompanied the rollout.
Eight weeks after the rollout of Google’s “Generative AI” performance reporting in Search Console, industry experts are warning that the dataset functions as a misleading trap for corporate budgeting. Dan Taylor, Head of Innovation and Consultancy at 10 Spurs, highlighted that the dashboard displays impression metrics without click numbers or query details because generative summaries satisfy user intent directly on the SERP. Due to Google’s “single position rule,” every URL inside an AI Overview (AIO) block receives a distorted rank-one metric regardless of whether it is a primary headline citation or buried in an expandable accordion menu, which corrupts central tendency averages and masks real revenue declines.
For search visibility and enterprise reporting, this means that relying on raw GSC impression growth can create a dangerous illusion of success while underlying web traffic collapses. Because an impression without a click delivers zero economic value, marketing teams must urgently decouple their KPI targets from average position metrics and high-level impression counts. Organisations are advised to pivot toward first-party analytics, median position tracking, and actual lead conversions to accurately gauge the true health of their digital visibility.
YouTube’s Shopping Affiliate Programme launched in the UK on 26 July, allowing creators with as few as 500 subscribers to tag products directly within videos using clickable stickers and timestamps. Commission rates range from 5% to 20% depending on the retailer, with creators also earning on wider basket purchases made after a click-through. YouTube says in-video product tagging drives up to 50% more clicks than description links. Retailers can join via supported affiliate networks including Awin alongside an active Google Merchant Centre feed. Early UK brand participants include M&S and Boots, with the relatively small retailer pool currently favouring early movers.
AI search visibility depends heavily on whether a model already remembers a brand before it searches the web. A new report found that remembered brands were searched 3.2 times more often than brands outside the model’s memory, with search rates of 55.7% versus 17.4% across nearly 4,000 AI responses. Category authority built through press coverage, analyst mentions, partnerships, and consistent brand association appears to increase the odds that AI models will search for and recommend a brand. Strong live content still helps newer brands appear in AI searches, but building long-term brand recognition gives a more reliable path to visibility across future AI answers.
Read more here.
Katie Passarello, Co-Founder of PitchWell, explains how AI is changing public relations by making storytelling and media outreach more accessible. The discussion explores how businesses can improve visibility while maintaining the human experiences that make stories compelling. It also highlights why third-party credibility remains important as brands compete with the growing volume of AI-generated content.
Read more here.
If you’d like more information on any of the stories we’ve mentioned, or for support with your digital marketing, get in touch with our expert team by booking a call.
