Francois Zaninotto, CEO of Marmelab, a web and mobile development studio based in Nancy, France, last week disclosed on LinkedIn that Google has stopped indexing pages on the company's website without issuing any crawling error, any Largest Contentful Paint warning, or any canonical URL conflict. The post, which accumulated 137 reactions and 39 comments within days, describes a situation that many smaller publishers and technical teams will recognise: a growing count of pages labelled "Crawled - currently not indexed" in Google Search Console, with no explanation attached and no appeal mechanism available.

"Google has silently stopped indexing pages on our website," Zaninotto wrote. "There is no crawling error reported, no LCP problem, no canonical URL hiccup, no 'written by AI' articles. They just decided, unilaterally, to drop arbitrary pages from their index."

The numbers he cites are stark. According to Zaninotto, incoming traffic from Google Search has dropped 50% in the past six months on the Marmelab website. The studio, which was founded in 2012, employs 22 people, lists more than 250 clients, and publishes a technical blog covering topics including AI architecture, shell tooling, and agile development - content it has maintained for over twelve years.

What "Crawled - currently not indexed" actually means

The status itself is not new. Google's documentation defines it as a state in which Google has successfully fetched the page but has chosen not to include it in the search index. Separately, a "Discovered - currently not indexed" status means Google is aware of the URL but has not yet crawled it at all. The distinction matters: crawled-but-not-indexed pages have been visited. Google made a decision after the visit.

PPC Land covered an explanation of this status from Google's Martin Splitt in August 2024, in which the search relations advocate described the situation as "not necessarily an error or a problem that requires immediate attention." Splitt noted that Google rarely indexes all content from a site, and that server performance issues and content quality are the two primary factors that can push pages into these limbo states. For sites with millions of pages, he said, the crawl stats report in Search Console is a useful diagnostic tool.

That official framing has not satisfied website operators facing real traffic losses. First-party data from an SEO tooling company cited in industry discussions suggests that 70-80% of pages carrying the "crawled - currently not indexed" label have, in fact, been indexed before - meaning the status often describes active removal from the index rather than a page still waiting its turn. The implication is significant: for a site like Marmelab, which has published consistently for more than a decade, the status may represent content that once ranked and has since been quietly dropped.

PPC Land reported in April 2026 on a parallel case involving a technical safety publisher who discovered that Google Search Console's "Request Indexing" tool imposes a daily quota, blocking manual re-indexing attempts after just five requests. Google's John Mueller stated in July 2025 that sites requiring frequent manual submissions may be exhibiting signals that Google's systems are "not convinced about the site overall" - a characterisation the affected publisher disputed, noting that Bing and DuckDuckGo were indexing the same content without friction.

Zaninotto's diagnosis: Google is hoarding the traffic

The Marmelab CEO goes further than a technical complaint. His post frames the indexing behaviour as part of a deliberate strategic shift. "We're not the only ones affected by this," he wrote. "Apparently, Google has decided that they no longer want to index the Internet. This appears to be in line with their recent announcement showing AI answers front and foremost in their user experience."

His conclusion is blunt: "Google doesn't pretend that they're sending back traffic anymore. Incoming traffic from Google search has dropped 50% in the past 6 months on our website. They just use the content they crawl to train their models, and they keep the traffic for themselves."

The observation aligns with a broader pattern documented across the industry. Google Web Search has dropped from sending 51% of traffic to news publishers in 2023 to just 27% in the fourth quarter of 2025, according to analysis of over 400 publishers worldwide by NewzDash. That research confirmed a fundamental redistribution of referral traffic away from organic search and toward Google's own Discover feed, which now accounts for 67.51% of Google's traffic to news organisations.

For smaller technical publishers and agency blogs - sites that do not qualify for Google's Discover feed and have no newsroom scale - the picture is different from news publishers, and arguably worse. They have no Discover fallback. Organic search is often the only meaningful channel Google controls for them. When those indexed pages contract, traffic falls and does not return.

The AI Overviews connection

Zaninotto draws a direct line between the indexing decisions and Google's AI search strategy. AI Overviews, which now serve more than two billion monthly users across more than 200 countries according to Google disclosures, synthesise answers directly on the results page. When a question can be answered without a click, the incentive to visit the source page diminishes. Research from Ahrefs examining 300,000 keywords found that AI-generated summaries reduce organic click-through rates by 58% compared to baseline measurements from December 2023, nearly doubling the 34.5% decline the company documented earlier in the same year.

UK businesses experienced an 86% collapse in website traffic growth following Google's deployment of AI search features, according to research released in October 2025 by digital marketing agency Tank. The study analyzed 800 companies across 16 sectors and found that average monthly organic traffic growth across all industries dropped to 3.7% from 26.3% the prior year - a fall of 22.6 percentage points. The hospitality sector recorded negative growth of 6.7%.

What Zaninotto describes at Marmelab sits at the intersection of two separate pressures: first, indexed pages disappearing without explanation, reducing the pool of pages that could theoretically rank; and second, even those pages that remain indexed generating fewer clicks as AI features intercept queries before users reach organic results. The effect compounds. Fewer indexed pages means fewer chances to appear. And for those that do appear, AI Overviews may absorb the click.

Google's own May 2026 announcement of five new outbound link features for AI Mode and AI Overviews signals awareness of the publisher traffic problem, but the structural tension between a system that synthesises answers and one that sends users to external websites remains unresolved. The update was framed as increasing link visibility rather than reversing the underlying traffic dynamics.

Marmelab's profile and the stakes involved

Marmelab is not an anonymous content farm or a low-quality link-building operation. Founded in 2012 and headquartered at 4 rue Girardet, 54000 Nancy, the studio lists specialties including Digital Innovation, Web development, Lean Startup, SCRUM, react.js, Symfony, Node.js, and AI. It maintains more than 80,000 stars on GitHub across its open-source projects, according to company materials. Its flagship product, React-admin - a low-code framework for building administration interfaces, dashboards, and B2B applications - powers around 3,000 new websites every month.

The studio has worked with clients including Qwant, the privacy-focused French search engine, where it built a comprehensive evaluation framework for generative AI responses. It has maintained a long-term infrastructure partnership with ARTE, the Franco-German public broadcaster. The company describes its blog as sharing learnings from customer projects, open-source products, and R&D, and notes it has done so weekly for over twelve years. Recent posts include "Agentic Software Factories: The Future Of Programming?" published May 22, 2026, and "Getting More Out of Claude Code in the Terminal" published May 12, 2026.

Zaninotto graduated from the Ecole Nationale Superieure des Mines de Nancy with an engineering degree in computer science in 1997. He has held roles including manager at Sensio from September 2003 to August 2007, and has been associated with the GreenFrame project - a tool that measures CO2 emissions of web applications - since February 2020. He is also the founder of react-admin.

The company's LinkedIn profile describes it as an innovation-focused web development agency with 11 to 50 employees and 21 associated members, founded in 2012. Its values include environmental responsibility, social impact through open-source contributions, and what it describes as "TechForGood" project selection.

The reaction: some sympathy, much technical pushback

The LinkedIn post generated a wide range of responses, and not all were sympathetic to Zaninotto's framing. Several SEO practitioners noted that "Crawled - currently not indexed" is not a new Google Search Console category - it has existed for years, and its presence does not automatically indicate intentional suppression.

One commenter pointed to a sitemap configuration issue at marmelab.com as a potential contributing factor, noting that the sitemap lives at a custom location - marmelab.com/sitemap-index.xml rather than the conventional /sitemap.xml - and indirectly references a sitemap0.xml file. Whether that configuration affects how Google prioritises the site's URLs is not confirmed.

Another commenter drew attention to what appears to be a redirect pattern, noting they had observed "huge increases in crawled - not indexed - but huge declines in page with redirect" that net out to stable impressions in some accounts. The suggestion is that some of what registers as lost indexing may actually reflect URL consolidation rather than content suppression.

A third response pointed to the blog's content structure rather than technical configuration. One practitioner noted that the blog "is very strong at showing technical expertise, but less structured around the buying decisions a potential client needs to make." The argument is that content built around user decisions - questions such as whether to build in-house or work with an agency, when to refactor versus rebuild, or when a freelancer is sufficient - may be indexed more readily than general-purpose technical posts. Google, the same commenter argued, appears to index decision-oriented content faster.

Zaninotto acknowledged the uncertainty but held to his central concern: "I find this concerning." His proposed solution is direct. "So if you want to hear from us, our good old RSS feed is probably your best choice." He linked to Marmelab's feed directly in the post.

RSS and direct distribution as a structural response

The recommendation to use an RSS feed is not nostalgia. It reflects a structural argument about dependency. When an organisation's content is visible to its audience only because a third-party platform has decided to index or surface it, the platform controls the relationship. RSS inverts that dependency: subscribers receive content directly, without algorithmic intermediation.

Small publishers have lost 60% of their search traffic over two years as AI reshapes the web, according to Chartbeat data covering thousands of publisher websites globally. The same research found that chatbots still account for less than 1% of all publisher page view referrals, meaning AI-driven traffic has not replaced what search has taken away. The math does not currently close for publishers who relied on organic discovery.

Flipboard moved in February 2026 to formally onboard dozens of independent publishers and ingest hundreds of RSS feeds from smaller quality sites, framing the move as an editorial commitment to the open web. Reddit opened its publisher toolkit to all users in March 2026, including RSS auto-import functionality that lets outlets connect their content feeds so new articles become immediately shareable on the platform. These moves suggest that RSS - as a distribution layer - is being quietly revalidated by platforms that see structural opportunity in the vacuum Google's changes are creating.

Whether or not Marmelab's specific indexing decline is fully attributable to the dynamics Zaninotto describes, the case he makes resonates precisely because it is concrete and verifiable. Traffic from Google Search fell 50% in six months. The "Crawled - currently not indexed" count in Search Console is growing. No recourse is available. For a 22-person engineering studio with a 12-year publishing record, those are not abstract concerns.

Timeline

Summary

Who: Francois Zaninotto, CEO of Marmelab - a 22-person web, mobile, and AI development studio founded in 2012 and based in Nancy, France. Marmelab is also the creator and maintainer of React-admin, an open-source low-code framework used to power approximately 3,000 new websites per month.

What: Zaninotto reported on LinkedIn that Google has silently stopped indexing pages on the Marmelab website, with the "Crawled - currently not indexed" count in Google Search Console growing without explanation and without any available appeal mechanism. He disclosed that organic traffic from Google Search has dropped 50% in the past six months. He attributed the behaviour to a strategic shift by Google toward AI-generated answers, arguing that Google crawls content to train its models but withholds the resulting traffic.

When: The LinkedIn post was published on May 21, 2026, describing a trend Zaninotto says has developed over the past six months. The broader context of declining publisher traffic from Google extends back at least to 2023.

Where: Marmelab operates from Nancy in the Grand Est region of France, with its website at marmelab.com. The indexing issues affect the company's publicly accessible technical blog, which publishes articles on AI, development methodology, and open-source tooling.

Why: The case matters because it illustrates, through a specific and verifiable example, what happens when a smaller technical publisher with a long publication record and high-quality open-source credentials loses organic search visibility without any technical explanation from Google. It raises questions about whether the "Crawled - currently not indexed" status in Search Console is functioning as a transparent quality signal or as an opaque editorial gatekeeping mechanism - and about whether the content Google crawls for AI training should generate reciprocal traffic benefits for the publishers that produced it.

Share this article
The link has been copied!