Ranking Signals
Behavioral Ranking Factors
CTR, dwell time, bounce rate — how search engines read user behaviour and convert it into ranking signals. Google vs Yandex: who uses what, and how.
Behavioural ranking factors are metrics that reflect how users interact with search results and web pages. CTR, dwell time, bounce rate, and pogo-sticking are signals that search engines collect in real time. Google uses them indirectly to train ML models; Yandex uses them directly as a weighted ranking factor.
What are behavioural ranking factors
Behavioural factors split into two classes: SERP signals (how users interact with the results page) and on-site signals (how users behave after clicking through). Search engines access both: clicks are logged directly from the SERP interface; on-site behaviour is captured through Chrome, proprietary counters, and webmaster tools.
Key behaviour metrics
Average organic CTR
Typical for positions 3–10. Position 1 earns 25–30% CTR on average
Healthy dwell time
Time between click and SERP return. Longer signals content satisfaction
Normal bounce rate for blogs
E-commerce targets below 40%. Varies heavily by page type
Target pogo-sticking rate
Rapid SERP returns signal unmet search intent — eliminate them
| Signal | What it measures | Data source |
|---|---|---|
| CTR (click-through rate) | Snippet attractiveness in SERP | Search engine click logs |
| Dwell time | Content satisfaction after click | Time between click and SERP return |
| Bounce rate | Session-level engagement | Analytics counter / browser |
| Pogo-sticking | How quickly user leaves | SERP logs + session timing |
| Direct visits | Brand trust and loyalty | Direct traffic in analytics |
How algorithms process behavioural signals
Behavioural data does not feed directly into a ranking formula. Instead, it serves as training signal for ML models — most notably RankBrain at Google. The model learns to predict "user satisfaction" from click and session patterns across hundreds of thousands of similar queries.
Google has never officially confirmed that behavioural metrics are a direct ranking signal — a key difference from Yandex, which openly acknowledges them as part of its algorithm. However, the 2024 internal documentation leak revealed a system called NavBoost that tracks clicks and sessions, providing strong indirect evidence of their influence.
User sees SERP and clicks. CTR is logged and position-normalised by the search engine relative to the average for that position.
Dwell time begins. The algorithm waits: will the user return to the SERP (pogo-sticking) or stay and engage with the content?
The ML model accumulates signals across hundreds of thousands of similar queries. A single session barely affects position — statistical significance across the full dataset matters.
Position adjustments don't happen instantly. Building sufficient statistical significance takes days to weeks after a change.
The page rises or falls depending on whether it satisfies user intent better than competitors ranking for the same query.
Google vs Yandex — different approaches to behaviour
The two major search engines treat behavioural factors very differently. Yandex openly acknowledges them as a direct ranking signal and has warned against manipulation since 2011. Google officially denies direct use of these metrics — though indirect evidence has steadily accumulated.
| Signal | Yandex | |
|---|---|---|
| CTR in SERP | ML training signal | Direct ranking factor |
| Dwell time | Not officially confirmed | High explicit weight |
| Bounce rate | Not officially confirmed | Direct quality signal |
| Pogo-sticking | Indirect via RankBrain | Direct negative signal |
| Direct visits | No direct data | Brand trust signal |
The practical takeaway: when targeting Russian-speaking audiences with significant Yandex share, behavioural factors should be a top priority. For global Google SEO, content quality and technical on-page optimisation are more reliable levers — they improve user behaviour as a side effect.
How to improve behavioural metrics
Improving behavioural signals means working on content quality and UX — not manipulating traffic. Bot-driven click manipulation is detected through session patterns: unnaturally uniform timing, IP clustering, zero in-page interactions, and absent scroll events. Real metric improvement requires a different set of tools.
Include a number («5 ways»), a question, or a clear benefit. A/B-test snippets via GSC → Search Appearance. A 1–2% CTR gain delivers measurable traffic impact over time.
LCP above 4 s is the leading cause of technical bounces. Optimise images, add a CDN, configure caching. Every 100 ms improvement lifts conversions by approximately 1%.
A table of contents, clear H2s, and short paragraphs help users see structure and stay. Tables, diagrams, and embedded video each anchor attention longer.
Relevant links to related content increase pages per session. A «Related articles» block at the end of a page is one of the cheapest audience-retention tools available.
FAQ, HowTo, and Review Schema create rich snippets that occupy more SERP space and lift CTR without changing position. Quick start: FAQ Schema on pages targeting question-based queries.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "What are behavioural SEO factors?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Behavioural factors are metrics that measure user interactions with search results and pages: CTR, dwell time, bounce rate, and pogo-sticking."
}
}
]
}Myths about behavioural factors
- "Google uses bounce rate from Google Analytics." No. GA is a separate product; its data is not fed into the organic ranking algorithm. Google collects behavioural signals through Chrome and its own search interface.
- "Higher CTR always improves rankings." Not always. CTR is position-normalised and niche-normalised. A CTR above the average for your position is a positive signal; below average is negative.
- "Behavioural factors take effect immediately." No. ML models accumulate data over weeks. Rapid rank changes after snippet edits are usually coincidence or a temporary A/B test effect.
- "Bot manipulation is an effective tactic." Algorithms detect non-human patterns: uniform click timing, IP clustering, zero in-page interactions, and absent scroll events.
- "A high bounce rate is always bad." Not always. A user who found their answer in 30 seconds and left — that's mission accomplished. Bounce is a problem only when the user leaves unsatisfied.