Your brand may be losing ground in AI search
Editor’s note: This is AI Impact, Newsweek’s weekly newsletter where each week, we will explore how business leaders are unlocking real value through artificial intelligence.
Tap or click here to get this newsletter delivered to your inbox.
Good morning and thanks for joining me.
One change I’ve kept coming back to this year is how much of the buying process can now happen before someone ever reaches a company’s website.
Earlier this year, I wrote about how strong Google performance no longer guarantees visibility when buyers turn to ChatGPT or another AI tool to research their options. Later, after speaking with Synchrony, I explored what happens when agents begin helping people compare products and narrow their choices on their behalf.
Those stories left me interested in a part of the customer journey that companies have historically had much less visibility into: what happens before a buyer reaches the site at all.
Marketers have spent years learning to read search rankings, traffic, clicks and conversions. Those still matter, of course, but they may not show which brands are appearing in AI answers, which sources those systems are relying on or whether an agent is visiting a site because someone is actively researching something the company sells.
AI visibility can deteriorate before the impact appears in website traffic, conversion or the sales pipeline.
Today, I’m diving into how marketers can tell whether that visibility is improving or slipping, which signals are worth tracking over time and what traffic from AI agents may reveal about customer intent.
Read on for what marketers may be missing in AI search.
Signal Capture
Signals from the frontlines of AI adoption
Sitecore, Scrunch CEOs: SEO Metrics Can Hide AI Search Problems
Search rankings and web traffic still tell marketers how customers find a brand online, but they can miss what is happening inside AI-generated answers. Among in-house marketers in a survey conducted by AI-search company Scrunch and professional services marketing agency Scribewise, 62 percent said leaders outside marketing or communications still judge their teams on traditional SEO metrics or web traffic.
Eric Stine, CEO of digital experience software company Sitecore, and Chris Andrew, CEO of Scrunch, a Sitecore company that helps brands track and improve how they appear in AI search, said a business can maintain strong Google rankings while competitors gain visibility in answers from ChatGPT, Gemini, Perplexity and other AI platforms.
“Healthy search rankings can hide an AI search problem,” Stine and Andrew told me over email.
The broader survey included 602 full-time U.S. marketing and PR professionals working both in-house and at agencies. Eighty-four percent of respondents said they were confident they could shape AI answers, even as 51 percent were unsure whether their current strategy was the right one. Fifty-six percent could not distinguish AI search optimization from traditional search engine optimization.
A single AI response is also a weak benchmark for brand visibility because the answer can change depending on how a question is phrased, what context the user provides and which platform responds.
“Think of it like meteorology: One prompt is a single reading, and it’ll bounce around,” Stine and Andrew said. “A consistent set of prompts tracked over time behaves like a network of weather stations.”
They recommend tracking more than whether a brand appears at all. Marketers can look at how prominently it appears, whether it is described favorably, how often it shows up relative to competitors and which sources are cited.
Citations can show which sources are shaping an AI answer, from a company’s own website to analyst research, customer reviews, news coverage, social media and other material outside its direct control.
“You can’t edit the answer, but you can change what the answer is built from,” they said.
Companies can respond by identifying where brand information is missing or inconsistent, making their content easier for AI systems to access and understand and monitoring the third-party sources those systems use.
Traditional web analytics may also be filtering out or missing another signal: visits from AI agents.
“For 25 years, marketers treated bot traffic as spam. That instinct is now a liability,” the two CEOs told me.
They distinguish between agents that train on or index web content and retrieval agents that gather information in response to a person’s question. When a retrieval agent visits a website, it can indicate that someone is actively researching a company, product or category through an AI service.
Only about a third of respondents to the survey said they analyze AI bot traffic. Tracking which AI systems visit a site, which pages they access and which ones they bypass can give marketers another view of what information is being found and used.
Stine and Andrew said people reaching a website after AI-assisted research should arrive better informed and more qualified. Companies can then compare AI visibility with conversion trends to gauge whether that exposure is producing business value.
AI visibility can deteriorate before traditional marketing metrics move at all. Competitors can begin appearing in answers where a company does not, or AI systems can increasingly cite sources that omit the brand, while its Google rankings remain steady.
“If a company waits for the impact to appear in website traffic, conversion, or pipeline, it is already behind,” they told me.
“And in a compressed funnel, there may not be a second chance to enter the consideration set.”
Upcoming Webinars
Overcoming Barriers to AI Transformation in Legacy Industries
AI investment is accelerating, and long-established companies are finding ways to bring the technology into organizations shaped by years of customer relationships, complex operations, critical systems and regulatory responsibilities. The challenge is turning promising pilots into lasting business results without disrupting the strengths that have made those companies successful.
In an upcoming Newsweek AI webinar presented by Cognizant, Gabriel Snyder, Newsweek’s executive editor, enterprise, will moderate a discussion titled “Overcoming Barriers to AI Transformation in Legacy Industries.” The conversation will delve into how established organizations are connecting AI to existing systems and workflows, clarifying ownership and governance and aligning technology investments with people, processes and measurable business outcomes.
Matt Sanchez, chief operating officer at Yahoo, Durga Malladi, executive vice president and general manager of technology planning, edge solutions and data center at Qualcomm and Katy George, corporate vice president of workforce transformation at Microsoft, will join the conversation.
Join the live discussion on October 21. Register for free.
AI Impact Forum Webinar
On October 22, Dr. Ranjit Tinaikar, host of Newsweek’s “AI Impact Forum,” will sit down with Firdaus Bhathena, executive vice president and chief technology and transformation officer at S&P Global, to discuss how large companies are approaching AI, productivity and transformation.
Drawing on Bhathena’s experience at S&P Global, FIS Global and CVS Health, the conversation will examine the decisions executives face as new capabilities reshape technology strategy, operating models and the way work gets done.
Join the live discussion Thursday, October 22, at 2 p.m. ET. Register for free.
Prompt Injection
What’s one recent insight you’ve learned about AI?
“One realization that has stayed with me is that AI is only as valuable as the clinical truth it helps preserve. In the mid-revenue cycle, the issue is often not that the care was not delivered, but that the final medical record does not fully reflect the patient’s true clinical complexity.
That is the problem we built Accuity’s physician governed AI engine, Amplifi, to address. The goal is not simply to find more revenue. It is to help ensure that the documented and coded record is accurate, complete and compliant, and that it faithfully represents the care the patient actually received.
As both a physician and a CMIO, I see that as the real opportunity for AI: not to replace clinical judgment, but to help clinicians, CDI teams and coders identify where the clinical story may be incomplete before the record is finalized.
It has made me much more disciplined about where AI should—and should not—be used.
AI is very effective at reviewing large volumes of clinical information, identifying patterns and surfacing cases that deserve a closer look. But it should not make the final clinical judgment. In healthcare, context matters, and many decisions sit in a gray area that requires physician, CDI and coding expertise.
That is why we designed Amplifi around clinical governance and human review. The technology helps focus attention and scale the work, but qualified clinicians and coders remain accountable for the final decision.
For me, successful healthcare AI is not defined by how much it automates. It is defined by whether it improves the accuracy, integrity and defensibility of the medical record while preserving human clinical judgment.”
Have your own lesson to share? Email me at: a.mills@newsweek.com
Run Log
AI use case of the week
Across a multi-site specialty practice, having enough drug inventory overall does not mean each clinic has what its scheduled patients need. One location can hold excess vials while another runs short. Oncology and retina clinics also buy costly specialty drugs before reimbursement arrives from payers, so unused inventory ties up cash and may expire.
Ravi Seshadri, chief technology officer of AllyGPO, a specialty group purchasing organization, said most practices forecast drug demand from recent purchase history. Those records show what they bought, rather than what patients are scheduled to need next.
AllyGPO built an AI forecasting system that combines medication orders for upcoming treatment with scheduled patient appointments. A medication order shows which treatment a patient is expected to receive, while the scheduled appointment shows when that treatment is expected to occur. The system uses those two inputs to predict how many vials of each drug each location will need.
Forecasting by location addresses a problem that an overall inventory count can mask: enough medication across the network, but too much at one clinic and too little at another. Because drugs can be difficult to transfer between clinics, excess inventory at one site may not solve a shortage at another.
Seshadri said practices using the forecasts have reduced the amount of medication they keep on hand, freeing cash that would otherwise be tied up in inventory without compromising their ability to treat scheduled patients. Purchasing can then be based on expected patient demand instead of previous consumption alone.
Finance and operations teams still have to be willing to act on the forecast.
“If they can’t understand and audit how a forecast was produced, they won’t make a multimillion-dollar purchasing decision based on it,” Seshadri said.
AllyGPO expresses the forecast in vials rather than milligrams or dollars because clinics buy and manage these drugs in fixed-size vials, and dosing can produce waste. With practices paying for these drugs before reimbursement arrives, getting the count right is as much a working-capital decision as an inventory one.
Have an interesting AI use case to share with us? Email me at: a.mills@newsweek.com
Context Window
■ OpenAI introduced a framework for publicly reporting model misalignment and released six initial cases involving behaviors such as concealing mistakes, using exposed API keys, uploading files without permission and unauthorized communication between agents. [OpenAI]
■ Agency executives say rising AI costs and metered model usage are forcing firms to scrutinize token spending, rethink their technology stacks and experiment with passing some AI costs on to clients. [Digiday]
■ State banking regulators released a voluntary framework that gives examiners a three-tier system for assessing AI risk at financial institutions, including how AI affects customers, uses sensitive data and relies on human oversight. [Banking Dive]
■ Salesforce and Nvidia introduced Koa, an open-weight reasoning model for sales, marketing and customer-service tasks that Salesforce says can handle some enterprise workloads with fewer tokens while keeping customer data within its existing security controls. [TechCrunch]
■ A survey of more than 500 CISOs found that 69 percent identified AI as their top priority for new cybersecurity budget dollars, even as security budgets grew only modestly and questions remain about which AI use cases deliver the strongest returns. [Dark Reading]
Transfer Protocol
Tracking executive moves across the AI landscape
Brett Kelsey, a four-time CISO who most recently advised security and technology companies through Omni Cyber Solutions, has been named chief AI officer at Athena Agentic, leading development of its agentic reasoning capabilities and autonomous decision models.
Senthil Velayutham, coming from the chief product and technology officer role at Nextiva, has joined Omega Healthcare as chief technology officer, leading its global technology and product organization and accelerating its AI strategy and Omega Digital Platform.
Arjun Sainath, after serving as chief technology officer at Cart.com, has been appointed chief technology officer at OnTrac, overseeing technology strategy and the use of AI, automation and advanced analytics across its operations and customer systems.
Grant Davis-Denny, a partner at Munger, Tolles & Olson, has taken on the firm’s newly created head of legal AI role, directing AI strategy, developing workflows for attorneys and guiding the evaluation, adoption and safe use of new tools.
George Llado, a TileDB board director and former chief information officer at Merck and Alexion Pharmaceuticals, is the new CEO of Tile.ai, leading the company as it builds a governed enterprise data layer designed to make data securely usable by AI agents.
Know someone on the move in AI? Send job change info to a.mills@newsweek.com
Magic Moment
What’s the most fun or unexpected way you’ve used AI lately?
“One of the most fun uses of AI recently was watching my 2.5-year-old son interact with Gemini voice AI to create a story about Peter Rabbit, who is actually a firefighter, working together with an aerospace engineer on a rocketship.
It was fascinating how naturally he interacted with it. He was able to develop the story in real time by shaping where it went next, which characters to include, and which topics to bring in, including something as specific as aerospace engineering. The story itself was still a little average, but it was impressive to see the creativity on both sides.”
Experience some AI magic? Tell us about it at a.mills@newsweek.com
Tap or click here to get this newsletter delivered to your inbox.
