Your favorite website might be rotting away

Site visitors to the web’s most established web sites fell greater than 11% over the previous 5 years, revealing the structural pressures conventional publishers face as synthetic intelligence reshapes how folks discover info on-line, in accordance with data from Similarweb analyzed by Axios and published January 3, 2026.

The decline impacts older web sites that constructed their companies throughout the search-dominated period of the web. These websites do not merely disappear when site visitors drops. They persist on the open internet for years, accumulating damaged hyperlinks and outdated info that clutters search outcomes whereas offering diminishing worth to each human readers and AI techniques coaching on internet content material.

Web utilization continues rising globally, however the site visitors patterns inform a narrative of redistribution quite than enlargement. General site visitors to the highest 1,000 web sites remained comparatively secure at roughly 300 billion common month-to-month internet visits over the previous 5 years. The mixture even confirmed modest development of 1.8% between November 2024 and November 2025.

Take away newer web sites from that calculation, nevertheless, and a special image emerges. Websites that existed 5 years in the past skilled a 1.6% site visitors decline over the previous yr when measured on the identical foundation. The determine excludes web sites that disappeared fully throughout the measurement interval.

1 / 4 of internet pages have vanished

Analysis from Pew Analysis Middle documented the dimensions of content material decay throughout the web. The examine discovered that 25% of all webpages that existed sooner or later between 2013 and 2023 are not accessible. The disappearance charge varies considerably by content material age, with pages created in 2013 displaying 38% inaccessibility by October 2023, whereas pages from 2023 confirmed solely 8% inaccessibility.

Information organizations face explicit vulnerability to hyperlink decay. In line with the Pew findings, 23% of reports webpages include at the least one damaged hyperlink. Authorities web sites reveal comparable issues, with 21% of presidency pages containing damaged hyperlinks.

The damaged hyperlink drawback compounds as time passes. Pages created in 2014 confirmed 35% inaccessibility, declining progressively to 31% for 2015 content material, 30% for 2016 materials, and 26% for 2017 pages. The sample accelerated once more in subsequent years, with 2018 displaying 31% inaccessibility and 2019 reaching 32% earlier than declining to 27% for 2020 content material.

Content material from newer years confirmed higher preservation charges. Pages from 2021 demonstrated 22% inaccessibility, 2022 confirmed 15% inaccessibility, and 2023 content material reached solely 8% inaccessibility by the October 2023 measurement date.

The sluggish degradation of older web sites creates important challenges for each customers searching for info and AI techniques making an attempt to collect coaching information. Engines like google return outcomes pointing to pages that not exist or redirect elsewhere. Customers encounter irritating lifeless ends. AI coaching techniques ingest outdated or incorrect info that undermines mannequin high quality.

Publishers battle whereas some websites neglect upkeep

Some publishers have invested sources in sustaining and updating their internet properties regardless of site visitors challenges. These efforts embody refreshing content material, fixing damaged hyperlinks, updating outdated info, and guaranteeing technical infrastructure stays purposeful. The investments characterize important ongoing prices with out corresponding income will increase.

Many web sites, nevertheless, take a special method. In line with the Axios evaluation, quite a few older websites stay on-line however obtain minimal upkeep. These properties proceed consuming server sources and area registrations whereas offering more and more restricted worth to guests. The uncared for websites accumulate technical debt, outdated content material, and damaged performance that degrades person expertise.

The upkeep problem extends past particular person pages to whole website architectures. Older content material administration techniques fall behind on safety updates. Third-party companies built-in into pages disappear or change APIs. Photographs hosted on exterior companies turn into unavailable. Embedded content material from social media platforms breaks as these platforms evolve their embedding techniques.

Publishing business advocates have spent years arguing that AI companies scraping writer content material would finally eradicate incentives for creating new materials. The priority facilities on a possible suggestions loop the place AI techniques cut back writer site visitors, publishers cut back content material creation, and AI techniques then lack contemporary coaching information past more and more outdated materials.

Knowledge from a number of sources helps the site visitors discount thesis. Analysis from Ahrefs discovered that AI Overviews lowered natural clicks to top-ranking web sites by 34.5% when current in search outcomes. The measurement in contrast click-through charges for similar search queries with and with out AI abstract options.

Additional research confirmed AI chatbot referral site visitors to prime media and information web sites working roughly 96% decrease than conventional Google search referrals. The dramatic distinction displays basic modifications in how AI-mediated discovery differs from link-based search.

OpenAI leads compensation makes an attempt

The popularity of potential content material exhaustion has prompted some AI firms to ascertain direct writer relationships. OpenAI has emerged as essentially the most lively participant in these preparations, in accordance with business reporting. The corporate has signed a number of agreements compensating publishers for content material entry.

The offers differ in construction and phrases. Some embody direct licensing charges paid upfront or on recurring schedules. Others incorporate revenue-sharing elements the place publishers obtain funds primarily based on how regularly AI techniques reference their content material. The precise phrases usually stay confidential beneath non-disclosure provisions.

Publishers have advocated for establishing a “two-sided internet market” the place AI firms pay for content material on a per-usage foundation. The idea mirrors dynamics from the search period when publishers obtained site visitors and promoting income from search engine referrals. The proposed framework would create direct fee flows from AI firms to content material creators.

The compensation preparations stay restricted in scale relative to the publishing business’s dimension. Dozens of publishers worldwide have signed offers with numerous AI firms, however 1000’s of publishers exist with out such agreements. The funds characterize small fractions of what these publishers beforehand earned from search site visitors.

Implementation challenges stop fast scaling of compensation fashions. Measuring content material utilization in AI techniques proves technically complicated. Figuring out honest fee charges lacks business requirements. Negotiating particular person offers with quite a few publishers requires substantial administrative sources. Smaller publishers lack negotiating leverage to safe favorable phrases.

Some publishers have taken defensive measures by blocking AI crawlers from accessing their content material. Analysis findings recommend this technique typically backfires. Studies documented that publishers blocking AI crawlers skilled 23% whole site visitors losses and 14% human site visitors declines in comparison with publishers permitting AI entry.

Site visitors patterns shift towards AI experiences

The site visitors redistribution displays altering person behaviors round info discovery. Conventional search workflows concerned submitting queries, reviewing a number of outcome listings, clicking by way of to numerous web sites, and synthesizing info throughout sources. This multi-step course of generated substantial site visitors to writer websites.

AI-powered search experiences compress these workflows into single interactions. Customers obtain synthesized solutions inside search interfaces, typically eliminating the necessity to go to exterior web sites. The compression improves person comfort whereas lowering writer site visitors and income alternatives.

Google’s Liz Reid has defended the corporate’s AI search options by arguing they generate “increased high quality” clicks than conventional search. The manager outlined high quality clicks as these the place customers do not instantly return to go looking outcomes, suggesting real curiosity in vacation spot content material.

The standard argument supplies restricted consolation to publishers experiencing absolute site visitors declines. Increased engagement per go to does not offset income losses when total go to quantity drops considerably. Publishers require particular site visitors volumes to help their enterprise fashions no matter per-visit high quality metrics.

Google has claimed that web sites experiencing site visitors losses had different components contributing to declines past AI options. The corporate pointed to methodology considerations in third-party research and recommended some publishers would have misplaced site visitors no matter AI implementation.

Writer teams have disputed these characterizations. Trade organizations representing information retailers have described AI options as systematic appropriation of writer work for utilizing content material with out producing corresponding site visitors and income.

Google has expanded entry to AI-powered search experiences to offer seamless transitions from conventional search outcomes into conversational interfaces. The enlargement additional reduces friction in AI-mediated discovery whereas probably lowering writer referrals. Customers can now have interaction with AI options extra immediately inside search interfaces the place writer hyperlinks seem much less prominently.

The measurement hole

Monitoring site visitors sources has turn into extra difficult within the AI period. Google has addressed some writer considerations about attribution and measurement capabilities. The additions do not absolutely remedy measurement challenges. AI techniques typically synthesize info from a number of sources with out clearly attributing particular claims to particular person publishers. Customers might eat info by way of AI interfaces with out visiting supply web sites, creating info flows that bypass conventional analytics fully.

Industry research has quantified the acceleration of site visitors declines. Teads, a significant promoting expertise platform working with 1000’s of premium publishers, reported 10-15% pageview declines throughout the third quarter of 2025. CEO David Kostman attributed the losses to “AI summaries and modifications in discovery” throughout the firm’s November 6, 2025 earnings name.

The Teads disclosure offered one of many first quantified assessments from an promoting platform with direct visibility into writer site visitors patterns. The corporate’s place serving ads throughout quite a few writer web sites offers it complete information about mixture site visitors traits that particular person publishers can’t observe.

Site visitors diversification has turn into important for publishers searching for to scale back dependence on search referrals. Advertising groups more and more discover various channels together with paid social media, native promoting, e-mail advertising, and direct viewers relationships. The diversification efforts try and construct income streams much less susceptible to algorithmic modifications at giant platforms.

Some publishers have achieved site visitors recoveries after algorithm updates, demonstrating that algorithmic modifications stay important components alongside AI function implementation.

The interaction between AI options and conventional search algorithms creates complicated causation questions when publishers analyze site visitors modifications. Isolating AI influence from broader algorithmic shifts requires subtle analytical approaches that many publishers lack sources to implement.

Trying ahead

The “internet rot” phenomenon described within the Axios evaluation represents greater than momentary disruption. The mix of declining site visitors to older websites, insufficient upkeep incentives, and AI-mediated discovery creates structural modifications in how the web capabilities as an info ecosystem.

Publishers face stark decisions about useful resource allocation. Sustaining in depth archives of older content material requires ongoing funding with diminishing returns. Some publishers have begun eradicating or consolidating older materials to focus sources on present content material extra more likely to generate site visitors. These choices additional cut back the pool of accessible historic info.

The query of who pays for content material creation and upkeep in an AI-mediated internet stays unresolved. Present compensation preparations cowl small fractions of the publishing business. Scaling these techniques requires technical infrastructure, business requirements, and enterprise fashions that do not but exist at obligatory ranges.

The deterioration of older web sites impacts greater than industrial publishers. Authorities info, tutorial content material, civic sources, and cultural supplies all face comparable preservation challenges. The decay extends throughout the net’s institutional reminiscence, not simply industrial media properties.

Hyperlink rot compounds as time passes. Every damaged hyperlink makes content material much less invaluable and tougher to confirm. Customers lose belief in on-line info when sources regularly show inaccessible. AI coaching techniques battle to judge supply high quality when unique supplies have disappeared.

The publishing business’s push for a two-sided market faces scaling challenges that stop fast implementation. Whereas OpenAI and some different firms have established pilot applications, these efforts stay restricted in comparison with the scope of content material getting used for AI coaching. The hole between present compensation ranges and publishers’ income losses continues widening.

The five-year site visitors decline documented by Similarweb represents solely the start of those structural shifts. AI capabilities proceed advancing quickly. Person adoption of AI-powered discovery instruments accelerates. Publishers proceed adapting methods, however the basic rigidity between AI effectivity and writer enterprise fashions persists with out clear decision paths.

Timeline

Abstract

Who: Conventional web site publishers, content material creators, information organizations, and AI firms together with OpenAI compete for person consideration and navigate altering web economics documented by Similarweb and analyzed by Axios reporter Sara Fischer.

What: Site visitors to established web sites that existed 5 years in the past declined greater than 11% throughout that interval whereas total web utilization remained secure, creating “internet rot” the place older websites stick with damaged hyperlinks and outdated content material. Analysis discovered 25% of internet pages from 2013-2023 grew to become inaccessible, with 23% of reports pages and 21% of presidency pages containing damaged hyperlinks.

When: The five-year site visitors decline was measured from 2020 by way of November 2025, with the Axios evaluation printed January 3, 2026. Pew Analysis Middle documented internet web page inaccessibility by way of October 2023, whereas Teads reported accelerating pageview declines within the third quarter of 2025.

The place: The site visitors patterns have an effect on the worldwide internet ecosystem, with explicit influence on English-language publishers in the US and different developed markets the place AI-powered search options have launched first. The phenomenon extends throughout information websites, authorities pages, and normal internet content material.

Why: This issues for the advertising neighborhood as a result of it represents a basic restructuring of how info discovery works on-line. Publishers losing 34.5% of organic clicks to AI options face existential income challenges whereas older content material accumulates damaged hyperlinks and turns into unreliable for each human readers and AI coaching techniques. The shortage of scaled compensation fashions means publishers proceed dropping income sooner than new monetization approaches can change it. Advertising methods constructed round search site visitors and content material advertising should adapt to AI-mediated discovery that reduces web site visits whereas the net’s infrastructure slowly degrades by way of insufficient upkeep of older properties.