Do clicks move rankings? Documented vs. leaked
Few SEO debates run hotter than click signals. Our evidence base holds three distinct layers here — what engines document, what researchers measured, and what leak interpretations suggest — and they deserve to be kept apart.
Last reviewed: 2026-09-15
What Google documents
Supported, from Google’s own documentation and statements: aggregated and anonymized interaction data is used to assess whether results are relevant; interaction data is transformed into signals that help machine-learned systems estimate relevance; Google can use strategies that predict what users would click; and Gary Illyes has said click behavior can be used directly in certain cases, including personalized results. That is a real role for clicks — aggregated, transformed, mostly indirect.
What research measured
Also supported: Microsoft researchers described query refinement, result clicks and dwell times as behavioral signals for estimating search satisfaction — and showed that a single fixed dwell-time threshold misclassifies satisfaction, because reading time is not uniform. That second finding matters for practitioners: crude dwell-time interpretations of your own analytics inherit the same flaw.
What the leaks suggest — kept in its own box
The 2024 leak commentary is in our evidence base as claims about interpretations: Rand Fishkin interpreted the purported documents as suggesting Google filters clicks, measures click length, and measures pogo-like return behavior. Those interpretation claims are supported as descriptions of what was said — which is not the same as the underlying behavior being confirmed. And the strongest folk version, that widespread pogo-sticking directly drops a result several positions, ended our evaluation with insufficient evidence.
Editory’s recommendation
Earn clicks the durable way — titles and descriptions that match intent — and ignore anything promising to move rankings by manufacturing engagement. The documented mechanism is aggregated and machine-learned; it is not a dial you can turn, and click-bot schemes bet real money on an unproven mechanism.
Common misconceptions
“CTR is a direct ranking factor — buy clicks, gain positions.”
Documented use of clicks is aggregated, anonymized and transformed into machine-learned relevance signals; the direct-manipulation version rests on leak interpretations and a pogo-sticking claim our evaluation found unproven.
What the engine does
audit:rank:serp-scanRank tracking reads real search results for tracked keywords rather than inferring positions from engagement metrics — and Editory never manufactures clicks or engagement.
Evidence
The claims behind this page, with their status in the evidence release. Click a claim for its sources.
Supported — asserted as written
Google’s current How Search Works documentation says aggregated and anonymized interaction data is used to assess whether search results are relevant to queries.Supported
Claim ID: clm_019ff289-27f4-7186-b5d3-4df5f001c935
Sources
- How Does Google Determine Ranking Results - Google Search — Google (google.com How Search Works) · primary source
“We also use aggregated and anonymized interaction data to assess whether search results are relevant to queries.”
- Automatically generating and ranking results — Google Search · primary source
“We also use aggregated and anonymized interaction data to assess whether search results are relevant to queries.”
- Google Search Won't Explain How It Uses User Interaction Signals — Search Engine Roundtable
“it says this on the How Search Works site,”
- Unbiased Learning-to-Rank with Biased Feedback — arXiv
“Implicit feedback (e.g., clicks, dwell times, etc.) is an abundant source of data in human-interactive systems.”
Google’s current documentation says interaction data is transformed into signals that help machine-learned systems estimate relevance.Supported
Claim ID: clm_019ff289-27f4-79c3-87cb-ecd39991cf6c
Sources
- How Does Google Determine Ranking Results - Google Search — Google (google.com How Search Works) · primary source
“We transform that data into signals that help our machine-learned systems better estimate relevance.”
- How Does Google Determine Ranking Results - Google Search — Google · primary source
“We transform that data into signals that help our machine-learned systems better estimate relevance.”
- Unbiased Learning-to-Rank with Biased Feedback — arXiv · primary source
“directly using click data as a training signal in Learning-to-Rank”
- Is Using Google Analytics A Search Ranking Factor? — Search Engine Journal
“We transform that data into signals that help our machine-learned systems better estimate relevance.”
Google can use a strategy that predicts what users will click.Supported
Claim ID: clm_a6e6427a-1329-4e83-9d5e-4879cd174d56
Sources
- US8938463B1 - Modifying search result ranking based on implicit user feedback and a model of presentation bias — Google Patents / USPTO (assignee Google Inc.; inventors Hyung-Jin Kim, Adrian D. Corduneanu) · primary source
“use a comparison of the predicted click through rate and the click fraction to adjust the respective ranking scores”
- US8938463B1 - Modifying search result ranking based on implicit user feedback and a model of presentation bias — Google Patents · primary source
“the model is trained to predict a click through rate”
- Trial Exhibit - UPXD104: U.S. and Plaintiff States v. Google LLC — U.S. Department of Justice · primary source
“Exploiting user feedback, principally clicks, has been the major theme of ranking work for the past decade”
- United States v. Google LLC - opinion excerpts (Picker, Antitrust Fall 2025 supplement) — University of Chicago Law School (reproducing federal court opinion text)
“RankEmbed and its later iteration RankEmbedBERT are ranking models that rely on two main sources of data”
Search Engine Land reported that Illyes said click behavior could be used directly in certain cases, including personalized search results.Supported
Claim ID: clm_019ff289-27f4-740c-82a9-4836521132d0
Sources
- Former Googler: Google 'using clicks in rankings' — Search Engine Land · primary source
“In certain cases it makes sense to use clicks directly. For instance if you have personalized results”
- #SMX Advanced keynote: Google's Gary Illyes talks RankBrain, Penguin update & more — Search Engine Land · primary source
“We use clicks for very specific things, like personalization”
- SMX Advanced 2015 Live Blog: AMA With Google Search — Search Engine Land · primary source
“In certain cases, it does make sense to use clicks — in personalized results, for example.”
Microsoft researchers described query refinement, result clicks, and dwell times as behavioral signals used to estimate search satisfaction.Supported
Claim ID: clm_019ff289-27f0-7c22-82b0-6e63ae03c51d
Sources
- Personalized Models of Search Satisfaction (CIKM 2013) — Microsoft Research (ACM CIKM 2013) · primary source
“estimate satisfaction from behavioral signals such as query refinement, result clicks, and dwell times”
- Modeling Dwell Time to Predict Click-level Satisfaction (WSDM 2014) — Microsoft Research (ACM WSDM 2014) · primary source
“Clicks on search results are the most widely used behavioral signals for predicting search satisfaction”
- Beyond Clicks: Query Reformulation as a Predictor of Search Satisfaction (CIKM 2013) — Microsoft Research (ACM CIKM 2013) · primary source
“implicit behavior is a useful data source, with clicks being the best-known implicit signal”
- Personalized Models of Search Satisfaction — Microsoft Research · primary source
“behavioral signals such as query refinement, result clicks, and dwell times”
The Microsoft researchers argued that a single fixed dwell-time threshold assumes uniform reading time and can misclassify click-level satisfaction.Supported
Claim ID: clm_019ff289-27f1-7a7d-94d0-165bdfbb8bee
Sources
- Modeling Dwell Time to Predict Click-level Satisfaction (WSDM 2014) — Microsoft Research / ACM WSDM 2014 (Kim, Hassan Awadallah, White, Zitouni) · primary source
“having a single threshold value assumes that users need a fixed amount of time to be satisfied with any result click”
- Modeling Dwell Time to Predict Click-level Satisfaction - Microsoft Research publication page — Microsoft Research · primary source
“the amount of dwell time needed to predict whether any click is associated with satisfaction”
- Explicit In Situ User Feedback for Web Search Results — Microsoft Research · primary source
“the dwell time threshold between negative and positive in situ feedback being 87 seconds, longer than the more common heuristic of 30 seconds”
- Personalized Models of Search Satisfaction — Microsoft Research · primary source
“proposed models can predict search satisfaction more accurately than a global baseline that applies the same satisfaction model across all users”
Fishkin interpreted the purported documents as suggesting that Google filters clicks.Supported
Claim ID: clm_019ff289-27f5-7e10-95da-29e104aa99cc
Sources
- An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them — SparkToro (Rand Fishkin) · primary source
“Google appears to have ways to filter out clicks”
- GoogleApi.ContentWarehouse.V1.Model.QualityNavboostCrapsCrapsClickSignals — HexDocs (Google API Content Warehouse mirror) · primary source
“badClicks; goodClicks; unsquashedClicks”
- Ahrefs monthly digest: takeaways from Rand Fishkin's Google leak article — Ahrefs
“ways to filter out clicks they don’t want to count”
- Google Confirms Leaked Search Documents are Real — KBS Sidhu (Substack)
“used to fight manual and automated click spam”
Fishkin interpreted the purported documents as suggesting that Google measures click length.Supported
Claim ID: clm_218216e6-b1be-5288-b6a7-e1fc04f56b91
Sources
- An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them — SparkToro (Rand Fishkin) · primary source
“They also seem to measure length of clicks (i.e. pogo-sticking”
- Google Algorithm Leak: The Source Awakens — Atlas SEO
“Google also appears to measure the length of clicks”
- Breaking Down the Google Search Documentation Leak — Neil Patel / NP Digital
“It also looks like they measure the length of clicks.”
Fishkin interpreted the purported documents as suggesting that Google measures pogo-like return behavior.Supported
Claim ID: clm_ff3b20e7-8d66-525c-9d25-8473ae727fce
Sources
- An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them — SparkToro (Rand Fishkin) · primary source
“They also seem to measure length of clicks (i.e. pogo-sticking - when a searcher clicks a result and then quickly clicks”
- Google leak: what the secret document on Search reveals — SEOZoom
“rehabilitates in some senses the analysis of behaviors such as pogo-sticking”
- How A Recent Google Algorithm Leak Impacts Contractors And Home Service Companies — Footbridge Media
“The docs refer to a variety of "click" tracking, including badClicks, goodClicks, lastLongestClicks, and more.”
Tracked — not yet established
Widespread pogo-sticking after a result click signals user dissatisfaction to Google and can cause that result to drop several ranking positions.Uncertain
Claim ID: clm_e271ef88-56dd-4a5c-99fb-37ab9b4e1da2
Sources
- Plaintiffs' Proposed Findings of Fact [Redacted]: U.S. and Plaintiff States v. Google LLC — U.S. Department of Justice · primary source
“Navboost memorizes all clicks that have been issued for all queries received in the last 13 months”
- Google Does Not Use Pogo-Sticking As a Ranking Factor — Search Engine Journal
“We try not to use signals like that when it comes to search.”
- 2024 Google Search documentation leak — Wikipedia
“Google representatives had previously downplayed or denied the use of click-based signals in rankings.”
- NavBoost: Google's Hidden Click Ranking System Revealed — Hueston
“Pandu Nayak—Google's VP of Search—confirmed that NavBoost, a click-based ranking system”
Related topics
Link myths: .edu magic and other unproven signals
Link building attracts folklore because links are hard to measure from outside. Here is one myth our evaluation found contradicted, one signal that remains genuinely uncertain, and what is actually supported about risky link profiles.
E-E-A-T: what Google actually says it is
E-E-A-T is the most quoted and most misquoted concept in content SEO. The evidence supports a narrower reading than most advice assumes.