Your search traffic is not down because your SEO broke. The search results changed.

If your inquiry volume from organic search has been sliding for four straight quarters and nobody on your team can tell you why, the answer is not that your agency stopped doing its job. The results page your prospective adult learners see is a different product than it was two years ago, and the traffic that used to arrive as a byproduct of ranking well is now being absorbed before it ever reaches you.

Here is the number that should reset your expectations: 68.01 percent of Google searches ended without a click to anyone in early 2026, according to SparkToro's June 9, 2026 analysis of Similarweb US panel data covering January through April. That is up from 60.45 percent in 2024. SparkToro is candid that its panel excludes Google's mobile search app, where zero-click behavior runs more aggressive than on the web, which means the real figure is probably worse than 68 percent rather than better.

Pew Research Center measured the mechanism directly. Pew recruited about 900 US adults who consented to browser tracking and captured 68,879 real Google searches during March 2025, published July 22, 2025. When an AI summary appeared on the page, users clicked a traditional search result on 8 percent of visits. When no AI summary appeared, they clicked on 15 percent of visits. Close to a halving. The links inside the AI summary itself were clicked on 1 percent of visits, which should end any strategy built on getting cited inside an AI Overview as a traffic play. Users ended their browsing session entirely on 26 percent of pages with an AI summary against 16 percent without, which means the summary frequently does not redirect the visit, it finishes the question. Google publicly disputed the Pew study on methodology, but it objected without publishing a counter-dataset.

You are not losing to a competitor who out-optimized you. You are losing to a results page that answers the question itself.

The fix your vendors are selling does not exist

The moment enrollment leaders understand the traffic problem, a whole industry arrives with a technical solution: add schema markup, publish an llms.txt file, restructure your program pages for the answer engines. It is a clean story, it comes with an implementation invoice, and Google's own documentation contradicts it in three sentences.

From Google Search Central's page on AI features and your website: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." On the markup question specifically: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add." Google states the eligibility bar as simply that "a page must be indexed and eligible to be shown in Google Search with a snippet."

That is the primary source, published by the company that operates the system, and the firms selling answer engine optimization talk around it rather than through it. When pressed, the argument usually retreats to schema helping indirectly through entity understanding, which is unfalsifiable by design, and every source we found making the claim sells schema implementation as a service. The llms.txt story is thinner still: Google says you do not need to create machine readable files, and no major answer engine has demonstrated adoption of the format. If your team is holding a proposal to make your program pages legible to AI, you are being sold a product that the platform's own documentation says is not required.

What did not change

The strongest pushback we get from enrollment leaders is the reasonable one: if the summary eats the click, is content dead, and should we stop publishing? No, and the distinction matters because it determines where your budget goes.

Informational and explanatory queries are what got absorbed. The article titled along the lines of "what is a competency based bachelor's degree" is now, in effect, free training material for a summary that answers the question and keeps the reader. Those queries were also never the ones that filled your class. Adult learners at the decision moment search differently: they search for a named program, a named credential, a schedule, a cost, a transfer policy, a licensure outcome in a specific state. Those are program-specific, local, and transactional queries, and they still convert a ranking into a visit and a visit into an inquiry, because a summary paragraph cannot tell a working parent in your metro whether her sixty existing credits transfer into your evening RN to BSN cohort.

The second objection is sharper: AI Overviews clearly cite well-structured pages, so is structure not itself an optimization? Structure helps, and it helps for a reason that has nothing to do with markup. Content that answers a question completely and self-containedly in a single passage gets lifted, because that is how retrieval works. Content that requires assembling context across a long page does not. That is a writing and information-architecture behavior, not a technical product you install, and it is precisely what Google is describing when it says no special optimization is necessary. The other thing that appears to matter is corroboration: when your program, your outcomes, and your claims appear consistently across independent places rather than only on your own domain, they surface more readily in generated answers. Being referenced elsewhere is not a plugin either.

Where the enrollment dollars should sit

EAB's 2026 Higher Ed Marketing Outlook, drawn from more than 120 higher education marketing executives, reports that 61 percent of enrollment marketing dollars now support digital while spending growth has stalled, and names the growing influence of AI-driven search as a force reshaping visibility. Those are the findings EAB published on the landing page. The full report is gated, so treat this as the summary the firm chose to make public rather than a dataset anyone has audited.

Read alongside the click data, that combination is uncomfortable: flat budgets, a majority of them in digital, and the free half of digital quietly repricing itself. Which means the reallocation is the decision in front of you, not a technical remediation project.

Four things hold up. Program-level content that answers a prospective adult learner's actual decision questions completely, on the page, without making her assemble the answer from four places. Corroboration, meaning your outcomes and credentials show up in places you do not own, from licensure boards to employer partners to regional press. Paid search at the decision moment, because the queries that still convert are exactly the ones with commercial intent, and paid placement does not care what the summary above it says. And owned channels, your email list and your inquiry nurture in particular, because they are the only channels where no algorithm sits between you and the student.

None of that is a new build. Most institutions already run all four and fund them as if organic search were still delivering what it delivered in 2023. The correction is a reallocation, and it starts with your team no longer being able to say the traffic decline is a mystery.