Open Web Visits Fell 12% in 5 Years, New Data Shows

Visits to the world's top one million websites fell about 12% between early 2022 and mid-2026, according to an analysis SparkToro published on November 10, 2026, using Similarweb data. Per internet user, the drop is nearly 30%. It measures the web as a whole, not any single site.

What we know

Who published it: the study is by Rand Fishkin, SparkToro's co-founder, with help from Similarweb's research team. It went out on November 10, 2026.
The headline figure: total visits to the top 1 million sites went from a 2.243 trillion peak in Q1 2022 to 1.970 trillion in Q3 2026, a 12.2% drop, per the study. The low point was Q1 2026 (1.914 trillion, 14.7% below peak).
Per user, the fall is larger: visits per internet user dropped 29.2%, SparkToro says, because more people online share fewer visits. Fishkin warns the user count (from DataReportal) was revised mid-period and is less reliable than the visit totals.
The giants aren't pulling away: the top 100 sites' share versus everyone else rose about 3 points (roughly 33% to 36%), per the study. Fishkin calls the distribution largely stagnant over five years.
Where the data comes from: Similarweb's worldwide panel, which observes desktop and mobile web through the browser. It does not include in-app traffic, per the study.
The correction: the first version carried wrong data and a headline claiming a 24% decline. Fishkin acknowledged the error on X on November 10 and corrected the figure to 12%. The page's URL still keeps "24" in the link.

What changes and what doesn't

The figure does not say any one site lost 12% of its visits. It is an aggregate of a million sites measured by a panel, not a census of each site's real traffic. It measures only what goes through the browser, leaving out apps and AI assistants, where some attention has moved over these years.

It also doesn't prove a cause. The study describes a trend — fewer visits per person, a stagnant split between big and small — and its author offers a reading, but the number alone doesn't separate how much is AI in search, the shift to apps, or the measurement method itself. What it does make clear is that the open web, seen through the browser, is not growing at the pace of its users.

How to tell if you're affected

Don't treat −12% as your number. It's the whole web. Your real trend is in your own analytics and Search Console, not in this study.
Look at your long series. In Search Console's Performance report, compare the last 16 months of clicks and impressions. If your drop resembles the aggregate, it's context; if it's very different, the cause is yours (a page, a query, a technical change), not the general trend.
Account for apps and AI. The panel measures browsers. If your audience moved to apps or asks AI assistants, your browser traffic can fall even if your reach holds. Check where your visits come from (search, direct, social, referral).
Compare with your sector, not the world. A 12% global average says nothing about your niche. Look for data on your sector or competitors before drawing conclusions.
Diversify your entry points. If almost all your visits come from one source, a change there moves you as a whole. Working to be cited in AI answers, your own email list, and direct visits spreads the risk.
Log the date and come back in a quarter. It's a quarterly series; one month isn't a trend. Mark today's figure and compare it in three months.

Sources

Updates: if SparkToro or Similarweb publish new quarters or correct the figures again, they will be added here with the link.