Google like to keep these things secret but the two engineers at the heart of Panda, Matt Cutts and Amit Singhal, gave us some strong clues in an interview with Wired.
Cutts and Singhal revealed their process which I’ll summarize as:
• Conduct qualitative research (that’s speaking with individuals and not a big questionnaire) to find out which of a sample of sites they considered to be low quality and why.
• Use the results to define low quality sites with the factors that Google can measure. This gives Google a mathematical definition of low quality.
If we start here, we can think of a number of factors that Google might be able to measure to define low quality, including:
• A high % of duplicate content. This might apply to a page, a site or both. If it’s a site measure then that might contribute to each page’s evaluation.
• A low amount of original content on a page or site.
• A high % (or number) of pages with a low amount of original content.
• A high amount of inappropriate (they don’t match the search queries a page does well for) adverts, especially high on the page.
• Page content (and page title tag) not matching the search queries a page does well for.
• Unnatural language on a page including heavy-handed on-page SEO (‘over-optimization’ to use a common oxymoron). Eg unnatural overuse of a word on a page.
• High bounce rate on page or site.
• Low visit times on page or site.
• Low % of users returning to a site.
• Low clickthrough % from Google’s results pages (for page or site).
• High % of boilerplate content (the same on every page).
• Low or no quality inbound links to a page or site (by count or %).
• Low or no mentions or links to a page or site in social media and from other sites.
If any of these factors is relevant to Panda, it is unlikely that they will be so on their own.
Multiple factors will likley be required to get ‘Panda points’ (and points do not mean prizes in this game). Panda points will be added up. Cross a threshold (the Panda Line) and all the pages on your site seem to be affected. This includes quality original pages being ranked well below useless scraper sites that have stolen your content.
Google have said that “low quality content on part of a site can impact a site’s ranking as a whole.”
It's important to define the difference between an algo change and a penalty.
A penalty must be served if it has a time limit and lifted if it is to be removed.
An algo change exists and its results will continue until it is changed, your site changes (or your site gets whitelisted).
Panda is an algo change but no ordinary one. It's an algo change that works like a penalty because if your site crosses the Panda Line then the whole site is affected, quality pages too.
Panda is penalty by algo.
Thanks to Franz Enzenhofer for pointing out a misreading in an earlier version of this article of Matt Cutts use of the word 'block' in the Wired interview.
Cutts and Singhal revealed their process which I’ll summarize as:
• Conduct qualitative research (that’s speaking with individuals and not a big questionnaire) to find out which of a sample of sites they considered to be low quality and why.
• Use the results to define low quality sites with the factors that Google can measure. This gives Google a mathematical definition of low quality.
If we start here, we can think of a number of factors that Google might be able to measure to define low quality, including:
• A high % of duplicate content. This might apply to a page, a site or both. If it’s a site measure then that might contribute to each page’s evaluation.
• A low amount of original content on a page or site.
• A high % (or number) of pages with a low amount of original content.
• A high amount of inappropriate (they don’t match the search queries a page does well for) adverts, especially high on the page.
• Page content (and page title tag) not matching the search queries a page does well for.
• Unnatural language on a page including heavy-handed on-page SEO (‘over-optimization’ to use a common oxymoron). Eg unnatural overuse of a word on a page.
• High bounce rate on page or site.
• Low visit times on page or site.
• Low % of users returning to a site.
• Low clickthrough % from Google’s results pages (for page or site).
• High % of boilerplate content (the same on every page).
• Low or no quality inbound links to a page or site (by count or %).
• Low or no mentions or links to a page or site in social media and from other sites.
If any of these factors is relevant to Panda, it is unlikely that they will be so on their own.
Multiple factors will likley be required to get ‘Panda points’ (and points do not mean prizes in this game). Panda points will be added up. Cross a threshold (the Panda Line) and all the pages on your site seem to be affected. This includes quality original pages being ranked well below useless scraper sites that have stolen your content.
Google have said that “low quality content on part of a site can impact a site’s ranking as a whole.”
It's important to define the difference between an algo change and a penalty.
A penalty must be served if it has a time limit and lifted if it is to be removed.
An algo change exists and its results will continue until it is changed, your site changes (or your site gets whitelisted).
Panda is an algo change but no ordinary one. It's an algo change that works like a penalty because if your site crosses the Panda Line then the whole site is affected, quality pages too.
Panda is penalty by algo.
Thanks to Franz Enzenhofer for pointing out a misreading in an earlier version of this article of Matt Cutts use of the word 'block' in the Wired interview.
No comments:
Post a Comment