[Live] SMX Sydney ’13 – Google Webspam – What are they looking for?

Speaker Profile:

Name: Brian White

Company: Google

Twitter: @brianwhite

Google+: https://plus.google.com/117362177282595168386/posts

Notes:

• Brian White is the program manager of the Search Quality Webspam from Google
• Brian works in the manual webspam team in Google
• How Search Works infographic is up on screen – It is to provide more transparency
• Search quality is a large group in Google broken up in to two teams: Quality Evaluation & Manual Webspam teams
• Quality Eval – They tackle ratings for algorithm development (no changes to rankings)
• Ratings are aggregated based on manual side by side evaluation on algorithm changes
• Manual Webspam – they do manual investigation, actions (demotions/removals)
• Manual webspam teams investigate and find out how people are trying to game the system, result of their work ends up with a manual action
• Live spam screenshots – check it out! Google are providing real time removal of spam websites. Check and see what type of websites are getting removed from the index.
• Brian is showing a graph on manual spam actions from 2004 – 2012. They are committed to fighting spam and created several offices around the world
• Google classifies their spam in to types: pure spam, hacked sites, legacy, unnatural links, cloaking, user generated spam, etc
• Brian just handed out a collectible android by guesssing Parked Domain girl
• Brian is now showing some examples or Pure Spam – mainly in the casino vertical
• Links embedded between unreadable content articles are classified for pure spam
• Hacked sites, cloaking and sneaky redirects are considered as spam too
• Typical hacked site behaviour would be injection of gibberish content with links without the webmasters knowledge
• The footprint that web spam leaves is astonishing!
• Brian White recommends using Fetch as Googlebot – see what google sees with their user agent
• Battle thin content by adding value to the user: provide useful features. Brian is using example of Amazon Light
• Create functionality that is useful to users, Amazon Light generate tonnes of links and mentions for years
• Doorway pages are dreadful webspam – Brian demonstrates a crushing rabbits example on a rock crushing business
• User generated spam on social bookmarking sites are being looked at by Google. Blog comment spamming spam – anchor text links as identity/name is also being looked at closely by Google.
• Hidden text & keyword stuffing not only happens in english but in other foreign languages as well
• They look for patterns
• Automatically generated pages such as infinite Calendar entries on a website can trigger a manual action – use robots.txt to block those pages
• Digg.com was not a manual penalty, it was a mistake
• In 2005, Google accidentally removed all .com.au websites from the index – shit happens, mistakes do happen
• Messages sent via GWT spiked significantly in 2012 – Google’s way of providing more transparency
• Reconsideration requests have also been increasing between 2006 – 2012
• Brians advice when you do a reconsideration request is to be frank and honest. Be straightforward and don’t lie.
• Example Brian used: Hi…Yes we buy links from XXX. This backlink is auto generated and cannot change html code. Regards…
• Google ignores shit like that
• Algos are awesome but not perfect, humans tend to be perfect but with no scale. Together, they provide better results.
• Hints and reports from webmaster tools on spam get some priority – starting reporting dodgy sites via GWT!
• There is going to be a major crackdown on spam in 2013, watch out blackhatters
• Next penguin update is going to be significant!
• Penalties tend to be very granular, they tend to penalise page level and not site level

Disclaimer:
Please excuse the typos, broken links, incomplete sentences, etc.