In Q3 2026, total visits across the million most-visited sites sat 12.2% below their Q1 2022 peak, according to Similarweb data that SparkToro published and Search Engine Journal covered Thursday. The share of those visits going to the top 100 sites stayed roughly flat across the period. The pie shrank, but the slices kept their proportions.
SparkToro co-founder and CEO Rand Fishkin started the project to test a theory that Big Tech properties were absorbing a growing share of all web traffic. The data did not support it. Fishkin described the top-100 share change as small, writing that “stagnation is a more accurate description of the trend.”
The totals trace a clear arc. Visits peaked at 2.243 trillion in Q1 2022, fell to a low of 1.914 trillion in Q1 2026 (14.7% below the peak), and recovered slightly to 1.970 trillion in Q3 2026. The chart stays roughly level through 2023, dips in early 2024, and declines again during early-to-mid 2025.
The per-user figure is the weaker number. SparkToro reports visits per internet user down 29.2%, but it divides Similarweb’s visits by DataReportal’s user estimates, which were revised in the middle of the timeline. DataReportal’s October 2025 report attributed a jump from 5.65 billion to 6.04 billion users in a single quarter to “significant revisions” in figures from India and China, and warned against reading it as new people coming online. The SparkToro post did not note that revision. Of the two headline numbers, the 12.2% visit decline depends on fewer assumptions.
The data has hard limits. Similarweb’s panel counts desktop and mobile browser sessions but excludes apps, and Fishkin wrote that Big Tech consolidation probably shows up far more inside apps, which the data cannot see. The figures are also worldwide. Fishkin wrote that sites serving American, British, and European audiences may have lost traffic while Asian and African sites, where internet access is expanding, gained it. The post neither breaks out regions nor states the panel’s size.
Fishkin did not offer a cause for the decline. His list of possible factors covers apps replacing browsers, dependence on search results, AI tools, social platforms, and changed behavior after the COVID-19 pandemic. Without a causal test, each of those stays a hypothesis, and none can be blamed or cleared on this evidence.
Two provenance notes matter for anyone citing the study. SparkToro corrected its article after publication: Fishkin posted on X on Nov. 10 that the earlier version “had an incorrect graph/data.” The old headline claimed a 24% decline, and the URL still carries that number. SparkToro makes its money from audience research software, and Fishkin is a co-author of “Zero Click Marketing.” Weigh the findings with that commercial context in mind.
For search teams, the useful reading is narrow. A multi-year decline in browser visits is a plausible backdrop when your own Search Console and analytics trends slope down, and the flat top-100 share suggests that large platforms are not the sole explanation for your losses. It does not tell you whether your market, your language, or your device mix matches the global panel. Search Engine Journal’s writeup also separates this measure from the earlier zero-click data, which counted clicks for each American Google search rather than total browser visits.
Until SparkToro or Similarweb publishes regional cuts, use the 12.2% figure as context in stakeholder conversations, not as a forecast. Over the next ninety days, compare your own year-over-year visit trends against this benchmark, and pull app and regional traffic into the analysis before attributing any drop to AI answers or search itself.
Search Engine Journal, Matt G. Southern, November 12, 2026, reporting Similarweb data shared by SparkToro.
