AI Search & GEO News

Parsnipp commentary

Kevin Indig and Amanda Johnson argue that traffic-based planning is breaking down, because zero-click searches undercount visibility and attribution can't reliably credit AI-driven influence. Their alternative is to set presence goals, such as visibility in the AI engines your customers use, and to plan surface by surface, since AI citations overlap little between engines. That makes it worth tracking your brand's visibility on each AI platform separately, rather than treating AI search as a single channel.

Parsnipp commentary

A Stella Rising survey of 511 beauty consumers found Google is the first stop for 43%, with the rest starting on Instagram, YouTube, TikTok, or AI tools. Many discover a product in one place, check it on Google, and buy somewhere else, such as Amazon. It's one category and a small sample, but the pattern is a useful reminder not to judge any single channel on conversions alone, since discovery and purchase often happen in different places.

Parsnipp commentary

OpenAI is testing labeled visual ads that appear alongside images users generate in ChatGPT, starting in the US later this month. It also named a long list of measurement partners, including attribution tools and geo-based incrementality testing, so advertisers can evaluate the channel in the reporting they already use. The early results are single-brand figures reported by the vendors, so treat them as hypotheses, and a holdout test is the best way to check whether the format works for your own brand.

Parsnipp commentary

OpenAI added bulk product campaigns to ChatGPT Ads, so advertisers can create multiple product feed campaigns and ad groups from one spreadsheet. A new Products tab also shows each item's review status and why it was rejected. This points to ChatGPT Ads moving from testing to something retailers can run at catalog scale, so it's worth checking how your product feed holds up here if it's already clean for other channels.

Parsnipp commentary

The Information's reporting puts numbers on Google's AI contribution pilot. Around 100 sites are participating, with one reportedly earning over $1 million a year and several smaller sites under $1,000 over several months. Payment applies only when content contributes significantly to an answer, and recipients say they don't know how amounts are calculated. That makes contribution to AI answers something that's starting to be measured and priced, which raises the value of tracking where your own brand shows up in them.

Parsnipp commentary

Andreessen Horowitz's Alex Immerman and Santiago Rodriguez argue that shopping assistants could own product discovery even when orders still flow through existing fulfillment, putting marketplaces' high-margin ad revenue at risk. That helps explain why Amazon blocks assistants while platforms like Instacart and Shopify are more welcoming. For brands, visibility to shopping agents will vary by platform, so product data needs to be readable by agents wherever they're allowed in.

Parsnipp commentary

The seller side of eCommerce is moving into LLM interfaces at the same pace as the shopper side. That means product data and listings now have to be built for agents to read correctly on both ends of a transaction, not just optimized for human shoppers.

Parsnipp commentary

This is the first real fight over who controls agent-led purchases, and it won't be the last. Any brand banking on agentic commerce needs a plan for each platform individually, since every walled garden is going to set its own rules for which agents get to transact.

Parsnipp commentary

Search Engine Roundtable reported that Google is testing anchor links inside AI Overviews that open a follow-up question in AI Mode instead of sending users to a website. It's another step toward keeping discovery inside Google's own AI surfaces rather than the open web. If your reporting still leans on organic click-through to prove AI search performance, this is one more sign that being cited inside the answer matters more than being linked below it.

Parsnipp commentary

A new NP Digital study of 68,000+ AI answers found challenger brand SoFi outperforming Bank of America on every visibility metric measured, including appearance rate and sentiment. It's concrete proof that AI search doesn't just preserve the existing pecking order — a smaller brand with the right content and citation footprint can out-perform a much larger, better-known competitor. Worth remembering that model behavior varies too: the same study found Gemini surfaces nearly double the brands per answer that Perplexity does.