Decodo, a company that provides web-scraping tools, today published its third annual list of the internet's most-scraped websites, and ChatGPT and Perplexity appeared for the first time. This matters to marketers because it shows that businesses have started collecting data from AI chatbots the same way they once tracked search rankings, a practice now called generative engine optimization. The practical change is that seven of last year's ten most-scraped sites dropped off the list entirely, replaced by AI platforms, a workflow-automation tool, and two developer-focused sites, meaning the places companies look for competitive intelligence are shifting faster than in the report's two previous editions.

Decodo, a web data and proxy infrastructure company headquartered in Vilnius, Lithuania, today published the third edition of its annual Most Scraped Websites report, recording ChatGPT and Perplexity among the ten most-scraped destinations for the first time since the series began. The 2026 edition, which follows the 2025 report published on September 9 of last year, marks what the company describes as the largest turnover in the report's history: seven of the ten websites that appeared in the 2025 ranking have dropped out entirely, replaced by artificial intelligence answer engines, a regional search platform, an automation tool, and two developer-focused sites.

TikTok retained the top position for a second consecutive year, according to Decodo, while Google held second place. ChatGPT entered directly at third, and Perplexity followed at fifth. Naver, South Korea's dominant search engine, took fourth place. Make.com, a workflow-automation platform, claimed sixth. Target debuted at seventh. Amazon remained in the top ten at eighth, despite falling five positions from third place a year earlier. GitHub and Stack Overflow rounded out the list at ninth and tenth, respectively, marking the first appearance of developer-focused platforms in the report's three-year history.

The findings rest on anonymized usage data that Decodo compiled from its own customer base rather than from a fixed panel of external websites, according to the company's methodology notes. That distinction matters for how the numbers should be read: the ranking reflects where Decodo's paying customers directed scraping requests during 2026, not a comprehensive census of every automated request across the open web. Even so, the pattern the data shows lines up with independent measurements published elsewhere this year, including Cloudflare Radar figures showing automated traffic crossing 50 percent of web requests for the first time in June, as PPC Land reported at the time.

What changed between 2025 and 2026

Only three websites carried over from last year's ranking: TikTok, Google, and Amazon. Everything else on the 2026 list is new. The seven departures were YouTube, which had held fourth place; Walmart, previously fifth; Coupang, previously sixth; eBay, previously seventh; ScienceDirect, previously eighth; Crunchbase, previously ninth; and Airbnb, previously tenth.

"Two years ago, businesses scraped the web to see what people were searching for. Now they scrape AI platforms to see what people are being told. The use case for data collection is changing," said Vaidotas Juknys, chief executive officer at Decodo, according to the company's report.

The scale of that turnover stands out against the report's own history. In the 2025 edition, TikTok had jumped from outside the top ten to claim first place, a shift Decodo attributed at the time to a 321 percent surge in traffic and a broader move toward multimodal AI training data, as PPC Land covered when that report was released. That earlier reshuffling replaced six of the ten positions from the 2024 list. This year's seven-position turnover exceeds even that figure, according to Decodo's own framing of the data.

Decodo groups the ten entries into six categories: AI answer engines, search, social video, eCommerce, developer platforms, and automation tools. Three of those categories, developer platforms, AI answer engines, and automation tools, did not appear in either of the report's two prior editions, according to the company.

AI answer engines enter the ranking

ChatGPT's arrival at third place followed a rapid expansion in the platform's user base. OpenAI disclosed that ChatGPT had reached 900 million weekly active users in February 2026, a figure the company announced alongside a $110 billion funding round and a $50 billion strategic investment from Amazon, as PPC Land reported at the time. That 900 million figure, which Decodo's report cites directly, has itself since been superseded: OpenAI's David Dugan referenced a base of 1 billion weekly active users in a subsequent update this month, according to PPC Land's coverage of that disclosure, underscoring how quickly the underlying user numbers behind this ranking continue to move.

Decodo's report attributes the scraping demand for ChatGPT to four distinct business motivations. Product and marketing teams check how ChatGPT ranks their company against named competitors within a category, according to the report. Some track how the platform describes their pricing, features, or policies, since an inaccurate answer could shape a purchasing decision before a prospective customer visits the company's own website. A third group collects prompt-and-response pairs to train or fine-tune their own models on conversational data. A fourth tracks which sources ChatGPT cites most frequently, then works to get published on those same sources.

That fourth motivation describes what the industry has begun calling generative engine optimization, a practice PPC Land has tracked extensively as it has moved from a fringe concern to a contested discipline. Google's own Search Advocate, John Mueller, argued in comments made during a discussion on the social platform Bluesky that nothing special needs to be done specifically for generative AI answers in search, and that the practice is ultimately the same discipline as conventional search engine optimization, according to PPC Land's coverage of that exchange. That framing has not gone unchallenged. A survey of 45 academic studies on the topic, posted to the preprint server arXiv and covered separately by PPC Land, concluded that the strongest available evidence shows content already retrieved by a generative system can influence how it is cited, while far weaker evidence exists that any technique reliably improves whether a page gets retrieved in the first place, as PPC Land's reporting on that survey detailed.

Perplexity entered the ranking at fifth place. Decodo's report notes that the platform's answer format, which cites sources directly inside each response, distinguishes it from ChatGPT's more conversational output and makes it a specific target for citation tracking. The report lists the same four collection motivations for Perplexity as for ChatGPT: competitive ranking checks, source-citation tracking, query-and-response pair collection, and monitoring of how the platform describes company pricing and policies.

Perplexity's presence on a scraping-demand ranking carries a certain irony, given the company's own documented history as a subject of scraping disputes rather than solely a target of them. Cloudflare accused Perplexity of stealth crawling in August 2025, alleging the company used undeclared crawlers disguised as a standard CAPTCHA-passing browser to fetch pages that had explicitly blocked its declared crawler through robots.txt directives, as PPC Land reported at the time. Cloudflare's chief executive, Matthew Prince, compared the practice to the conduct of state-linked hackers rather than that of a reputable AI company, according to that same reporting. Reddit subsequently sued Perplexity and several data brokers over related conduct in October 2025, PPC Land reported, and CNN filed its own copyright complaint against the company in May 2026 alleging similar circumvention of robots.txt blocks, according to PPC Land's coverage of that filing. Decodo's report does not reference any of that litigation history; it treats Perplexity strictly as a data-collection target for its own customer base.

Search and eCommerce hold their ground, with one new entrant each

Google retained second place in the 2026 ranking. Decodo's report states that the company holds 89.2 percent of global search market share this year, a figure the report presents as unchanged despite the growing prominence of AI-generated summaries. The report frames the persistence of Google scraping demand around a specific shift in what marketers now need to track: ranking first on a target keyword used to be the objective, but AI Overviews now answer many queries before a searcher scrolls past the summary panel, meaning a first-place ranking no longer guarantees a click. Teams now scrape Google, according to the report, to capture both the classic organic results and the AI Overview content displayed above them, since that comparison is described as the only way to track how a brand actually appears on the page.

That framing aligns with a body of independent research PPC Land has tracked through the year. Ahrefs research examining 300,000 keywords, published in February 2026, found that AI Overviews correlate with a 58 percent reduction in click-through rates for pages ranked in the top organic position, according to PPC Land's reporting on that study. A subsequent randomized field experiment involving 1,065 desktop Chrome users, described by its authors as the first causal evidence on the question, found that AI Overviews cut outbound organic clicks by 39.8 percent and raised zero-click searches by 34.5 percent, while clicks on sponsored results remained flat, as PPC Land reported when that paper was published. Google has disputed the framing of that research without disputing the underlying arithmetic; Search Central documentation and public remarks from company executives have characterized AI Overview clicks as higher-quality traffic in which users spend more time on the destination site, according to PPC Land's coverage of that dispute.

Naver entered the ranking at fourth place, a debut Decodo's report attributes to the search engine's dominance in South Korea specifically. The report states that Naver holds 63.8 percent of South Korea's search market as of March 2026, with Google trailing at 28.7 percent in that same market. Decodo frames the entry as evidence that businesses expanding internationally cannot assume Google's global dominance extends to every regional market; teams now collect data from Naver to localize product listings, track pricing against Korean competitors, and monitor brand visibility on a platform where the dominant search engine is not the one most Western marketing teams default to tracking.

In eCommerce, Target debuted at seventh place. Decodo's report links the platform's appearance to the continued growth of Target Plus, the company's third-party marketplace, which the report states has added sellers every quarter. Retail teams track pricing and promotions against Target's private-label and national-brand assortments, according to the report, while others monitor Target Plus listings specifically to identify which third-party sellers are gaining shelf space. Inventory data, the report notes, helps competing teams spot stock-outs and restocks before rivals do.

Amazon remained in the top ten at eighth place, despite falling five positions from third in the 2025 edition. Decodo's report frames that decline not as evidence of reduced demand for eCommerce data overall, but as evidence that scraping demand has fragmented across a wider set of retail targets. Sellers continue to track pricing and Buy Box ownership against competing listings, according to the report, while others monitor customer reviews to catch product issues before they escalate into broader reputational problems.

Developer platforms and automation tools mark a structural shift

GitHub's appearance at ninth place represents the first time a developer-focused platform has entered Decodo's ranking. The report states that GitHub added 36 million new developers in 2025, its fastest growth rate on record, and now hosts more than 180 million developers across 630 million repositories. Code-generation AI models require real repositories to train on, according to the report, and GitHub remains the largest public source of that material.

"GitHub and Stack Overflow mark a new kind of entry for this list. Both sites hold code and technical answers instead of search results or product listings. Teams pull that content straight into their own AI models," said Gabriele Vitke, product marketing team lead at Decodo, according to the company's report.

That growth in scraping demand for GitHub sits alongside a separate and consequential ownership change PPC Land has covered this year. Microsoft absorbed GitHub into its CoreAI division in January 2026, ending the developer platform's operational independence within the company roughly eight years after Microsoft's original 2018 acquisition, as PPC Land reported at the time. That reorganization placed GitHub's code repositories under the same corporate structure that controls Microsoft's Azure AI infrastructure and model-development partnerships, a detail Decodo's scraping report does not address but which bears directly on how that training data now flows internally, separate from the external scraping the report measures.

Stack Overflow entered at tenth place under circumstances that Decodo's report describes as counterintuitive. The report states that the platform's monthly question volume fell 78 percent in December 2025 compared with the same month a year earlier, a decline the report attributes to developers switching to AI coding assistants and asking questions directly rather than posting to the forum. Despite that traffic collapse, Decodo's report argues that Stack Overflow's archive, which the report describes as holding 16 years of accumulated technical answers, retains commercial value because AI companies use that historical data to train coding models. Teams also track question and tag volume on the platform as an early indicator of shifting developer interest, according to the report, watching which programming languages are rising and which tools are falling out of favor.

Make.com's sixth-place entry breaks from the pattern set by every other website on the list, according to Decodo's own framing. Every other entry holds content of one kind or another, whether video, text, product listings, or code. Make.com, by contrast, holds workflows. The platform launched its own AI agents in 2025, according to the report, letting users construct goal-driven automations that reason through sequential steps rather than following fixed, pre-programmed rules. Teams scrape Make.com to monitor which integrations and platform connections the company supports, and to track scenario templates as a signal for which automated workflows are gaining adoption across the broader market. That use case, the report states plainly, did not exist in either of the report's prior two editions.

Six categories now define the field

Decodo's report organizes the ranking's ten entries into six distinct categories: AI answer engines, search, social video, eCommerce, developer platforms, and automation tools. AI answer engines and search together account for the largest share of scraping requests recorded across Decodo's customer base, according to the report, driven by the combined pull of generative-engine-optimization tracking and conventional search-engine-optimization monitoring. Developer and technical platforms form a new cluster of their own, the report states, distinct from the eCommerce and social-media categories that had defined the 2024 and 2025 editions. Automation tools represent the smallest of the six categories, represented on this year's list solely by Make.com.

What the report anticipates for 2027

Decodo's report offers several forward-looking projections about how the ranking might shift again next year, though these remain the company's own stated expectations rather than confirmed outcomes. Generative-engine-optimization tracking is likely to extend beyond ChatGPT and Perplexity, according to the report, with Google's Gemini and Microsoft's Copilot named as strong candidates for next year's list as their respective user bases continue to close the gap with ChatGPT's scale. That competitive dynamic has already begun to play out in measurable terms this year: Google's Gemini app crossed 900 million monthly active users in May 2026, a milestone that one industry analyst compared directly against ChatGPT's weekly figures despite the two platforms reporting fundamentally different engagement metrics, as PPC Land detailed in its coverage of that comparison.

Naver's entry, according to Decodo's report, may open the door to other regional search engines appearing in future editions, with China's Baidu named specifically. The report frames this as part of a broader pattern in which more businesses expanding into markets where Google is not the default search engine will need country-specific competitive intelligence, since a single global search strategy will not cover every regional market.

Make.com's appearance is unlikely to be an isolated event, according to the report, which names Zapier and n8n as two likely automation-platform candidates for the 2027 edition as agentic AI adoption continues to grow. Decodo's report also raises a more structural question about who performs the scraping itself. Currently, the report states, scraping activity is driven mostly by people and the tools they operate directly. By 2027, the report suggests, AI agents may begin pulling this data themselves, meaning automated workflows could trigger scraping requests without a person actively initiating the request in the loop.

"Every year this report gets less about websites and more about where decisions get made," said Juknys, according to the company's report. "In 2025 that was search. In 2026 it's AI answers and automation. Next year the agents might be doing the deciding instead of just reading the answers."

Why this matters for the marketing community

The turnover Decodo documents in this report is not an isolated data point; it lands within a broader pattern PPC Land has tracked across multiple independent measurements throughout 2026. Cloudflare Radar data showed automated traffic crossing 57.4 percent of tracked HTML web requests for the first time in a seven-day window ending June 5, 2026, with human visitors accounting for the remaining 42.6 percent, as PPC Land reported at the time. Within that automated majority, training-related crawlers accounted for 50.6 percent of the total, while search-related bots made up a comparatively small 10.7 percent share, according to that same reporting, a composition that indicates most of the web's machine traffic exists to build AI models rather than to index content for results pages that send visitors back to publishers.

That asymmetry between crawling volume and the traffic sent back to publishers has itself become a subject of independent measurement this year. Cloudflare's own attribution data, published in July 2026, documented crawl-to-referral ratios ranging from 118 crawls per referred visitor at the low end to nearly 50,000 at the high end, according to PPC Land's coverage of that dashboard's launch. Microsoft later built a comparable ranking of its own into its Clarity analytics product, comparing which AI operators scrape most against which refer the least traffic back, as PPC Land reported when that feature launched.

For advertisers and publishers specifically, the practical significance of Decodo's report lies less in the exact rank order of ten websites than in what the category shift represents. Three of the report's six categories, AI answer engines, developer platforms, and automation tools, did not exist as distinct groupings in either prior edition. That structural change mirrors what PPC Land's own reporting has documented across search-visibility research this year: an IAB-set measurement framework found that only 16 percent of brands actively track their AI visibility at all, even as ChatGPT's weekly active user base surpassed 900 million and Google's AI Overviews reached over 2.5 billion monthly users, according to PPC Land's reporting on that measurement gap. A separate Semrush survey of 481 marketers, also covered by PPC Land, found that 45 percent could not properly measure their brand's visibility in AI-generated answers, and only 9 percent said they could measure the full set of metrics that the accompanying research identified as relevant.

The 2026 ranking also documents, in its own limited way, a data point about measurement itself: the report's own methodology draws on Decodo's customer base rather than a comprehensive census of the open web, meaning the ranking reflects a commercial vendor's own book of business rather than an independent, neutral audit. Decodo sells proxy infrastructure and web-scraping application programming interfaces, placing the company itself on the collection side of the very dynamic its report documents, a positioning consistent with how the company has framed its commercial recommendations in prior editions of the same research, according to PPC Land's earlier reporting on Decodo's bot-traffic analysis.

Timeline

Summary

Who: Decodo, a web-scraping and proxy-infrastructure company headquartered in Vilnius, Lithuania, led by chief executive officer Vaidotas Juknys and product marketing team lead Gabriele Vitke.

What: The company published the third annual edition of its Most Scraped Websites report, finding that ChatGPT and Perplexity entered the top ten most-scraped destinations for the first time, alongside Naver, Make.com, Target, GitHub, and Stack Overflow, while seven of the ten websites from the 2025 ranking dropped out entirely.

When: The report was published today, September 23, 2026, based on Decodo's anonymized customer usage data covering scraping activity throughout 2026.

Where: The findings apply to Decodo's global customer base, with specific market data cited for South Korea's search landscape and general references to worldwide AI-platform adoption.

Why: The report documents a structural shift in what businesses scrape and why, moving from traditional search-ranking and product-pricing monitoring toward tracking how AI platforms cite, describe, and rank companies, a practice the industry has termed generative engine optimization, alongside new demand for developer-platform code and workflow-automation templates as AI-training and agentic-AI use cases expand.