Senior Engineer
As a tech lead, Jim works with clients through the full project cycle, translating their business requirements into actionable development work and working with them to find technical solutions to their challenges.
June 2, 2026
Last winter I was trying to plan something memorable for my daughter’s birthday. Our family goes to a lot of sporting events at the local university, so I started looking for party options.
I did what more and more of us are doing now. I didn’t open Google. I opened ChatGPT.
The answer was better than I expected. Specific, confident, detailed. Group ticket packages. University-branded gift bags. Special seating sections. And then one option that stopped me: a hockey package where the birthday kid gets to ride the Zamboni during intermission.
I could be parent of the year. I was already mentally writing the invitations. Then I went to find the page to book it. It didn’t exist.
After enough searching, I found what ChatGPT had actually read: a news article from eight years ago, describing a program the university had since discontinued. The page existed, the program had run, the details were accurate. They were just accurate about something that hadn’t been available for years. No one had retired the article. No one had marked it outdated or linked it to whatever replaced it. It simply sat there, long past the point any human searcher would find it, invisible to everyone except a system that reads everything and treats it all as equally current.
That distinction is worth holding onto. This wasn’t a hallucination. It was something quieter and harder to catch: accurate information about something that no longer exists, still present, still teaching. For most of the web’s history that kind of oversight was harmless. Stale content drifted past relevance and no one noticed. That’s no longer the case.
This isn’t a technology question. It’s a reputation question.
I spend a lot of time in rooms with people who are cautious about AI. That caution is legitimate. There are real obligations around privacy, accuracy, equity, and trust. The deliberation is appropriate. I’m not here to argue against it.
But there’s a distinction we’re not making clearly enough.
Adoption is a choice. Exposure is not.
You don’t have to deploy an AI tool, build an AI workflow, or take any institutional position on generative AI for it to be actively shaping your reputation. These systems are not waiting for your policy committee to finish its work. They’re already reading your website, processing your content, and forming a portrait of who you are and whether you’re worth recommending, right now, before your next prospective customer, student, or partner clicks a single link.
And you can’t see it happening. There is no report. No dashboard. The conversation is already underway. You’re just not in the room.
The first impression no longer happens on your website
This isn’t a higher education story, or a B2B story, or a consumer story. It’s all of them at once.
Consider what’s happened to search behavior in the last 18 months. According to Adobe’s May 2025 survey of 800 U.S. consumers, 77% of Americans who use ChatGPT now use it as a search engine. Nearly one in four goes to ChatGPT first, before Google. And 36% of all consumers have already discovered a new product or brand through ChatGPT. For Gen Z, nearly half have.
In B2B, the shift is equally stark. G2’s March 2026 survey of over a thousand software buyers found that half of B2B buyers now begin their software research with an AI chatbot more often than with Google, up from just 29% in April 2025. That’s nearly doubled in less than a year.
In higher education, EAB published survey data this past February showing that 46% of high school students now use AI tools during their college search, nearly double the rate from six months prior. Nearly one in five have already removed a college from consideration based on what AI told them.
Picture what that actually looks like. A 17-year-old, Sunday night, asking ChatGPT whether a particular school has a strong environmental engineering program. The answer comes back thin, vague, assembled from whatever that institution has left lying around. The school never makes the list. No one in admissions ever knows the conversation happened.
Not after a campus visit. Not after talking to an admissions counselor. Based entirely on what a language model assembled from everything that institution had ever published.
The common thread across all three of these: The decision happens before the website visit. Before a sales call. Before a form submission. Before anyone in your organization even knows someone was considering you.
The traffic you’re expecting may not arrive
There’s one more data point that deserves attention, because it names what many organizations are starting to feel without yet having language for it.
Semrush analyzed billions of visits across more than 50,000 sites and found that total web traffic was nearly flat in 2025, while AI traffic grew 66% and organic search declined in 13 of 17 industries. A separate field study published in 2026 found that when a Google AI Overview appears, organic click-through rates drop 61%. In Google’s newer AI Mode, the zero-click rate reaches 93%.
Demand didn’t disappear. It moved.
Since this post was first drafted, Google has made that trajectory official. At Google I/O in May 2026, the company announced that AI Overviews now reach 2.5 billion monthly users and that its conversational AI Mode has surpassed 1 billion. More significantly, Google confirmed it is building agentic search capabilities designed to perform information-gathering on users’ behalf — with traditional links described internally as increasingly secondary to the experience. The shift the data was pointing to is no longer a prediction. Google has named it as the direction.
The harder problem is that your analytics won’t tell you. Traffic may hold steady, or decline slowly enough to explain away, while the decisions that used to happen on your site are now happening somewhere you have no instrumentation for.
The people who would have visited your site to answer a question are now getting that question answered before they arrive. The top and middle of your funnel, awareness, consideration, initial shortlisting, may already be happening somewhere you can’t see, can’t measure, and currently have no ability to influence.
This doesn’t mean websites are dying. It means the website has lost its monopoly on first impressions. For many organizations, the AI answer is the first impression, and it’s being assembled from whatever you’ve left lying around.
The 20-minute diagnostic you should run this week
Before you commission a content audit, redesign a page, or draft a governance policy, do something simpler.
Ask.
Open ChatGPT. Open Perplexity. Open Claude. Open whatever tool your audience is using. Ask about your organization the way a prospective customer would. Ask what you’re known for. Ask what you offer. Ask whether you’d be a good fit for someone in a specific situation.
Then read the answer carefully. Not just for accuracy, but for recognition. Does it sound like you? Does it reflect who you are now, or who you were five years ago? Is it specific and confident, or vague in ways that suggest the underlying signal was weak?
Then do it for a peer organization. Compare the answers. You’ll feel the difference immediately between an organization whose content model is coherent and one whose isn’t.
Run that experiment this week. Project it on a screen with the people who make decisions about your content and let them watch it happen in real time. The conversation that follows is the one that needs to happen.
Your organization is already being summarized by machines. The only question is whether the story they tell is the one you intended.
Making the web a better place to teach, learn, and advocate starts here...
When you subscribe to our newsletter!