Google is testing AI-generated review replies in Google Business Profile.
Why we care. Responding to reviews can impact conversions and trust. But generic AI replies could be risky and erode trust, especially on negative reviews where authenticity matters most. Response quality matters more than whether a business replies to reviews.
What it looks like. Here’s a screenshot:
The details. Google appears to be rolling out a limited test of Reply to reviews with AI inside Google Business Profile.
The feature generates suggested responses to customer reviews.
Users can review, edit, and manually submit replies.
Availability is inconsistent across accounts and reviews.
The feature has been spotted in the U.S., Brazil, and India, but not widely in Europe.
Early behavior. Some users report prompts focused on older, unanswered negative reviews.
In at least one test, users could trigger AI responses in bulk.
There are conflicting reports on automation — some users say bulk responses still require review; others report fully automated replies can be published without edits.
First seen. The feature was first shared on LinkedIn by Chandan Mishra, a freelance local SEO specialist, and amplified by Darren Shaw, founder of Whitespark.
Google is testing AI-generated headline rewrites in Search results, describing it as a small, narrow experiment for now.
What’s happening. Google confirmed to The Verge (subscription required) that it’s testing AI-generated titles in traditional Search results, not just Discover.
The test is “small” and “narrow,” and not approved for broader rollout.
It impacts news sites but isn’t limited to them.
The goal is to better match titles to queries and improve engagement, Google said.
One example showed Google replacing original headlines with shorter or reworded versions, sometimes changing tone or intent (e.g., reducing “I used the ‘cheat on everything’ AI tool and it didn’t help me cheat on anything” to “‘Cheat on everything’ AI tool.”).
Why we care. Google Search is already sending fewer clicks. Now you also have to contend with Google generating entirely new headlines with AI, risking changes to meaning, brand voice, and click-through rates.
What they’re saying.Sean Hollister, senior editor at The Verge, wrote:
“This is like a bookstore ripping the covers off the books it puts on display and changing their titles. We spend a lot of time trying to write headlines that are true, interesting, fun, and worthy of your attention without resorting to clickbait, but Google seems to believe we don’t have an inherent right to market our own work that way.”
Google’s generation of title links on the Google Search results page is completely automated and takes into account both the content of a page and references to it that appear on the web. The goal of the title link is to best represent and describe each result.
Google said it uses these sources to “automatically determine title links”
Content in <title> elements
Main visual title shown on the page
Heading elements, such as <h1> elements
Content in og:titlemeta tags
Other content that’s large and prominent through the use of style treatments
Other text contained in the page
Anchor text on the page
Text within links that point to the page
WebSite structured data
What to watch. Google called this one of many routine experiments, but that’s no guarantee it stays small. The Verge noted a similar “experiment” in Discover later became a full feature.
Any future launch may not rely on generative AI, but Google didn’t explain how that would work.
Reaction. After seeing this news, Louise Frahm, SEO director at ESPN, wrote on LinkedIn:
“After 10+ years in news SEO, I’ve come to find that a headline is the most prominent element for attracting readers in timely windows, to provide a targeted synopsis that elevates your brand voice. If that vision gets altered and facts are misrepresented, long-term audience trust will be compromised.”
AI bots could outnumber humans on the web by 2027, according to Cloudflare CEO Matthew Prince, as agent-driven browsing explodes alongside generative AI adoption.
Prince made the prediction at SXSW, warning that bots are already reshaping how the internet is used — and how it’s monetized.
Why we care. Search is shifting from human clicks to AI-generated answers. If bots become the web’s primary “users,” you’ll need to reshape your strategy to ensure AI systems can access, trust, and use your content.
The details. Prince said AI agents generate far more web activity than humans because they gather information differently. A person shopping might visit five sites. An AI agent could hit thousands.
“If a human were doing a task… you might go to five websites. Your agent… will often go to a thousand times the number of sites.”
“So it might go to 5,000 sites. And that’s real traffic, and that’s real load.”
He also noted the web’s baseline is shifting fast.
“For a long time, the internet was about 20% bot traffic.”
“We suspect that in 2027 the amount of bot traffic online will exceed the amount of human traffic.”
Prince said this growth isn’t spiking like COVID-era traffic. It’s rising steadily with no end in sight.
Between the lines. Prince compared AI to past shifts like mobile and social. The difference: users may no longer visit websites directly. Instead, they rely on AI interfaces that aggregate and answer.
“The business model of the internet was… create content, drive traffic, and then sell things… That was the business model.”
“That breaks down because… bots don’t click on ads.”
“Customers are trusting the output from the helpful robot. They’re not clicking through the footnotes.”
AI sandboxes. AI agents also change how computing works behind the scenes. Prince described a future where “sandboxes” — temporary environments for AI agents — spin up and shut down instantly, potentially millions of times per second.
“You can… as easily as you open a new tab in your browser… spin up new code which can then run and service the agents.”
“We think that there will be literally millions of times a second these sort of sandboxes… being created… and then torn back down.”
The result: sustained pressure on internet infrastructure.
“We’re seeing internet traffic grow and grow and grow. And we don’t see anything that’s going to slow it down or stop it.”
The business impact. Companies are already split on how to respond to AI agents. Prince pointed to diverging strategies across major retailers.
“There are three radically different strategies about how they are going to interact with the bots.”
At the core is a bigger risk: losing the customer relationship.
“The nature of bots is going to be that it disintermediates the relationship between you and your customer.”
“Agents… don’t care about brand.”
For publishers. Prince argued AI could both hurt and help media. While AI reduces direct traffic and breaks ad-based models, AI companies need unique, original data — especially local and hard-to-replicate information — and may pay for it.
“Traffic has always been a really bad proxy for value.”
“What they actually want is… unique local interesting information they can’t get elsewhere.”
He pointed to local media as an example.
“If you don’t have the Park Record, then you don’t get that information.”
“We may make more off licensing our content to AI companies than we do off digital advertising.”
For small businesses. Prince was more blunt. AI agents optimize for price, quality and efficiency — not brand loyalty or proximity.
“My bot doesn’t care.”
“My bot is going to figure out actually who is the best… and route that traffic.”
That could erode traditional advantages.
“The shortcuts of trust that small business had in the past… are going to be much more difficult.”
“The natural tendency of AI is towards that level of aggregation.”
What to watch. The next phase of the web will hinge on control and compensation. Prince said:
“There has to be some exchange of value.”
“We’ve got to figure out… what’s going to pay for it.”
Prince said the core question is still unresolved:
“What is the future business model of the internet?… I don’t know what it’s going to be, but it’s going to change.”
Walmart said conversion rates for purchases made directly inside ChatGPT were three times lower than when users clicked through to its website.
Why we care. This suggests agentic commerce isn’t ready to replace traditional shopping. Sending users to owned environments still drives higher conversion rates.
The details. Starting in November, Walmart offered about 200,000 products through OpenAI’s Instant Checkout. Users could complete purchases inside ChatGPT without visiting Walmart’s site.
Daniel Danker, Walmart’s EVP of product and design, said those in-chat purchases converted at one-third the rate of click-out transactions.
He called the experience “unsatisfying” and confirmed Walmart is moving away from it.
Goodbye, Instant Checkout. Instant Checkout was designed to let users complete purchases directly inside ChatGPT without visiting a retailer’s website. However, earlier this month, OpenAI confirmed it was phasing out Instant Checkout in favor of app-based checkout handled by merchants.
What’s changing. Walmart will embed its own chatbot, Sparky, inside ChatGPT. Users will log into Walmart, sync carts across platforms, and complete purchases within Walmart’s system.
A similar integration is coming to Google Gemini next month.
Perplexity’s new Comet browser for iOS defaults to Google Search. That’s because mobile queries often focus on navigation, local results, and transactions, where “Google does a much better job … than anyone else … including Perplexity,” according to Perplexity CEO Aravind Srinivas.
Comet for iOS. It includes Perplexity’s AI assistant directly in the browser. Comet for iOS also blends AI answers with standard search results. For many queries, you’ll still see a traditional results page.
You can ask questions by voice while browsing.
The assistant can summarize pages, answer questions, and take actions like drafting emails.
Deep Research features generate cited summaries and prep materials.
What Comet does. According to Perplexity, the assistant can act on your behalf. Examples include:
Summarizing articles and sharing outputs.
Researching people or topics across tabs.
Assisting with bookings or form fills.
What Perplexity is saying.
“The search experience in Comet iOS provides traditional search results pages for fast, local, and high-intent queries that are more common on mobile. Meanwhile, the Comet Assistant easily allows for more advanced knowledge and intelligence powered by the Perplexity answer engine. The intention is for users to have the smoothest browsing experience possible for the real use cases of iOS.”
Why we care. The near future of search increasingly looks hybrid, which means you’ll need to optimize for traditional Google results and AI-driven answers. This also reinforces Google’s dominance in commercial and local search while shifting competition to the AI layer.
Google’s AI Overviews now appear on 14% of shopping queries, up 5.6x from 2.1% in November 2025, according to new Visibility Labs analysis.
Ecommerce brands have been mostly unaffected by AI-driven click loss in Search. That seems to be changing.
Why we care. As Google’s AI Overviews expand across product searches, ecommerce brands face a growing risk of losing visibility and clicks before shoppers reach standard organic or Shopping listings.
The details. The analysis targeted product-intent keywords tied to results with a Shopping box, paid or organic — terms like “weighted blanket,” “mushroom coffee,” “protein powder,” and “blue T-shirts.”
That produced 20,900,323 shopping keywords.
Of those, 2,919,229 triggered an AI Overview — 14.0% penetration.
What they’re saying. Report author Jeff Oxford, founder and CEO of Visibility Labs, concluded:
“Focusing on AI SEO is no longer a luxury, it’s becoming a necessity. Ecommerce sites need to think beyond traditional SEO and start incorporating AI SEO best practices into their search optimization strategy.”
Small publishers are seeing sharp traffic declines from AI search experiences, according to new data from thousands of global sites using Chartbeat analytics.
The details. Publishers with 1,000 to 10,000 daily pageviews lost 60% of search referral traffic over two years, Chartbeat found.
Mid-sized sites with 10,000 to 100,000 daily pageviews lost 47%.
Large publishers with more than 100,000 daily pageviews were down 22%.
Reality check. AI referrals aren’t replacing lost search traffic.
Google Search pageviews fell 34% year over year.
Google Discover dropped 15%.
ChatGPT referrals rose 200% but still account for less than 1% of total traffic.
Yes, but. Traffic is shifting, not disappearing. Total weekly pageviews across publishers fell just 6% from 2024 to 2025, a typical swing tied partly to the news cycle. Search is shrinking as a share of traffic, while direct, internal, and messaging channels are growing.
Why we care. SEO has long been the growth engine for smaller sites. That’s no longer true. If you don’t have a strong brand, direct audience relationships, repeat visitors, or differentiated value, you face the biggest risk as search referrals decline.
Visibility is no longer just about ranking. It depends on whether your content is discovered, evaluated, and selected in AI-driven search experiences.
We’re kicking off our new monthly SMX Now webinar series on April 1 at 1 p.m. ET with iPullRank’s Zach Chahalis, Patrick Schofield, and Garrett Sussman on how you must adapt.
The session introduces iPullRank’s Relevance Engineering (r19g) framework for executing Generative Engine Optimization (GEO) through an omnichannel content strategy. You’ll learn how AI search uses query fan-outs to discover and select sources, and how to structure content so it’s retrieved, surfaced, and cited.
It also emphasizes that GEO success isn’t universal. It requires testing, tailored strategies, and a three-tier measurement model spanning discovery, selection, and citation impact.
Google is expanding Personal Intelligence across AI Mode, Gemini, and Chrome in the U.S., moving it beyond beta into broader consumer use.
Why we care. Personal Intelligence pushes Google further into fully personalized search, using first-party data like Gmail and Photos. That makes results harder to replicate, rank against, or track — especially in AI Mode, where outputs may vary based on user history, purchases, and behavior.
The details. Personal Intelligence now works across:
AI Mode in Google Search (available now in the U.S.)
Gemini app (rolling out to free users)
Gemini in Chrome (rolling out)
How it works. Users can connect apps like Gmail and Google Photos so Google can tailor responses using personal context. Examples Google shared include:
Shopping recommendations based on past purchases and brand preferences.
Tech troubleshooting using receipt data to identify exact devices.
Travel suggestions using flight details, timing, and past trips.
Personalized itineraries and local recommendations.
Hobby suggestions inferred from user interests.
Availability. These features are available only for personal accounts, not Workspace users, Google said.
Yahoo CEO Jim Lanzone said AI-powered search — especially Google’s AI Mode — is putting the open web’s core traffic model at risk and argues AI search engines must send users back to publishers.
“I think that the LLMs are one big reason that they’re under threat, with AI Mode in Google being the biggest challenge.”
“Those publishers deserve [traffic], and we’re not going to have the content to consume to give great answers if publishers aren’t healthy.”
Why we care. Many websites are seeing less traffic from answer engines like Google and OpenAI — and I think it’ll only get worse. So it’s encouraging to see Yahoo trying to preserve the “search sends traffic” model. As he said: “We have very purposefully highlighted and linked very explicitly and bent over backwards to try to send more traffic downstream to the people who created the content.”
Yahoo’s AI stance. Yahoo is taking a different approach from chatbot-style interfaces, Lanzone said on the Decoder podcast. He added that Yahoo isn’t trying to compete as a full AI assistant:
“Ours looks a lot more like traditional search and it is more paragraph-driven. It’s not a chatbot that’s trying to act like it’s a person and be your friend.”
“We’re not a large language model. We’re not going to be the place you come to code. We’ve really launched Scout as an answer engine.”
What’s next: Personalization + agentic actions. Yahoo plans to expand Scout beyond basic answers and is embedding AI across its ecosystem:
“You are very shortly going to see us get into very personalized results. You’re going to see us get into very agentic actions that you can take.”
“There’s a button in Yahoo Finance that does analysis of a given stock on the fly… It is in Yahoo Mail to help summarize and process emails.”
Yahoo vs. Google isn’t a thing. Yahoo isn’t trying to win by converting Google users directly. Instead, Yahoo is prioritizing its existing audience and increasing usage frequency over immediate market share gains:
“Nobody chooses, you will not be surprised, Yahoo over Google or somewhere else to search. The way that we get our search volume is because we have 250 million US users and 700 million global users in the Yahoo network at any given time. There’s a search box there. And infrequently, they use it.”
A warning. Companies — including publishers — should be cautious about relying too heavily on AI platforms as intermediaries. Lanzone compared today’s AI partnerships to Yahoo’s past reliance on Google:
“You are tempting fate by opening up a way for consumers to access your product within a large language model.”
“The big bad wolf will come to your door and say everything’s cool.”
Login
