<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Surfil - SEO blog</title>
    <link>https://surfil.com</link>
    <description>Reports, research write-ups and case studies, each built on a cited primary source.</description>
    <language>en</language>
    <atom:link href="https://surfil.com/blog/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Ahrefs: ChatGPT searched the web in 14.4% of 66.6M responses</title>
      <link>https://surfil.com/blog/chatgpt-query-fan-out-ahrefs-15m-queries</link>
      <guid isPermaLink="true">https://surfil.com/blog/chatgpt-query-fan-out-ahrefs-15m-queries</guid>
      <pubDate>Mon, 12 Oct 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs studied 66.6M ChatGPT responses and 15M fan-out queries: 14.4% searched the web, 6.8% of fan-outs used site:. Each figure with its base and limits.</description>
      <content:encoded><![CDATA[Ahrefs analyzed 66.6 million logged-out responses from ChatGPT's free GPT-5.6 Luna model and their 15 million fan-out queries, from August and September 2026. It reports that 14.4% of responses triggered a web search, with about 1.6 fan-out queries each, and that 6.8% of fan-outs used a site: operator, mostly aimed at brand sites.

- Ahrefs reports that only 14.4% of 66.6 million responses required a web search, and those that did ran about 1.6 fan-out queries on average.
- Ahrefs says 6.8% of fan-out queries used a site: operator, and 75.9% of the most common site: searches pointed at a brand site.
- Per Ahrefs, the word "official" appeared in 13.1% of English fan-out queries in September, up from 10.1% in August, and "price" rose from 6.04% to 8.88%.
- Ahrefs reports from separate Brand Radar data that Reddit appeared in about 1 in 43 ChatGPT answers by late September, against about 1 in 5 in March.
- This is a vendor study of one model over two months, and the post does not describe how the responses were collected.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>How long Google takes to crawl, index and update results</title>
      <link>https://surfil.com/blog/google-crawling-indexing-serving-timelines-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-crawling-indexing-serving-timelines-2026</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>All 23 typical and slowest times Gary Illyes showed at Search Central Live Barcelona, as recapped by ROAST, plus caveats and Google&apos;s own docs.</description>
      <content:encoded><![CDATA[Google's Gary Illyes showed typical and slowest times for 23 crawling, indexing and serving processes at Search Central Live Deep Dive in Barcelona on 2 October 2026. Per the ROAST recap, a typical site move takes 1-3 months, a title update 1-2 days and core update recovery 3-6 months. The figures are Google's internal analysis, not published documentation.

- Per the ROAST recap, Gary Illyes showed a typical and a slowest time for 23 processes across crawling, indexing and serving, based on Google's internal analysis.
- The recap lists a typical site move at 1-3 months and core update recovery at 3-6 months, with slowest cases of 6 months to 1 year or more.
- Five of the 23 slowest times include the word never, and three of those add quality in brackets, so a fast technical fix cannot rescue a page Google does not value.
- Search Engine Journal notes that typical is undefined and that sample size and measurement period are not reported, so the figures are ranges, not deadlines.
- Illyes cautioned that the processes are linked, so a delay in crawling carries into indexing and serving.]]></content:encoded>
      <category>reports</category>
      <category>indexing</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Ahrefs survey: 93% of 301 marketers use AI for content</title>
      <link>https://surfil.com/blog/ai-content-usage-survey-ahrefs-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/ai-content-usage-survey-ahrefs-2026</guid>
      <pubDate>Wed, 30 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs&apos; 2026 survey of 301 marketers: 93% use AI for content, 85% of them work faster, 66% trust it only after review. Each figure with its base.</description>
      <content:encoded><![CDATA[Ahrefs surveyed 301 marketers and reports that 280 of them (93%) use AI to create or help create content, against 87% last year, when more people responded. Among AI users, 85% say they produce content faster, and 7% say more than 75% of what they publish is unedited AI output.

- Ahrefs reports that 280 of 301 respondents (93%) use AI to create or help create content, against 87% in last year's survey.
- Ahrefs reports that 85% of AI users produce content faster, and that 66% of marketers trust AI content only if a human reviews it first.
- Per Ahrefs, 7% of AI users say more than 75% of what they publish is unedited AI content, up from 4% in 2025, while 40% say none of it is.
- Ahrefs says 28.6% of AI users now use AI to check AI-assisted content, up from 14%, and 22% disclose their AI use, up from 16%.
- The sample is 301 self-reported answers, and Ahrefs says last year's survey had more respondents, so only high-level comparisons hold.]]></content:encoded>
      <category>reports</category>
      <category>content-strategy</category>
      <category>ai-search</category>
    </item>
    <item>
      <title>SearchPilot: brand links on PDPs lift PLP traffic by 18%</title>
      <link>https://surfil.com/blog/pdp-brand-link-plp-traffic-searchpilot</link>
      <guid isPermaLink="true">https://surfil.com/blog/pdp-brand-link-plp-traffic-searchpilot</guid>
      <pubDate>Wed, 30 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>SearchPilot split-tested turning unlinked brand text on product detail pages into links to brand category pages, finding an 18% uplift in organic traffic.</description>
      <content:encoded><![CDATA[SearchPilot split-tested converting plain-text brand names on ecommerce product detail pages into links to brand-level product listing pages. The change produced an 18% uplift in organic traffic to those listing pages with no negative impact on product page visits, isolating internal linking as the single driver.

- Converting plain-text brand labels on product detail pages into internal links produced an 18% uplift in organic traffic to brand category pages.
- The controlled split test confirmed no negative ranking or traffic impact on the product detail pages where the links were added.
- SearchPilot evaluated the test at a 95% credible interval across an enterprise ecommerce store over its testing window.
- The variant altered only the link attribute without changing the visible page layout or typography.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
      <category>ecommerce-seo</category>
    </item>
    <item>
      <title>What to do after a Google core update: Google&apos;s recovery steps</title>
      <link>https://surfil.com/blog/what-to-do-after-google-core-update</link>
      <guid isPermaLink="true">https://surfil.com/blog/what-to-do-after-google-core-update</guid>
      <pubDate>Sat, 26 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Google&apos;s own recovery steps after a core update: wait a week, compare the right dates, size the drop, self-assess and avoid quick fixes.</description>
      <content:encoded><![CDATA[After a Google core update, wait until the rollout ends plus at least a full week, then compare a week after with a week before it began. Small drops need no drastic action; large, sustained drops call for a site-wide self-assessment against Google's helpful content questions, with no quick fixes. Recovery can take until a later core update.

- Google recommends waiting at least a full week after a core update completes before analyzing your site in Search Console.
- The comparison Google suggests is a week after the rollout against a week before it started, never the rollout window itself.
- Google treats a fall from position 2 to 4 as needing no drastic action and a fall from 4 to 29 as needing a deeper assessment.
- For a sustained, large drop, Google points to its helpful content self-assessment questions and advises against quick-fix changes.
- Google says improvements can take several months to register and may not show until the next core update.]]></content:encoded>
      <category>reports</category>
      <category>core-updates</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>How to tell which Google update hit your site: a 5-step method</title>
      <link>https://surfil.com/blog/which-google-update-hit-my-site</link>
      <guid isPermaLink="true">https://surfil.com/blog/which-google-update-hit-my-site</guid>
      <pubDate>Sat, 26 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Match a Search Console drop to confirmed Google update dates using 7-day windows, search types and a manual actions check, then rule out overlaps.</description>
      <content:encoded><![CDATA[To tell which Google update hit your site, find the first day of a sustained drop in Search Console and match it to the Search Status Dashboard dates for each confirmed update. Compare 7-day windows, split search types, check Manual actions first, and rule out overlapping rollouts, seasonality and your own site changes.

- Google's Search Status Dashboard is the only first-hand record of when a ranking update started and finished, so every attribution starts from its dates.
- Across the 39 completed Google-confirmed updates since November 2021, the median rollout ran 14 days, so a drop can surface well after an update's start date.
- Comparing whole 7-day windows cancels the weekday cycle in search traffic, which a single day against another single day does not.
- A manual action shows up in the Search Console Manual actions report; check it before blaming any algorithm update.
- 10 of the 39 completed updates overlapped another rollout, so a drop inside an overlap cannot be pinned on one update from dates alone.]]></content:encoded>
      <category>reports</category>
      <category>core-updates</category>
      <category>search-console</category>
    </item>
    <item>
      <title>Structured Data Markup Helper: is Google&apos;s tool still available?</title>
      <link>https://surfil.com/blog/google-structured-data-markup-helper</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-structured-data-markup-helper</guid>
      <pubDate>Sat, 26 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>What Google&apos;s Structured Data Markup Helper does, whether it still works, and what Google recommends for writing and testing structured data.</description>
      <content:encoded><![CDATA[Yes, Google's Structured Data Markup Helper is still live and Google has published no deprecation notice for it. It lets you tag parts of a page and generates JSON-LD or Microdata. Google's current documentation, though, centers on hand-written JSON-LD checked with the Rich Results Test, and treats the helper as a basic starting aid.

- Google's Structured Data Markup Helper page loaded normally on 2026-09-26 and none of the Google pages reviewed announce its deprecation.
- The helper generates markup in two formats, JSON-LD, which Google's help page labels as preferred by Google, and Microdata.
- Google's own documentation describes the helper as a way to generate basic structured data, not a complete implementation.
- Google recommends JSON-LD for structured data and names the Rich Results Test as the tool for validating it.
- Data Highlighter, the no-code alternative in Search Console, can break if a page layout changes, according to Google.]]></content:encoded>
      <category>research</category>
      <category>structured-data</category>
    </item>
    <item>
      <title>SearchPilot: adding &apos;The Best&apos; to title tags lifted traffic 10%</title>
      <link>https://surfil.com/blog/the-best-title-tag-ab-test-searchpilot</link>
      <guid isPermaLink="true">https://surfil.com/blog/the-best-title-tag-ab-test-searchpilot</guid>
      <pubDate>Sat, 26 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>SearchPilot split-tested putting &apos;The Best&apos; at the start of an editorial site&apos;s title tags. The variant pages gained 10% more organic traffic.</description>
      <content:encoded><![CDATA[Adding "The Best" to the start of title tags lifted organic traffic 10% in a SearchPilot split test on an editorial site. SearchPilot estimated the change was worth about 11,000 extra organic sessions a month across the site section, and says Google kept showing the new titles in results. Test dates and confidence interval are not published.

- SearchPilot's split test found that adding "The Best" to the start of an editorial site's title tags produced a 10% uplift in organic traffic.
- SearchPilot estimated that deploying the new titles across the tested site section would add about 11,000 organic sessions per month.
- The test ran after Google began rewriting most title tags in results, yet SearchPilot reports Google still displayed the variant titles.
- SearchPilot cannot say whether the gain came from a higher click-through rate or from ranking for a different set of queries.
- The published case study does not state the test dates, the number of pages tested or the exact confidence interval.]]></content:encoded>
      <category>studies</category>
      <category>serp-features</category>
    </item>
    <item>
      <title>Pages with no backlinks and Google traffic: Ahrefs&apos; 14B-page data</title>
      <link>https://surfil.com/blog/backlinkless-pages-search-traffic-ahrefs-14b</link>
      <guid isPermaLink="true">https://surfil.com/blog/backlinkless-pages-search-traffic-ahrefs-14b</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs found ~20 million indexed pages with no referring domains, and only 2,997 of them earn over 1K monthly search visits.</description>
      <content:encoded><![CDATA[Pages with no backlinks almost never draw meaningful Google traffic. Ahrefs' index holds about 20 million pages with no referring domains, yet only 2,997 of them get more than 1K search visits a month, roughly 1 in every 6,671. Across its whole 14-billion-page database, 96.55% of pages get zero organic traffic.

- Ahrefs studied the full Content Explorer database, around 14 billion pages, to see how many earn organic search traffic.
- 96.55% of all pages in that index get zero traffic from Google, and 1.94% get between one and ten monthly visits.
- The index contains about 20 million pages with no referring domains, and only 2,997 of them exceed 1K monthly search visits.
- That works out to roughly 1 in every 6,671 pages with no backlinks reaching 1K visits a month.
- Ahrefs says most no-backlink pages that do rank target low-competition topics, and its two examples sit on high Domain Rating sites.]]></content:encoded>
      <category>reports</category>
      <category>link-building</category>
    </item>
    <item>
      <title>Search engine market share, August 2026: Statcounter</title>
      <link>https://surfil.com/blog/statcounter-search-engine-share-august-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/statcounter-search-engine-share-august-2026</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Statcounter puts Google at 91.02% of worldwide search in August 2026, ahead of Bing at 4.55%, against 90.37% across the prior twelve months.</description>
      <content:encoded><![CDATA[Statcounter Global Stats shows Google handled 91.02% of worldwide search engine page views in August 2026, with Bing second at 4.55%. Across the twelve months from August 2025 to August 2026, Google averaged 90.37% and Bing 4.49%. Yahoo!, Yandex, DuckDuckGo and Baidu each stayed at or below 1.5% in both windows.

- Statcounter recorded Google at 91.02% of worldwide search in August 2026, against 90.37% over the twelve months to August 2026.
- Bing ranked second at 4.55% in August 2026, compared with 4.49% across the twelve-month window.
- Yandex fell from 1.5% over twelve months to 1.01% in August 2026, and Yahoo! from 1.38% to 1.24%.
- DuckDuckGo held near 0.7% and Baidu at 0.62% in both windows, so neither moved in the latest month.
- Statcounter measures page views on its tracked sites, so the shares describe traffic referred to those sites, not queries.]]></content:encoded>
      <category>reports</category>
      <category>technical-seo</category>
      <category>analytics</category>
    </item>
    <item>
      <title>Median home page weight 2025: 2.86 MB desktop, 2.56 MB mobile</title>
      <link>https://surfil.com/blog/web-almanac-2025-median-home-page-weight-by-type</link>
      <guid isPermaLink="true">https://surfil.com/blog/web-almanac-2025-median-home-page-weight-by-type</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>The 2025 Web Almanac puts the median home page at 2.86 MB on desktop and 2.56 MB on mobile, with images and JavaScript the two largest parts.</description>
      <content:encoded><![CDATA[HTTP Archive's 2025 Web Almanac found the median home page weighs 2.86 MB on desktop and 2.56 MB on mobile. Images are the largest share at 1,058 KB on desktop and 911 KB on mobile, followed by JavaScript at 697 KB and 632 KB, with fonts, CSS and HTML making up the rest of the transfer.

- The median home page in the July 2025 crawl was 2,862 KB on desktop and 2,559 KB on mobile.
- Images were the heaviest resource type, at 1,058 KB on the median desktop home page and 911 KB on mobile.
- JavaScript came second at 697 KB on the median desktop home page and 632 KB on the median mobile home page.
- Inner pages carry far fewer image bytes than home pages, at 442 KB on desktop and 354 KB on mobile.
- HTML is the smallest measured slice, at 22 KB on both the median desktop and mobile home page.]]></content:encoded>
      <category>reports</category>
      <category>page-speed</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>YouTube citation share across AI search platforms, BrightEdge</title>
      <link>https://surfil.com/blog/youtube-citation-share-across-ai-platforms-brightedge</link>
      <guid isPermaLink="true">https://surfil.com/blog/youtube-citation-share-across-ai-platforms-brightedge</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>BrightEdge measured YouTube&apos;s citation share in Google AI Overviews, AI Mode, Perplexity and ChatGPT from May 2024 to September 2025. Here are its figures.</description>
      <content:encoded><![CDATA[YouTube is the most cited video platform in AI search, averaging a 20% citation share across platforms, according to BrightEdge. Its share is 29.5% in Google AI Overviews, 16.6% in AI Mode, 9.7% in Perplexity and 0.2% in ChatGPT. No other video platform registers above 0.1%. The figures are BrightEdge's own.

- BrightEdge reports that YouTube averages a 20% citation share across the AI platforms it tracked, from May 2024 to September 2025.
- In Google AI Overviews, YouTube was the number one cited domain at 29.5%, ahead of Mayo Clinic at 12.5%.
- YouTube's share varies sharply by platform, from 29.5% in AI Overviews down to 0.2% in ChatGPT.
- BrightEdge found Vimeo and TikTok at 0.1% each, while Dailymotion, Twitch and all other video platforms sat at 0%.
- YouTube typically ranked in positions 3 to 10 within citations, which BrightEdge reads as video serving as supporting evidence.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>Ahrefs 600,000-page study: AI content and Google rankings</title>
      <link>https://surfil.com/blog/ahrefs-600k-pages-ai-content-ranking-correlation</link>
      <guid isPermaLink="true">https://surfil.com/blog/ahrefs-600k-pages-ai-content-ranking-correlation</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs checked the top 20 Google results for 100,000 keywords with its AI detector and found almost no link between AI share and ranking position.</description>
      <content:encoded><![CDATA[Ahrefs analyzed the top 20 Google results for 100,000 random keywords, 600,000 pages in total, with its own AI detector. It found 4.6% pure AI pages, 13.5% pure human pages and 81.9% mixed pages, and a 0.011 correlation between AI share and ranking position, effectively zero.

- Ahrefs classified 4.6% of the 600,000 sampled top-20 pages as pure AI, 13.5% as pure human and 81.9% as a mix of both.
- The correlation between a page's detected AI percentage and its ranking position was 0.011, which Ahrefs describes as effectively zero.
- Pages ranking number one showed slightly less AI-generated content than pages ranking lower in the top 20.
- All figures come from Ahrefs' own detector, which the authors say is probabilistic and can produce false positives.]]></content:encoded>
      <category>research</category>
      <category>content-strategy</category>
      <category>core-updates</category>
    </item>
    <item>
      <title>Title tags, H1s and content grade: what Backlinko found</title>
      <link>https://surfil.com/blog/backlinko-title-h1-clearscope-ranking-correlation</link>
      <guid isPermaLink="true">https://surfil.com/blog/backlinko-title-h1-clearscope-ranking-correlation</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Backlinko&apos;s 11.8M-result study found exact-match titles and H1s common on page one but unrelated to rank, while Clearscope content grade tracked position.</description>
      <content:encoded><![CDATA[Backlinko's 11.8 million result study found that keyword-matched title tags and H1s are common on Google's first page but do not correlate with higher rankings. A Clearscope Content Grade analysis on a subset did track position, roughly one rank per grade point. The authors say correlation cannot show why.

- Backlinko found that 65 to 85 percent of the keyword appears in most first page title tags, yet an optimized title did not correlate with higher rankings.
- Most first page results had a keyword-matched H1 tag, but keyword-matched H1s had essentially no relationship with higher rankings.
- Within the top 30 results of a Clearscope-analyzed subset, one extra Content Grade point approximated one better ranking position.
- The authors say it is impossible to tell from their data whether Google prefers comprehensive content or users are simply more satisfied by it.
- The public article does not disclose the size of the Clearscope subset or the grade scale, so that finding cannot be independently checked from it.]]></content:encoded>
      <category>research</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>Does Google use Analytics bounce rate to rank pages?</title>
      <link>https://surfil.com/blog/google-analytics-bounce-rate-not-ranking-signal</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-analytics-bounce-rate-not-ranking-signal</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>John Mueller said on record that Google does not use Analytics bounce rate for ranking. What that statement covers, and what it leaves open.</description>
      <content:encoded><![CDATA[Google does not use Google Analytics bounce rate to rank websites, according to John Mueller. In a webmaster hangout reported in June 2020 he called the idea a misconception. The statement covers Analytics data only, not click behavior inside Google Search itself, and it is one on-record answer, not documentation of how ranking works.

- John Mueller said Google is not looking at the Analytics bounce rate when ranking websites, and called the idea a misconception.
- The statement was made in a public webmaster hangout and reported by Search Engine Roundtable on June 16, 2020.
- The statement is about Google Analytics bounce rate, a site-side metric, and does not describe click behavior measured inside Google Search.
- Search Engine Roundtable notes Google made similar denials in 2017, 2018, 2019 and 2020, so the position is consistent rather than new.
- A spoken denial is not documentation, so treat it as a strong statement of intent and not as proof of every ranking system's inputs.]]></content:encoded>
      <category>research</category>
      <category>search-console</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Mobile vs desktop parity in the 2024 Web Almanac SEO chapter</title>
      <link>https://surfil.com/blog/mobile-desktop-parity-web-almanac-2024-seo</link>
      <guid isPermaLink="true">https://surfil.com/blog/mobile-desktop-parity-web-almanac-2024-seo</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>HTTP Archive&apos;s 2024 Web Almanac SEO chapter compared mobile and desktop crawls. Titles, canonicals and word counts nearly match; a few gaps remain.</description>
      <content:encoded><![CDATA[Do mobile and desktop pages match closely enough for mobile-first indexing? Mostly yes. The 2024 Web Almanac SEO chapter found near-identical title, canonical and hreflang usage across both crawls, while the median home page rendered 364 words on mobile against 400 on desktop, a gap that narrowed from 55 words in 2022 to 36.

- The 2024 Web Almanac found 98.0% of desktop pages and 98.2% of mobile pages carried a title tag, so title presence is effectively at parity.
- Canonical conflicts, such as header and rendered HTML disagreeing, hit 0.8% of both mobile and desktop pages, double the 0.4% of 2022.
- The median rendered home page held 364 words on mobile and 400 on desktop, and that gap narrowed from 55 words in 2022 to 36 words in 2024.
- The chapter attributes the narrower word-count gap to Google's completed move to mobile-first indexing, but describes that as a reasonable conclusion, not a measured cause.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
      <category>indexing</category>
    </item>
    <item>
      <title>Subdomain vs subfolder: what Ahrefs found in Mention.com</title>
      <link>https://surfil.com/blog/subdomain-vs-subfolder-mention-com-ahrefs-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/subdomain-vs-subfolder-mention-com-ahrefs-case-study</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs checked Mention.com&apos;s January 2016 move from subdomain to subfolder in Site Explorer. Traffic shifted, and only new content produced growth.</description>
      <content:encoded><![CDATA[Does moving a blog from a subdomain to a subfolder raise organic traffic? Not by itself, according to an Ahrefs case study of Mention.com. After its January 2016 migration, traffic shifted location rather than growing, and gains came only once new content was added. Google's own statements call both structures workable.

- Ahrefs examined Mention.com's January 2016 subdomain to subfolder migration in Site Explorer and saw traffic move to the new location without a clear increase.
- Growth on Mention.com appeared only after the site started adding additional content, so the URL change alone did not explain it.
- Patrick Stox argues that published subfolder success stories usually involved other changes, such as extra internal linking or merging several properties into one.
- Google's quoted position from Matt Cutts and John Mueller is that either subdomains or subdirectories are fine for search.
- This is a single observational case with no control group, so it challenges the subfolder-always-wins claim but does not settle it.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Backlinko content relaunch: one post, organic traffic up 260.7%</title>
      <link>https://surfil.com/blog/backlinko-content-relaunch-white-hat-seo-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/backlinko-content-relaunch-white-hat-seo-case-study</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Brian Dean reports a 14-day content relaunch of Backlinko&apos;s White Hat SEO post: organic traffic up 260.7% and the target keyword from 7th to 4th.</description>
      <content:encoded><![CDATA[Backlinko founder Brian Dean reports that relaunching his old White Hat SEO post lifted its organic search traffic 260.7% in 14 days and moved its target keyword from 7th to 4th position. The relaunch combined updated content, email outreach and a new publish date. The figures are self-reported and come from a single page.

- Brian Dean reports that one relaunched Backlinko post gained 260.7% in organic search traffic within 14 days.
- The post's target keyword, white hat SEO, is reported to have moved from the 7th spot to position 4.
- The relaunch improved an existing post and involved no new article, plus outreach, a newsletter send and a new publish date.
- Dean says the newsletter send drove about 7,000 clicks against 111 from a tweet, so promotion was a large part of the result.
- The case covers one page with no absolute traffic counts, so treat it as an illustration rather than a benchmark.]]></content:encoded>
      <category>studies</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>BODYBAR Pilates: unique studio pages lifted organic sessions 116%</title>
      <link>https://surfil.com/blog/multi-location-studio-pages-bodybar-oneupweb</link>
      <guid isPermaLink="true">https://surfil.com/blog/multi-location-studio-pages-bodybar-oneupweb</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Oneupweb rewrote 17 duplicate franchise studio pages for BODYBAR Pilates. Organic sessions rose 115.9% and conversions 198.1% year over year.</description>
      <content:encoded><![CDATA[Oneupweb replaced duplicate copy on 17 BODYBAR Pilates franchise studio pages with unique local content. In its July 2023 analysis, organic sessions to the optimized pages rose 115.9% year over year and conversions rose 198.1%. Oneupweb notes that unoptimized pages were newer, so the comparison has caveats.

- Oneupweb rewrote the copy on 17 franchise studio pages for BODYBAR Pilates, which had relied on duplicate content for scalability.
- Organic sessions to the optimized pages rose 115.9% year over year and engaged sessions rose 124.7% in July 2023.
- Organic conversions on the optimized pages rose 198.1% year over year, while engagement rate rose only 5.3%.
- Oneupweb reports pages optimized six or more months earlier performed best, and those optimized three months earlier were just starting to gain traction.
- Oneupweb cautions that many unoptimized pages were newer studios, so its optimized versus unoptimized comparison has complicating factors.]]></content:encoded>
      <category>studies</category>
      <category>local-seo</category>
    </item>
    <item>
      <title>NFON merged its country domains into one gTLD, and Austria grew</title>
      <link>https://surfil.com/blog/nfon-cctld-consolidation-international-gtld-searchviu</link>
      <guid isPermaLink="true">https://surfil.com/blog/nfon-cctld-consolidation-international-gtld-searchviu</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>searchviu&apos;s client case: NFON merged nfon.at into nfon.com/at/ and saw Austrian organic traffic up 90% in five months, with a Germany dip.</description>
      <content:encoded><![CDATA[searchviu documents how NFON, a European cloud telephony provider, merged separate country domains into one international gTLD with subdirectories. In Austria, organic traffic grew 90% in the first five months after the early 2015 migration versus the prior year, and organic leads grew 100%. Germany lost visibility first.

- After NFON moved nfon.at to nfon.com/at/ in early 2015, Austrian organic traffic grew 90% in the first five months versus the previous year.
- Leads generated through organic search in Austria grew 100% year over year in the same five-month window after the migration.
- The Germany migration lost most of its SEO visibility right after the switch, which the author ties mainly to redirect chains rather than to the domain strategy.
- The author names several factors behind the growth, including content and technical work, so the domain move is a likely driver and not the only one.
- Once country versions share one domain their performance is closely connected, which cuts both ways when one market has a problem.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Brafton&apos;s accidental sitewide noindex: a six-week recovery</title>
      <link>https://surfil.com/blog/noindex-deploy-brafton-recovery-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/noindex-deploy-brafton-recovery-case-study</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>A sprint pushed noindex onto every Brafton page in August 2019. Search traffic fell 33.2% at the peak and took six weeks to return to baseline.</description>
      <content:encoded><![CDATA[Brafton accidentally deployed a sitewide noindex tag in August 2019. Search traffic fell 33.2% in the first week. It recovered to baseline after six weeks, while some pages needed eight-plus weeks to re-index. Jeff Baker's account, published on Moz, found that pages regained their full visibility once indexed again.

- A code sprint deployed on August 1, 2019 put a noindex robots meta tag on every Brafton page, and the team noticed almost three days later.
- Search traffic was down 33.2% in the first week after the incident, which Jeff Baker reports as the peak level of damage during the incident.
- Search visibility returned to baseline after six weeks, and all pages were re-indexed after about eight to nine weeks.
- Requesting indexing, resubmitting sitemaps and updating content did not force re-crawls of about 150 stubborn pages, and one product page stayed out until October 1.
- Pages that were indexed again recovered their original search visibility, so the slow part of recovery was re-indexing, not ranking.]]></content:encoded>
      <category>studies</category>
      <category>indexing</category>
    </item>
    <item>
      <title>Rootly&apos;s AI citation rate rose from 3% to 30% in AthenaHQ case</title>
      <link>https://surfil.com/blog/rootly-athenahq-ai-citation-rate-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/rootly-athenahq-ai-citation-rate-case-study</guid>
      <pubDate>Fri, 25 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>AthenaHQ says Rootly&apos;s citation rate across tracked AI prompts rose from 3% to 30%. Here is what it reports, and what a vendor case cannot prove.</description>
      <content:encoded><![CDATA[AthenaHQ reports that incident management vendor Rootly grew its citation rate across tracked AI prompts from 3% to 30%, and its mention rate from 7.5% to 18.3%, within months. The figures come from AthenaHQ's own tracking tool, published by the vendor, with no control group and no list of the prompts or AI assistants tracked.

- AthenaHQ reports Rootly's citation rate across its tracked prompts rose from 3% to 30%, a tenfold increase.
- The same page reports Rootly's mention rate rising from 7.5% to 18.3%, a gain of 10.8 percentage points.
- The changes were shipped across documentation, content, messaging and third-party sources, guided by AthenaHQ's recommendations.
- The case is vendor-published and self-reported, with no control group, so the lift cannot be separated from other marketing.
- The page calls the mention gain a 2.5x lift on non-branded prompts but never labels which prompts the 7.5% to 18.3% figure covers.]]></content:encoded>
      <category>studies</category>
      <category>ai-search</category>
    </item>
    <item>
      <title>AI shoppers now convert 54% better than non-AI traffic, Adobe finds</title>
      <link>https://surfil.com/blog/ai-referral-conversion-vs-non-ai-retail-adobe-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/ai-referral-conversion-vs-non-ai-retail-adobe-2026</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Adobe&apos;s May 2026 data shows AI-referred retail visits converting 54% better than non-AI traffic, a reversal from a year earlier.</description>
      <content:encoded><![CDATA[Adobe Digital Insights found AI-referred visits to U.S. retail sites converted at a rate 54% higher than non-AI traffic in May 2026, reversing the deficit of a year earlier. Those visitors also showed 15% higher engagement, spent 53% more time on site, viewed 23% more pages and generated 53% more revenue per visit.

- Adobe reports AI-referred retail visitors converted at a rate 54% higher than non-AI traffic in May 2026.
- AI-referred visitors showed 15% higher engagement than non-AI traffic in May 2026, the strongest advantage Adobe has recorded since tracking began.
- AI-referred retail visits generated 53% more revenue per visit in May 2026, against non-AI visits worth 128% more twelve months earlier.
- AI-referred visitors spent 53% more time on site and viewed 23% more pages per visit than visitors from other channels in May 2026.
- AI-referred visitors were 36% less likely to bounce, with AI bounce rates in a 17-20% band while non-AI rates hovered near 27%.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
      <category>ecommerce-seo</category>
    </item>
    <item>
      <title>Mobile vs desktop request share: Cloudflare Radar 2025</title>
      <link>https://surfil.com/blog/cloudflare-radar-2025-mobile-desktop-request-share</link>
      <guid isPermaLink="true">https://surfil.com/blog/cloudflare-radar-2025-mobile-desktop-request-share</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Cloudflare Radar&apos;s 2025 Year in Review puts mobile at 43% of global requests, up from 41%, with mobile ahead in 117 countries and regions.</description>
      <content:encoded><![CDATA[Cloudflare Radar's 2025 Year in Review found that 43% of global requests came from mobile devices, up from 41% in 2024, with the remainder from desktop-class devices. Mobile led in 117 countries and regions, peaking at 75% in Sudan. Cloudflare calls the global split a steady state because it matches Year in Review readings back to 2022.

- Cloudflare found that 43% of global requests in 2025 came from mobile devices, up from 41% in 2024.
- More than half of requests came from mobile devices in 117 countries and regions, led by Sudan at 75% and Malawi at 74%.
- Gibraltar was the only location below 10% mobile share, at 5.1%, and six others fell under 25%, down from a dozen in 2024.
- Cloudflare describes the global split as a steady state, in line with its Year in Review reports dating back to 2022.
- The figures count requests seen on Cloudflare's network, so they are not the same as visits, searches or organic clicks.]]></content:encoded>
      <category>reports</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Did sites recover after the HCU? Amsive&apos;s 400-site tracking</title>
      <link>https://surfil.com/blog/hcu-recovery-amsive-400-site-sistrix-tracking</link>
      <guid isPermaLink="true">https://surfil.com/blog/hcu-recovery-amsive-400-site-sistrix-tracking</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Amsive tracked about 400 sites hit by the September 2023 Helpful Content Update and found no known significant recoveries by late March 2024.</description>
      <content:encoded><![CDATA[Lily Ray of Amsive tracked about 400 sites hit by Google's September 2023 Helpful Content Update and, as of late March 2024, found no known significant recoveries. Google's own guidance says recovery can take several months and may wait for the next core update. Neither source promises any timeline.

- Amsive monitored about 400 sites hurt by the September 2023 Helpful Content Update and further hit by the March 2024 core and spam updates.
- As of late March 2024, Amsive reported no known significant recoveries among sites negatively impacted by the September update.
- Google says some changes can take effect in a few days, but confirming a site is helpful can take several months.
- Google adds that if a few months pass without effect, the site may have to wait for the next core update.
- Amsive did not publish how the 400 sites were selected or per-category counts, so the result cannot be generalised to all sites.]]></content:encoded>
      <category>research</category>
      <category>core-updates</category>
    </item>
    <item>
      <title>Does adding JSON-LD schema raise AI citations? Ahrefs test</title>
      <link>https://surfil.com/blog/json-ld-schema-ai-citations-ahrefs-1885-pages</link>
      <guid isPermaLink="true">https://surfil.com/blog/json-ld-schema-ai-citations-ahrefs-1885-pages</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs tracked 1,885 pages that added JSON-LD against matched controls. Here is the method, the citation changes by platform and the limits.</description>
      <content:encoded><![CDATA[Adding JSON-LD schema did not raise AI citations in Ahrefs' test of 1,885 pages against matched controls. Google AI Mode (+2.4 percent) and ChatGPT (+2.2 percent) were statistically indistinguishable from zero, and AI Overviews fell 4.6 percent relative to controls, so schema alone is no citation lever.

- Ahrefs compared 1,885 pages that added JSON-LD between August 2025 and March 2026 with about 4,000 matched control pages that did not.
- Google AI Mode (+2.4 percent) and ChatGPT (+2.2 percent) changes were statistically indistinguishable from zero after schema was added.
- Google AI Overviews citations fell 4.6 percent relative to controls after schema was added, a small decline Ahrefs says it cannot explain.
- Pages already cited by AI are a different case from pages AI never sees, so the result does not prove schema is useless everywhere.
- Pages that add schema often change other things at the same time, so even this matched design cannot fully isolate schema.]]></content:encoded>
      <category>research</category>
      <category>structured-data</category>
      <category>ai-search</category>
    </item>
    <item>
      <title>Digital PR case: 12 links, 143k visits in 6 weeks, per OneLittleWeb</title>
      <link>https://surfil.com/blog/digital-pr-canada-affordability-study-onelittleweb</link>
      <guid isPermaLink="true">https://surfil.com/blog/digital-pr-canada-affordability-study-onelittleweb</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>OneLittleWeb reports a six-week Canadian digital PR campaign earned a finance blog 12+ editorial links and 143k+ new organic visits. What the case shows.</description>
      <content:encoded><![CDATA[OneLittleWeb says a six-week Canadian digital PR campaign earned an anonymous personal finance blog 12 or more editorial backlinks, over 143,000 additional organic visitors and 199,000 newly ranking keywords. The client is unnamed and the agency published the case itself, so these figures are self-reported.

- OneLittleWeb reports one campaign in six weeks earned an unnamed Canadian personal finance blog 12 or more editorial backlinks.
- The agency credits the campaign with over 143,000 additional organic visitors and 199,000 newly ranking keywords, read from an Ahrefs report dated 1 August 2023.
- The linkable asset was a data study, The 10 Most Affordable Places to Live in Canada, built from government income and spending data plus Numbeo cost-of-living data.
- The client is unnamed under an NDA and the report itself is not linked, so the figures are the agency's own and cannot be independently checked.
- The page gives two traffic figures, 143k in its headline and 50K in its title tag, so treat the traffic number as approximate.]]></content:encoded>
      <category>studies</category>
      <category>content-strategy</category>
      <category>link-building</category>
    </item>
    <item>
      <title>Product schema on a B2B store: +111% clicks, per Simplified Search</title>
      <link>https://surfil.com/blog/product-schema-b2b-ecommerce-impressions-clicks-simplified-search</link>
      <guid isPermaLink="true">https://surfil.com/blog/product-schema-b2b-ecommerce-impressions-clicks-simplified-search</guid>
      <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Simplified Search reports Product markup on one unnamed B2B ecommerce site lifted impressions 119% and clicks 111%, with no control group.</description>
      <content:encoded><![CDATA[Simplified Search reports that adding Product markup dynamically to every product page of one unnamed large B2B ecommerce site lifted impressions 119.14% and clicks 111.13%, with revenue from those pages up more than 500%. There was no control group, and the client is anonymous.

- The author added Product markup dynamically to every product page of one large B2B ecommerce site and reports impressions up 119.14% and clicks up 111.13%.
- Organic users rose 42.21% and organic sessions rose 37.44% after the Product rich feature was earned, according to the write-up.
- The author states revenue from those pages rose more than 500%, or over $80K more than the previous period, but gives no baseline revenue.
- There was no control group and the client is unnamed, so the result is a documented claim from one engagement, not a controlled test.
- The same study found review markup on about 1,500 pages raised impressions 18.23% but cut clicks 8.47%, so structured data does not lift clicks by default.]]></content:encoded>
      <category>studies</category>
      <category>structured-data</category>
      <category>ecommerce-seo</category>
    </item>
    <item>
      <title>Google AI Mode adoption and zero-click rate, Semrush 2025</title>
      <link>https://surfil.com/blog/google-ai-mode-adoption-and-zero-click-semrush</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-ai-mode-adoption-and-zero-click-semrush</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Semrush clickstream data on Google AI Mode&apos;s first two months: usage grew about 4x to just over 1% of sessions, and 92-94% ended without an external click.</description>
      <content:encoded><![CDATA[Google AI Mode was a small but fast-growing slice of Google search in its first two months, and nearly all of it stayed on Google. Semrush's clickstream study found usage rose roughly 4x to just over 1% of sessions, while 92-94% of AI Mode sessions produced no visit to an external domain.

- Semrush found AI Mode's share of Google search sessions grew roughly 4x, from 0.25% in early May 2025 to a little over 1% by early July 2025.
- On average only 6-8% of AI Mode sessions led to a visit to an external domain, so 92-94% were zero-click.
- Semrush cites separate earlier studies putting zero-click at about 34% without an AI Overview and about 43% with one, but warns the comparison is imperfect.
- The average AI Mode query ran 7.22 words against 4.0 words for traditional Google search, and sessions took 2-3 searches.
- The data covers US desktop sessions only, from May 1 to July 5, 2025, so it says nothing about mobile or later months.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
      <category>analytics</category>
    </item>
    <item>
      <title>Google Discover now drives 67.5% of Google traffic to news sites</title>
      <link>https://surfil.com/blog/google-discover-share-newzdash-publisher-traffic</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-discover-share-newzdash-publisher-traffic</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>NewzDash&apos;s analysis of 400+ news publishers shows Discover&apos;s share of Google traffic rising from 37.03% in 2023 to 67.51% in Q4 2025.</description>
      <content:encoded><![CDATA[NewzDash's analysis of more than 400 news publishers worldwide found Google Discover supplied 67.51% of Google traffic to news sites in late 2025, up from 37.03% in 2023. Traditional web search fell from 51.10% to 27.42% over the same period. The figures describe share of Google surfaces, not absolute clicks.

- NewzDash's analysis of over 400 news publishers worldwide put Google Discover's share of Google traffic to news sites at 67.51%.
- Discover's share climbed from 37.03% in 2023 to 67.51%, which NewzDash's John Shehata described as nearly doubling in two years.
- Traditional web search fell from 51.10% of Google traffic to news publishers in 2023 to 27.42% over the same period.
- The percentages are shares of Google surface traffic, so they show a shift in mix and do not report absolute clicks gained or lost.]]></content:encoded>
      <category>reports</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Which SERP features sit next to AI Overviews, per Semrush</title>
      <link>https://surfil.com/blog/serp-features-alongside-ai-overviews-semrush</link>
      <guid isPermaLink="true">https://surfil.com/blog/serp-features-alongside-ai-overviews-semrush</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Semrush&apos;s refreshed study: Related searches (95.32%) and People Also Ask (90.03%) almost always appear beside an AI Overview, and ads now reach 25%.</description>
      <content:encoded><![CDATA[Semrush's refreshed AI Overviews study found that Related searches (95.32%) and People Also Ask (90.03%) almost always appear on the same results page as an AI Overview. Featured snippets, image carousels and organic reviews now co-occur less often than in March 2025, while Google Ads now sit on 25% of AI Overview pages.

- Semrush found Related searches appear alongside an AI Overview 95.32% of the time.
- People Also Ask appears on the same results page as an AI Overview 90.03% of the time, according to Semrush.
- Featured snippets, image carousels and organic reviews are all seen less often beside an AI Overview than in March 2025.
- Google Ads at the bottom of the page now appear on 25% of AI Overview results pages, against less than 1% in March 2025.
- Semrush concludes AI Overviews are layered on top of existing formats rather than replacing them, at least for now.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
      <category>serp-features</category>
    </item>
    <item>
      <title>HTTPS adoption in the 2025 Web Almanac: 97% of homepages</title>
      <link>https://surfil.com/blog/https-adoption-web-almanac-2025-httparchive</link>
      <guid isPermaLink="true">https://surfil.com/blog/https-adoption-web-almanac-2025-httparchive</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>HTTP Archive&apos;s 2025 Web Almanac measured HTTPS on 97.3% of mobile homepages and 98.8% of requests. What the crawl method can and cannot show.</description>
      <content:encoded><![CDATA[HTTPS is now near-universal: the 2025 Web Almanac from HTTP Archive found 97.3% of mobile homepages and over 98.8% of mobile requests served over HTTPS, from a July 2025 crawl of about 16 million websites. The data measures adoption only, not ranking effect, so it explains why HTTPS no longer separates competing pages.

- The 2025 Web Almanac reports that over 98.8% of all requests on mobile were sent over HTTPS.
- Mobile homepage HTTPS adoption rose from 95.6% to 97.3% in the 2025 edition, with desktop at 97.5%.
- The Almanac notes homepage adoption runs below the request share because pages load many third-party resources.
- The crawl covers 16,213,084 websites from July 2025, tested from a datacenter, so it is a lab measurement and not real user data.
- The dataset measures adoption, not ranking, so it cannot show whether HTTPS moves any page up or down in results.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>How old are the pages that rank in Google&apos;s top 10?</title>
      <link>https://surfil.com/blog/page-age-top-ranking-pages-ahrefs-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/page-age-top-ranking-pages-ahrefs-study</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs checked the age of top 10 pages for 1.3 million US keywords: 72.9 percent are over three years old, and the average number one page is five.</description>
      <content:encoded><![CDATA[Ahrefs measured page age, not domain age, across top 10 results for 1.3 million US keywords and found 72.9 percent of ranking pages are over three years old, while the average number one page is five years old. This is an observed correlation, not proof that age causes rankings.

- Ahrefs found that 72.9 percent of pages in Google's top 10 are more than three years old, up from 59 percent in its 2017 study.
- The average page ranking number one in the Ahrefs data is five years old, compared with two years old in the 2017 version of the study.
- Only 13.7 percent of top 10 pages are under one year old, down from 22 percent in 2017.
- Ahrefs measured page age as the date its crawler first saw a URL, so the study is about pages and not about domain registration age.
- Only 1.74 percent of newly published pages reached the top 10 within a year, so age in the data reflects who survives, not proof that age ranks.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>URL length and rankings: what the Backlinko study found</title>
      <link>https://surfil.com/blog/url-length-ranking-correlation-backlinko-11-8m</link>
      <guid isPermaLink="true">https://surfil.com/blog/url-length-ranking-correlation-backlinko-11-8m</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Backlinko&apos;s 11.8M-result study found shorter URLs correlate with slightly better Google rankings. Here is the method, the numbers and the limits.</description>
      <content:encoded><![CDATA[Backlinko's analysis of 11.8 million Google search results found that shorter URLs tend to rank slightly better. Position #1 URLs averaged 9.2 characters shorter than position #10 URLs, and the average top 10 URL ran 66 characters. The authors stress it is correlation, not causation.

- Backlinko's 11.8 million result study found that first place URLs are on average 9.2 characters shorter than tenth place URLs.
- The average URL length for a top 10 Google result in the study was 66 characters, and most first page URLs fell between 40 and 100 characters.
- The authors offer a possible explanation, that a long URL tends to sit several clicks from the homepage and so receives less authority.
- Backlinko states the work is a correlation study, so the data cannot show that shortening a URL raises a ranking.
- The study does not report on keywords in the URL, so no claim about keyword placement in URLs can be sourced to it.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>One publisher&apos;s Discover-driven traffic and revenue, per Adsterra</title>
      <link>https://surfil.com/blog/discover-traffic-independent-publisher-adsterra-case</link>
      <guid isPermaLink="true">https://surfil.com/blog/discover-traffic-independent-publisher-adsterra-case</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Adsterra&apos;s case study of an independent Indian publisher: 8.45 million impressions, 703,948 clicks and $6,976.06 in revenue, about 95% from Discover.</description>
      <content:encoded><![CDATA[Adsterra's ad network published a case study of Karan, an independent publisher in India whose traffic came about 95% from Google Discover. He posted four to seven trend-led articles a day, mostly to mobile readers. The reported totals were 8.45 million impressions, 703,948 clicks and $6,976.06 in revenue, with no dates given.

- Adsterra reports that roughly 95% of this independent publisher's traffic came from Google Discover, with about 85% of it on mobile devices.
- The publisher posted four to seven articles a day, built around watching trends and reacting to them quickly.
- Reported totals were 8.45 million impressions, 703,948 clicks, an 8.32% average CTR and $6,976.06 in revenue.
- Revenue depended heavily on geography, with India's CPM of $0.339 far below the United States at $23.30 and Canada at $19.97.
- The article gives no dates and no before figures, and it is an ad network's own promotion, so treat it as one self-reported case, not a benchmark.]]></content:encoded>
      <category>studies</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>On reports +67% Google traffic after prerendering its JavaScript site</title>
      <link>https://surfil.com/blog/on-prerender-javascript-rendering-traffic-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/on-prerender-javascript-rendering-traffic-case-study</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>On&apos;s Prerender.io case study reports +67% Google traffic and +98% Bing traffic after prerendering a JavaScript front end instead of building SSR.</description>
      <content:encoded><![CDATA[On, a sportswear retailer, says bots could not read its JavaScript pages before a 2023 front-end update. It chose Prerender.io over building server-side rendering and reports a 67% rise in Google traffic and 98% in Bing after nearly three years. The figures are percentages published by the vendor, with no baselines.

- On reports a 67% increase in Google traffic and a 98% increase in Bing traffic after nearly three years on Prerender.io.
- On evaluated server-side rendering first and rejected it because it would have tied up developers for several months.
- On says traffic fell to under 10% of normal in one low-traffic market when Prerender.io was switched off for a month.
- The case study is written by the vendor and gives percentage changes only, with no baseline traffic figures.
- Prerendering serves rendered HTML to bots, so it is a different fix from SSR, SSG or hydration in the browser.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Pillar pages lifted an agency&apos;s organic traffic 744% in 13 months</title>
      <link>https://surfil.com/blog/topic-cluster-pillar-pages-campaign-creators-case</link>
      <guid isPermaLink="true">https://surfil.com/blog/topic-cluster-pillar-pages-campaign-creators-case</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Campaign Creators reorganized six years of blog posts into three pillar pages and reports organic visits rising from 1,091 to 9,214.</description>
      <content:encoded><![CDATA[Campaign Creators, a HubSpot agency, organized six years of blog posts into three pillar pages in 2017. Its own analytics show organic visits rising from 1,091 in August 2017 to 3,497 by February 2018 and 9,214 by September 2018. It is a single site, self-reported, with no control group.

- Campaign Creators reports organic visits rising from 1,091 in August 2017 to 9,214 in September 2018, a 744% increase.
- The agency built three pillar pages in about three months, covering lead generation, lead nurturing and marketing technology.
- Most of the first three months went on auditing and relinking existing posts rather than writing new ones, per the agency's own account.
- The result is one site's self-reported analytics with no control group, so it shows the tactic can work, not that it will.
- The traffic baseline was low, so a large percentage gain still means a modest absolute number of visits.]]></content:encoded>
      <category>studies</category>
      <category>content-strategy</category>
    </item>
    <item>
      <title>Shopping Graph grew from 45 to 60 billion listings in 19 months</title>
      <link>https://surfil.com/blog/google-shopping-graph-visual-search-scale</link>
      <guid isPermaLink="true">https://surfil.com/blog/google-shopping-graph-visual-search-scale</guid>
      <pubDate>Tue, 22 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Google&apos;s own disclosures put the Shopping Graph at 45 billion listings in October 2024, 50 billion in January 2026, and past 60 billion by May 2026.</description>
      <content:encoded><![CDATA[Google has stated three dated Shopping Graph sizes: more than 45 billion product listings in October 2024, over 50 billion at NRF in January 2026, and past 60 billion at I/O 2026 in May 2026. The same October 2024 post disclosed that 20 percent of Google Lens searches are shopping-related, tying visual search directly to this catalog's growth.

- Google's Shopping Graph held more than 45 billion product listings on October 3, 2024, per a Google Shopping blog post announcing Lens shopping features.
- Sundar Pichai put the figure at over 50 billion listings, refreshing 2 billion of them hourly, in January 11, 2026 remarks at the National Retail Federation.
- Google's Ads and Commerce VP disclosed a catalog of over 60 billion listings when announcing Universal Cart on May 19, 2026, about a 20% jump in four months.
- The October 2024 post is also the source for Google's claim that 20% of all Lens searches are shopping-related, the only dated figure given for Lens composition.
- Google has never disclosed how a listing is counted, whether duplicate offers across retailers are merged, or what share of the catalog surfaces in visual search.]]></content:encoded>
      <category>reports</category>
      <category>structured-data</category>
      <category>ai-search</category>
      <category>ecommerce-seo</category>
    </item>
    <item>
      <title>Does anchor text type actually correlate with Google rankings?</title>
      <link>https://surfil.com/blog/anchor-text-type-ranking-correlation-ahrefs-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/anchor-text-type-ranking-correlation-ahrefs-study</guid>
      <pubDate>Tue, 22 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs&apos; 384,614-page study isolated exact, phrase, partial and random-match anchor text and found every correlation with rank was weak.</description>
      <content:encoded><![CDATA[Ahrefs studied 384,614 pages ranking for 19,840 keywords, controlling for URL Rating, and found exact-match anchor text correlates with rank at just 0.1436 (Spearman, average). Phrase and partial match scored lower still, and random anchor text barely correlated at all. Every anchor text type it tested showed a weak relationship with ranking position, not a strong one.

- Ahrefs analyzed 384,614 web pages ranking for 19,840 keywords and controlled for URL Rating to isolate anchor text as a variable.
- Exact-match anchor text had the strongest correlation with rankings in the dataset, and it was still only 0.1436 on average, a weak Spearman value.
- Phrase-match and partial-match anchors scored close to each other, around 0.106 and 0.108, both weaker than exact match.
- Random anchor text, unrelated to the target keyword, correlated at only 0.0161, nearly flat.
- A second test of 16,000 pages found the words surrounding a link's anchor text correlated with rank even more weakly, and one measure was slightly negative.]]></content:encoded>
      <category>research</category>
      <category>link-building</category>
    </item>
    <item>
      <title>Italiaonline: 841% video click growth after schema fixes</title>
      <link>https://surfil.com/blog/italiaonline-video-seo-schema-case-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/italiaonline-video-seo-schema-case-study</guid>
      <pubDate>Tue, 22 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Google&apos;s case study on Italiaonline reports an 841% rise in video clicks and 353% more impressions after fixing VideoObject structured data errors.</description>
      <content:encoded><![CDATA[Google's own case study on Italiaonline, one of Italy's largest digital publishers, states video clicks in Search rose 841% and impressions rose 353% after the company implemented video structured data and Videos on Search best practices. Indexing errors dropped 85%. Google published the figures without a baseline click count or a measurement window.

- Google's case study states Italiaonline's video clicks on Search increased 841% after implementing video structured data and Videos on Search best practices.
- The same case study states organic Search impressions for Italiaonline's video content rose 353% over the same rollout.
- Video indexing errors, tracked through the Search Console video indexing report, fell 85% once the structured data issues were fixed.
- Italiaonline reaches 62% of the Italian market across 27.2 million unique users, giving the case study a large, verifiable publisher rather than an anonymous site.
- Google names two Italiaonline executives but does not disclose a start date, end date, or the baseline click and impression counts behind either percentage.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Whitespark&apos;s 2026 local ranking factors: what actually moved</title>
      <link>https://surfil.com/blog/whitespark-local-ranking-factors-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/whitespark-local-ranking-factors-2026</guid>
      <pubDate>Mon, 21 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Whitespark surveyed 47 local search experts on 187 ranking factors for 2026. Here is the verified top 10 for the local pack and what changed since 2023.</description>
      <content:encoded><![CDATA[Whitespark's 2026 Local Search Ranking Factors survey asked 47 local search experts to score 187 factors across eight signal groups. Primary GBP category ranks first for local pack and map results, followed by proximity to the searcher and keywords in the business title. Review and behavioral signals grew in weight since 2023, while on-page, link and citation signals fell back.

- Whitespark surveyed 47 local search experts on 187 ranking factors across eight signal groups for the 2026 edition, published November 6, 2025.
- Primary GBP category is the top-scored local pack and Maps factor, followed by proximity to the point of search and keywords in the Google Business Profile title.
- Business hours entered the survey as a new factor and landed at fifth place, ahead of high numerical ratings and address visibility on the profile.
- Review signals and behavioral signals grew in importance compared to the 2023 edition, while on-page, link and citation signals declined.
- The 2026 survey added an AI search visibility category for the first time, asking experts which signals affect visibility in AI responses rather than classic local pack ranking.]]></content:encoded>
      <category>reports</category>
      <category>local-seo</category>
    </item>
    <item>
      <title>Page weight and rankings: what the Backlinko study actually found</title>
      <link>https://surfil.com/blog/page-weight-byte-size-ranking-correlation-backlinko</link>
      <guid isPermaLink="true">https://surfil.com/blog/page-weight-byte-size-ranking-correlation-backlinko</guid>
      <pubDate>Mon, 21 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Backlinko and Ahrefs analyzed 11.8 million search results and found no correlation between page HTML size and Google rankings.</description>
      <content:encoded><![CDATA[Backlinko and Ahrefs analyzed 11.8 million Google search results and found no correlation between page HTML size and ranking position. The same study also found zero correlation between site loading speed and rankings, with first-page results averaging 1.65 seconds to load. The authors call it correlation only, not causation.

- Backlinko and Ahrefs analyzed 11.8 million Google search results and found no correlation between page HTML size and ranking position.
- The same dataset showed zero correlation between site loading speed, a domain-level median, and first page rankings, with an average load time of 1.65 seconds.
- The study measured page HTML size and load speed only, not the number of HTTP requests or a mobile versus desktop split.
- Backlinko's own authors describe the results as correlational, stating it is impossible to determine the underlying reason behind any pattern.
- No correlation to rankings is not proof that page weight is irrelevant, since a bloated page can still hurt conversions and crawl efficiency on its own terms.]]></content:encoded>
      <category>research</category>
      <category>technical-seo</category>
      <category>page-speed</category>
    </item>
    <item>
      <title>Silly Santa&apos;s 7-country hreflang rollout: a Skalum case study</title>
      <link>https://surfil.com/blog/sillysanta-hreflang-cctld-expansion</link>
      <guid isPermaLink="true">https://surfil.com/blog/sillysanta-hreflang-cctld-expansion</guid>
      <pubDate>Mon, 21 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Skalum&apos;s case study on Shopify brand Silly Santa: hreflang across 7 ccTLDs, with UK organic users up 297.5% and Norway&apos;s local traffic share up to 93%.</description>
      <content:encoded><![CDATA[Skalum, an ecommerce SEO agency, published a case study on Silly Santa, a Shopify apparel brand running seven country domains. Alongside hreflang, it also reworked content, links and internal linking. Skalum reports UK organic users grew 297.5% and Norway's local traffic share rose from 80% to 93%, comparing the same six-month season a year apart.

- Silly Santa runs seven separate Shopify ccTLD domains for Germany, Denmark, Finland, the Netherlands, Norway, Sweden and the UK.
- Skalum's case study says it implemented hreflang signals across all seven domains to stop traffic spilling across similar-language markets and establish clean country targeting.
- Skalum reports organic users grew across every domain in the comparison, with the UK domain up 297.5% and the Netherlands up 141.1%.
- Norway's local traffic share, Skalum's own label for how local the domain's organic traffic is, rose from 80% to 93% after hreflang assigned the market.
- The comparison spans the same August to January season a year apart and Skalum also lists content and link work, so it does not isolate hreflang.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
      <category>ecommerce-seo</category>
      <category>analytics</category>
    </item>
    <item>
      <title>Which AI crawlers do news sites block most? CCBot tops 75%</title>
      <link>https://surfil.com/blog/news-site-ai-crawler-blocking-by-bot-2026</link>
      <guid isPermaLink="true">https://surfil.com/blog/news-site-ai-crawler-blocking-by-bot-2026</guid>
      <pubDate>Sun, 20 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>A robots.txt scan of 100 top UK and US news sites finds CCBot blocked by 75%, the most of any AI bot. Google-Extended is the least-blocked training bot.</description>
      <content:encoded><![CDATA[A robots.txt scan of the top 50 UK and top 50 US news sites finds CCBot blocked by 75%, the highest of any crawler checked, ahead of Anthropic-ai at 72%, ClaudeBot at 69% and GPTBot at 62%. Google-Extended is blocked least among training bots, at 46%. Overall, 79% of sites block at least one AI training bot.

- CCBot, Common Crawl's bot, is the single most-blocked AI crawler among the 100 news sites checked, disallowed by 75% of them.
- GPTBot is blocked by 62% of the sites studied, well behind CCBot, Anthropic-ai and ClaudeBot but still ahead of Google-Extended.
- Google-Extended is the least-blocked training bot at 46%, and US publishers block it far more often than UK publishers, 58% versus 29%.
- Live retrieval bots are blocked less than training bots overall, with Perplexity-User disallowed by only 17% of sites versus 67% for PerplexityBot, Perplexity's indexing bot.
- Just 14% of the news sites studied block every AI bot the researchers checked, while 18% block none of them.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
    </item>
    <item>
      <title>A year of SearchPilot split tests on adding FAQ schema</title>
      <link>https://surfil.com/blog/faq-schema-uplift-searchpilot</link>
      <guid isPermaLink="true">https://surfil.com/blog/faq-schema-uplift-searchpilot</guid>
      <pubDate>Sun, 20 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>SearchPilot&apos;s aggregate split tests found FAQ schema was positive 67% of the time, with organic traffic uplifts of 4-15% while the rich result existed.</description>
      <content:encoded><![CDATA[SearchPilot ran controlled split tests adding FAQ schema across many clients and industries from 2019 to 2020, finding 67% of tests positive with organic traffic uplifts typically between 4% and 15%, one test reaching 25%. One test turned negative when FAQ schema replaced review and price markup instead of running alongside it, a 5% decline.

- SearchPilot found that 67% of its FAQ schema split tests across clients and industries produced a positive result on organic traffic.
- Positive tests typically showed organic traffic uplifts between 4% and 15%, with one outlier test reaching a 25% uplift.
- The one negative result in SearchPilot's data came from a test that swapped review and price schema for FAQ schema rather than adding it alongside them, a 5% decline.
- Uplifts tended to run higher on mobile than desktop, which SearchPilot attributes to the FAQ rich result claiming more SERP real estate on smaller screens.
- This case study predates Google's later restriction and removal of the FAQ rich result, so its uplift figures describe a feature that no longer produces a visible result today.]]></content:encoded>
      <category>research</category>
      <category>structured-data</category>
      <category>serp-features</category>
    </item>
    <item>
      <title>Valery Real Estate: 324,580 404s, then a 1,379% traffic recovery</title>
      <link>https://surfil.com/blog/valery-real-estate-404-redirect-recovery</link>
      <guid isPermaLink="true">https://surfil.com/blog/valery-real-estate-404-redirect-recovery</guid>
      <pubDate>Sun, 20 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Croton Content&apos;s audit found 324,580 broken URLs after a rebrand with no redirects. Fixing them took weekly organic sessions from 974 to a 12,276 peak.</description>
      <content:encoded><![CDATA[Croton Content's technical SEO case study on Valery Real Estate found a September 2024 rebrand had left 324,580 URLs returning 404 errors with no redirects in place. Mapping 301 redirects at scale and cleaning the sitemap took weekly organic sessions from a 974 floor to a 12,276 peak, and 2025 sessions rose 1,379% on a four-month 2024 base.

- Croton Content's audit of Valery Real Estate found 324,580 pages returning 404 errors after a September 2024 rebrand left old URLs without redirects.
- Weekly organic sessions had fallen to a floor of 974 by May 2025, the point at which the audit was delivered and the fix was scoped.
- 301 redirects went live in late August 2025, three months after the audit; weekly sessions hit 5,847 in the week of 3 to 9 September and peaked at 12,276.
- Sessions rose from 14,918 in 2024 (only the four months after launch) to 220,570 in 2025, a 1,379% increase, and registrations more than doubled after the fix.
- Croton Content is explicit this was recovery of authority trapped behind broken URLs, not new growth, and the fix sat unimplemented for three months.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Portent: 1-second e-commerce sites convert 2.5x higher</title>
      <link>https://surfil.com/blog/portent-page-speed-conversion-rate-2022</link>
      <guid isPermaLink="true">https://surfil.com/blog/portent-page-speed-conversion-rate-2022</guid>
      <pubDate>Sat, 19 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Portent&apos;s aggregate analysis of 100 million page views across 20 sites found B2C conversion falls from 3.05% at 1 second to 0.67% at 4 seconds.</description>
      <content:encoded><![CDATA[Portent analyzed 5.6 million sessions across 20 B2B and B2C sites over 30 days. B2C e-commerce conversion rate fell from 3.05% at a 1-second load to 0.67% at 4 seconds, and Portent puts the 1-second versus 5-second gap at 2.5x. B2B lead-generation conversion was almost 40% at 1 second and 29% at 3 seconds, a 3x gap against 5 seconds.

- Portent's data covered just over 100 million page views and 5.6 million sessions across 20 B2B and B2C sites over a 30-day snapshot.
- B2C e-commerce conversion rate dropped from 3.05% at a 1-second load time to 0.67% at 4 seconds, and Portent reports a 2.5x difference between 1-second and 5-second loads.
- B2B lead-generation conversion rate was almost 40% at 1 second and 34% at 2 seconds, and Portent reports it 3x higher at 1 second than at 5 seconds and 5x higher than at 10.
- The six e-commerce sites converted far less often than the fourteen lead-generation sites, since a purchase is a higher barrier than submitting contact information.
- Portent flags its B2C data in the 10-second range as thinner than the rest, since it had fewer pages to work with at that load time.]]></content:encoded>
      <category>reports</category>
      <category>technical-seo</category>
    </item>
    <item>
      <title>Does faster backlink growth move rankings faster?</title>
      <link>https://surfil.com/blog/backlink-velocity-ranking-movement-ahrefs-study</link>
      <guid isPermaLink="true">https://surfil.com/blog/backlink-velocity-ranking-movement-ahrefs-study</guid>
      <pubDate>Sat, 19 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Ahrefs tracked backlink growth for 200,000 pages across 10,000 keywords and found no observable link between acquisition pace and ranking movement.</description>
      <content:encoded><![CDATA[Ahrefs tracked followed-referring-domain growth for 200,000 top-20 pages across 10,000 keywords over three months. Among pages that held a top-3 spot throughout, it found no observable relationship between backlink growth pace and ranking position. Position 1 pages typically gained referring domains at 5 to 14.5 percent per month, a pace that declined by rank.

- Ahrefs analyzed 200,000 pages across 10,000 non-branded keywords, tracking followed referring domain growth over three months using its own backlink index.
- Position 1 pages typically gained new referring domains at a pace of 5 to 14.5 percent per month, and the median pace declined as rank fell.
- Among pages in the top three both at the start and end, Ahrefs found no observable relationship between backlink pace and ranking position.
- Ahrefs states that its results can neither support nor disprove that pages get links as a byproduct of ranking, and that cause and effect would need a different experiment.
- Ahrefs notes that not all links carry equal weight and that backlinks are only one of many ranking factors the study could not isolate.]]></content:encoded>
      <category>research</category>
      <category>link-building</category>
    </item>
    <item>
      <title>REI cut 34 million URLs down to 300,000</title>
      <link>https://surfil.com/blog/rei-faceted-navigation-crawl-budget-botify</link>
      <guid isPermaLink="true">https://surfil.com/blog/rei-faceted-navigation-crawl-budget-botify</guid>
      <pubDate>Sat, 19 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>REI&apos;s technical SEO team told Botify&apos;s Crawl2Convert summit how it cut its site from 34 million URLs to 300,000, including faceted navigation.</description>
      <content:encoded><![CDATA[REI's technical SEO team disclosed that its site had 34 million URLs before a cleanup cut that figure to 300,000, including URLs crawled from faceted navigation. The team then slowly reintroduced significant URLs. Botify published the account from its own customer conference and gives no traffic or ranking figures.

- REI's technical SEO team found the retailer's site had roughly 34 million URLs, and Botify's recap names faceted navigation among the URL types that waste crawl budget.
- The team cut that figure down to 300,000 crawlable URLs partly by cutting down the URLs crawled from faceted navigation.
- REI reintroduced a smaller set of URLs afterward, described only as significant URLs.
- Ryan Ricketts, REI's Technical SEO Manager, presented the account at Botify's Crawl2Convert customer summit rather than in a written case study document.
- Botify's own recap of the same story does not publish organic traffic, ranking or revenue figures alongside the URL count, only the crawl-budget outcome.]]></content:encoded>
      <category>studies</category>
      <category>technical-seo</category>
      <category>ecommerce-seo</category>
    </item>
    <item>
      <title>AI Overview trigger rate by query type: comparison queries hit 95%</title>
      <link>https://surfil.com/blog/ai-overview-trigger-rate-by-query-type</link>
      <guid isPermaLink="true">https://surfil.com/blog/ai-overview-trigger-rate-by-query-type</guid>
      <pubDate>Fri, 18 Sep 2026 00:00:00 GMT</pubDate>
      <dc:creator>Adam Hafez</dc:creator>
      <description>Seer Interactive tracked 49,358 queries and found comparison searches trigger an AI Overview 95.4% of the time, versus 27.3% for single words.</description>
      <content:encoded><![CDATA[Seer Interactive's 2026 CTR study classified 49,358 queries by type and found comparison searches ("X vs Y") trigger an AI Overview 95.4% of the time, question-format queries 85.9%, and "near me" informational queries 76.9%. By intent, informational queries average 36%, commercial 8%, and transactional just 5%, a roughly sevenfold gap.

- Seer Interactive found comparison queries such as "X vs Y" triggered an AI Overview in 95.4% of 79,436 tracked organic impressions.
- Question-format queries beginning with what, why, how, is or are triggered an AI Overview 85.9% of the time across 1.4 million impressions.
- Single-word informational queries still triggered an AI Overview in 27.3% of cases across 25 million impressions, so query length alone is not a reliable safety signal.
- Informational queries averaged 36% AI Overview prevalence across the full 2025 dataset, compared with 8% for commercial intent and 5% for transactional intent.
- Informational queries that included "near me" triggered an AI Overview 76.9% of the time, a rate far above any length-based bucket despite carrying local intent.]]></content:encoded>
      <category>reports</category>
      <category>ai-search</category>
    </item>
  </channel>
</rss>
