A paywall is a technical and commercial barrier that stops a reader from seeing some or all of a website's content until they pay, most often through a recurring subscription. Merriam-Webster defines it as "a system that prevents Internet users from accessing certain web content without a paid subscription" and dates the word's first known use to 2004. Paywalls exist because advertising alone has rarely covered the cost of producing journalism online. A publisher that charges readers directly trades some reach for revenue that does not depend on ad prices, search referrals or the policies of a platform.

The news publishing sense leads here; app, streaming and academic uses are separated near the end.

How a paywall decides who sees what

On each page request, a paywall decides whether the visitor is entitled to the full article.

A hard paywall shows almost nothing to non-payers beyond a headline and perhaps a summary. The Wall Street Journal has run one in some form since 1996, and The Times in London adopted one in 2010.

A metered paywall grants a quota of free articles per period, typically a calendar month, and blocks access once the quota is used. The meter is usually counted with a first-party cookie or local storage, sometimes backed by a login. Clearing cookies can reset such a meter, one reason publishers ask readers to register.

A freemium model, also called premium or hybrid, keeps most content free and locks selected articles, often analysis, opinion or exclusive reporting, behind the wall regardless of how many pages a visitor has read.

A dynamic paywall replaces a fixed rule with a prediction. Each anonymous visitor receives a score estimating their likelihood to subscribe, and the system decides how many free articles to allow, or whether to show the wall at all. The Wall Street Journal's version scores non-subscribers on more than 60 signals, including device, clicks and location, according to a February 2018 account of a Nieman Lab report. Karl Wells, then the Journal's general manager for membership, said that with a reader's score, "I pretty much know how likely they will be to subscribe." Piano sold a commercial equivalent from 2019, starting from 76 metrics, according to Nieman Lab.

Implementation matters as much as the model. A server-side paywall withholds the full text until the subscription is confirmed. A client-side paywall sends the complete article to the browser and hides it with JavaScript or an overlay, which is cheaper to build but leaves the text in the page source. In August 2025 Google added paywall guidance to its JavaScript troubleshooting documentation warning that hiding delivered content "isn't a reliable way to limit access to the content."

What search engines are allowed to see

Publishers want paywalled articles indexed, but showing Googlebot text that human visitors cannot read resembles cloaking, which breaches Google's spam policies. Google resolves this through structured data. Publishers mark the article, in JSON-LD, with isAccessibleForFree set to false, the only required property, and can add a hasPart element of type WebPageElement whose cssSelector points to the class wrapping the locked section, such as .paywall. Google's documentation, last updated September 8, 2026, says the markup helps it "differentiate paywalled content from the practice of cloaking."

Google's crawling overview, published March 3, 2026, states that its crawlers never enter paywall or subscription content without permission, and that structured data is the means of granting it.

How much a publisher gives away to search visitors is covered by Google's Flexible Sampling guidance, which describes metering and lead-in, in which an article's first sentences appear above the wall. The version updated December 10, 2025 puts the optimal range for most daily news publishers at 6 to 10 free articles per user per month.

From the Journal to the meter

Early experiments were blunt. The New York Times put its opinion columnists behind TimesSelect in 2005 and scrapped it about two years later after columnists objected to shrinking audiences, according to NPR.

The Financial Times introduced a metered "third way" in October 2007, allowing 30 free articles a month, according to a Press Gazette timeline. The New York Times followed on March 28, 2011 with 20 free pages a month, then $15 to $35 a month depending on devices, according to NPR. Clicks from Google, Facebook and Twitter did not count against the limit, a deliberate leak to preserve traffic. The allowance fell to 10 in 2012.

Search shaped the economics. Google's First Click Free policy required subscription publishers to let search users read articles without paying. On December 1, 2009, amid Rupert Murdoch's threats to pull News Corp titles from Google, publishers were allowed to cap that access at five pages a day, according to The Register. Google ended the policy on October 2, 2017, replacing it with Flexible Sampling. Richard Gingras, Google's vice president of news, wrote that "10 articles per month is a good starting point."

By May 2019, 69% of daily newspapers in seven countries used some pay model, with freemium and metered designs at 33% each and hard paywalls "extremely rare," according to a Reuters Institute study of 212 news organisations.

Why it matters for marketers

A paywall changes what a publisher can sell to advertisers. Subscribers are logged-in, identified readers, a first-party audience addressable without third-party cookies, while fewer anonymous pageviews can mean fewer impressions for open-market buyers.

The scale is considerable. News Corp said in November 2025 that digital revenues made up 62% of the business. FIPP's Global Digital Subscription Snapshot, published April 23, 2026, counted 23.98 million title-level subscribers in the US and 7.07 million in the UK, with The New York Times Company alone at 12.21 million in the first quarter of 2026. Growth is uneven: the Washington Post fell 20% to 2 million.

Modern paywalls also trade subscription revenue against ad revenue explicitly. Reuters runs a dynamic paywall whose models weigh a user's behaviour, an article's conversion history and the ad revenue that user represents. Phil Andraos, Reuters' general manager for digital, said: "Our advertising business is doing better today than it was before launching the paywall."

Google now treats paid access as a display signal. Through Subscription Linking, readers connect a subscription to a Google account; links to those publications were highlighted in the Gemini app first, then AI Overviews and AI Mode, and since May 2026 carry a subscription label in AI Mode and AI Overviews.

Where paywalls fall short

The first limit is demand. The Reuters Institute's 2026 Digital News Report found that 17% of people across 20 countries paid for online news in the past year, against 18% in 2025, and described 10% to 20% as an apparent ceiling in most markets. Norway, at 40%, and Sweden, at 32%, are outliers; the UK sits at 10%. In the 2024 edition, 57% of respondents in those markets said they would never pay, according to Press Gazette.

The second is reach. When the BBC introduced a dynamic paywall for US readers in mid-2025, priced at $8.99 a month, US visits to bbc.com fell 15% month on month to 100.2 million in July, according to Similarweb data published by Press Gazette. Liz Reid, Google's head of Search, was blunt when asked about a publisher whose traffic fell after it added a paywall: "Yes, yes, that is what will happen if you charge", she said in June 2026.

The third is leakage, which artificial intelligence has made costlier. An investigation by Alex Reisner in The Atlantic found that Common Crawl, the non-profit archive used to train many AI models, captured complete paywalled articles because its scraper never runs the code that checks subscriptions; the News/Media Alliance took the dispute to Common Crawl in April 2026. In October 2025 OpenAI's Atlas and Perplexity's Comet browsers both retrieved a 9,000-word subscriber-only MIT Technology Review article, according to the Columbia Journalism Review.

Critics add a public-interest objection: if rigorous reporting sits behind subscriptions while free sources fill feeds, the quality of information people see splits along ability to pay.

Not the same as

Recent developments

The paywall's next frontier is machines rather than people. Cloudflare's pay per crawl, which entered private beta on July 1, 2025, lets sites charge AI crawlers per request using the long-dormant HTTP 402 "Payment Required" status code. Prebid.org has since mapped four AI payment models for publishers, priced from $0.001 a page, extending the logic of the wall from readers to software.

New products keep appearing. MS NOW launched a $7.99 monthly membership on September 9, 2026 without removing free content, and USA TODAY Sports put fantasy football tools behind a $39.99 annual fee in August. Both are freemium designs that keep news itself open. Advertisers are also being invited to pay the toll on readers' behalf: on October 8, 2026, Substack added formats that let brands pay to open paywalled posts.

Timeline

  • 1996 - The Wall Street Journal places its online edition behind a hard paywall.
  • 2004 - First known use of the word "paywall," according to Merriam-Webster.
  • 2005 - The New York Times launches TimesSelect, charging for opinion columns; it ends about two years later.
  • October 2007 - The Financial Times introduces a metered model with 30 free articles a month.
  • December 1, 2009 - Google lets publishers cap First Click Free access at five pages a day.
  • 2010 - The Times in London introduces a hard paywall.
  • March 28, 2011 - The New York Times launches its metered paywall with 20 free pages a month.
  • 2012 - The New York Times cuts its free allowance to 10 articles.
  • October 2, 2017 - Google ends First Click Free and introduces Flexible Sampling, suggesting 10 free articles a month.
  • February 2018 - The Wall Street Journal's dynamic paywall, scoring visitors on more than 60 signals, is reported.
  • May 2019 - The Reuters Institute finds 69% of daily newspapers in seven countries use a pay model.
  • August 2019 - Piano's propensity paywall is reported live for five clients.
  • June 26, 2025 - Google Ad Manager launches Offerwall.
  • June-July 2025 - The BBC introduces a dynamic paywall for US readers.
  • July 1, 2025 - Cloudflare opens pay per crawl in private beta.
  • August 2025 - Google adds JavaScript paywall guidance to its Search documentation.
  • October 30, 2025 - The Tow Center reports AI browsers reading paywalled articles.
  • November 2025 - The Atlantic reports Common Crawl's capture of paywalled articles.
  • December 2025 - Google announces subscription highlighting, starting in the Gemini app.
  • January 10, 2026 - AdSense machine-learning Offerwall metering takes effect.
  • March 3, 2026 - Google publishes its crawling overview stating crawlers do not enter paywalls without permission.
  • May 6, 2026 - Google adds subscription labels in AI Mode and AI Overviews.
  • June 16, 2026 - The Reuters Institute reports 17% paying for online news across 20 countries.
  • September 8, 2026 - Google updates its paywalled content structured data documentation.
  • September 9, 2026 - MS NOW launches its paid membership.

Summary

Who: News and magazine publishers such as The Wall Street Journal, The New York Times, the Financial Times, Reuters and the BBC operate paywalls, often using vendors such as Piano. Google sets the rules for how paywalled pages are indexed and sampled from search; readers, advertisers and AI companies sit on the other side.

What: A paywall withholds some or all digital content until a reader pays. The main models are hard, metered, freemium and dynamic, implemented either on the server or, less securely, in the browser.

When: The Wall Street Journal began charging in 1996, the word appeared by 2004, metering took hold with the FT in 2007 and The New York Times in 2011, and Google replaced First Click Free with Flexible Sampling in October 2017. Dynamic models spread from 2018.

Where: On publisher websites and apps worldwide, with the highest reader payment rates in Norway (40%) and Sweden (32%) as of June 2026, and in Google Search, where structured data tells crawlers which content is locked.

Why: Advertising has not covered the cost of journalism online. Paywalls give publishers direct revenue and identified audiences, at the price of reach, search traffic and, increasingly, disputes over AI systems that read what human non-subscribers cannot.