Call us at +1 (703) 907-9861

Google Panda ExplainedActionable and informative details

Guerilla's SEO Webinars
* indicates required


What is Google Panda?

Also known as Google Farmer

Named after Navneet Panda, a Google engineer

Navneet Panda is an expert at Machine Learning

Navneet has created a Machine Language Algorithm (MLA) for Google

The MLA used by Google is a program that learns from user surveys, which allows it to make predictive choices based on human preferences.

How Google Panda works

Acts as a quality filter on search results

Uses survey data to construct a model of desirable and undesirable site characteristics

Penalty appears to override domain authority

Refined and pushed manually on a monthly basis

Deployed internationally in English

Low scores on some deep URLs can impact the entire domain

Google's Algorithm before Panda

Keyword as anchors to determine relevancy

Link quality

Link volume

Link velocity

Domain age, domain trust

Simple onpage SEO factors

Rolling (live) updates

Google has been trying to achieve

Search result pages for users, not SEOs

Minimize exploitation of the ranking algorithm

Ordinal versus Cardinal number based rankings
ie. Qualitative versus Quantitative results

Determine trust through behavior models

Predict trust before the initial user activity

Increase differentiation from Bing

When did Panda take effect?

  • 1.0 - February 24, 2011 (Monday)
  • 2.0 - April 11, 2011 (Monday)
  • 2.1 - May 10, 2011 (Tuesday)
  • 2.2 - June 16, 2011 (Thursday)
  • 2.3 - July 23, 2011 (Saturday)

UX is about trust

Satisfaction and value are psychological

High trust should manifest itself through increased website engagement

Traditional trust was based on domain age and backlink quality

Panda trust is based on the predictive filter, which judges a site based on factors found on good and bad websites, as determined by user surveys

Panda factors we can monitor

Search impressions and click through rates

Time on site

Actions on site (browse rate)

Bounce rate

Repeat visits

Navigation paths

Onsite search queries

Useful monitoring tools

Webmaster Tools - (Google and Bing)

Analytics -(Clicky, Piwik, Woopra, Google)

Thoughts on split testing

Correlation is not causation (example)

Not enough SEOs split test

Testing should be repeatable

Random internet posters rarely provide enough data to understand the context of their tests, many of which are just conjecture

Tests have limits, understand when best practices will yield greater utility

Consequences of Google Panda

Thin or duplicate content is a low ROI tactic

Link building may not be enough to overcome a quality penalty

SEOs will have to acquire deeper skill sets

Ad blending will have to get more sophisticated

Time lapse updates makes testing almost impossible

Google Panda & Content

Keyword density of the top 3 results is typically just under 0.5%

The average content length of a top 3 search result, is over 2400 words (3C)

Keyword in the title appears to be a more consistent factor than keyword in the url, exact match domains or keyword in the meta description

Data courtesy of Darrin @

Content Engagement

Good content, regardless of length, is engaging

Use of videos can assist time on site

Images stimulate imagination, placement can increase readability

Authorship and the perception of unique content are intangibles that Google Panda is looking for

Good copy sells and earns its own links, the payoff is there and with Panda, the incentives are stronger than ever to provide good content

Google Panda is not about links

"If you only have a hammer, you tend to see every problem as a nail." - Abraham Maslow

Links haven't stopped working, Google has removed the perception of linear rewards for link building