- Google does not reveal ranking factors to avoid search being manipulated
- For example, we know first input delay is a tiny factor. Only breaks a tie between two otherwise equal results.
- They have a vested interested in protecting their algorithm
- You can look at patents to get clues
Click Signals
- Subject of many of their patents
- Google has spent years measuring and understanding
- These indicate: Are users happy with results?
The Signals
Be the first click
- Results ranking are a guess by Google about which is best
- Users may prefer a different link which will cause it to rise, potentially based on click signals
- First click indicates user preference
Earn long clicks
- Longer stays indicate greater satisfaction for some queries
- For simple queries (e.g.,
capital of North Dakota), this may not be a meaningful indicator
Be the last click
- If users return to Google, that indicates the page does a poor job answering the query
Does Google use them?
- We don’t know if they use them in search rankings
- They use them in machine learning, but we don’t know how
- Most others (like Bing) admit using them in search rankings
- It may not matter because the signals indicate user satisfaction. Build toward optimizing satisfaction instead of the signals.
- You can’t isolate a signal and test it because there are so many signals being tracked. May provide a hint though.
Optimizing for click through rate
- favicon
- brand (domain)
- URL keywords
- title
- description
- Google rewrites many of them
- rich snippets
Experiments
Description
Added keywords to meta description
In brackets at the beginning of the description ([word1 word2]). Lost 7.5% traffic and lower engagement.
Rewrote entire description
Similar to adding keywords
Removed meta description
Forces Google to write it. Lost 2.99% traffic and mixed engagement.
Add “GOOGLE GUARANTEED” to description
Gained 7% traffic and higher engagement. Don’t do this.
Title Tag
Add “[solved]” to questions that were
Gained 7% traffic
Remove boilerplate (like “Whiteboard Friday”)
Gained 21% when boilerplate is irrelevant
Optimizing for long clicks
Experiments
Adding Videos
Lost 1.39% traffic. Not statistically significant.
Adding related articles at the top
Even linked to competitors if they didn’t have a useful article. 18.5% increase in traffic. Engagement was worse. Google is probably not measuring these engagement ratings. SearchPilot also saw success by adding additional related links (e.g., 4 where they previously had only 2).
Case Study: Domain Authority
Page defines domain authority. Added a way to check a site’s DA on the page that defines and explains it. Bounce rate went down, all engagement metrics improved, traffic nearly tripled.
Notes
- If your content doesn’t support the changes, numbers will “return to earth”
- See SEO Master Class: Advanced Title Tag Optimization (For Any Site) - Moz
Optimizing for last click
Experiments
FAQs
Added FAQs (not marked up with special FAQ markup) and at bottom of page. Increased traffic 18.5% and all engagement metrics.
Case Study: Moz Q&A
Needed to migrate Q&A to new platform. Couldn’t use the same URLs (because of new platform) and needed to migrate 100k URLs. No new content; just a better user experience with increased speed, added breadcrumbs, and increase relevant links on page. 17% increase in traffic. No decrease. “Amazing success.” Engagement metrics increased as well with more pages and lower bounce rate.
Takeaway
User satisfaction is the #1 ranking factor.