2009 June 08

Liveblogging SES Toronto: What Relevancy Indicators Are Search Engineers Watching for Today.

Posted by Agustin Vazquez-Levi at 12:07 pm

Moderator: Andrew Goodman
Speakers: Mike Grehan, Dan Zarella

Marios Alexandrou (Acronym Media): New signals SE’s are looking for…

What are the changes going forward?

Phase I: It used to be simple (Altavista) to manipulate results. Searc engines were also faced with the “abundance problem” = Different content sources about the same topic. What should be first? How to differentiate the trustworthy source.

Phase II: Inbound links provides a new signal, so web content creators (not users) have the votes. But, unless you had a website couldn’t help it rank. This was Google’s concept of PageRank and was abused by linkspam.

Phase III: Text and links were strongest signals. Keeping up with new fresh content (resource constraints). Problem was users didn’t only want text results. Now we see video, images and othe types of results depending on the query. An example of the search for “terminator movie” shows all these things: Video and images, movie times, news reviews, etc.

So in Universal search what is the number 1 position? Regular results? Map listings?

What is the effect of Universal search?

1. Promotes alternative content other than just text

2. Push competing text listings down the page with other media.

3. Engage customers in new ways

Universal search results affect user behavior. People interact with these results differently. User behavior will affect results as well.

Toolbars also feed information back to search engines about how users behave with websites. Where are people going: bounce rates etc. Microsoft uses this also its called “Browse Rank”

Examples of Toolbar Data:

1. page discovery

2. gauges the quality of the content Good Signal: Quality content that is regularly visited Bad Signal: Millions of pages on a site never get visited?

Users shift towards social media. People are sometimes starting their web session on social sites instead of search engines. Why use Yahoo answers and wait for someone to answer when you can go to twitter and get answers instantaneously? Also people to people is percieved as being more accurate. If people trust the votes of other people, should search engines trust those votes?

Google Search Wiki, how you can rearrange your results for your personal search. In the future if something is obvious in this data gathered, they will use it for the ranking factors.

Connected marketing: extending beyond computers. iPhone and blackberry is changing user behavior as well. These communities will counterbalance one-sided brand messages. Also applications sidestep web browsers to deliver specialized content. No one really knows how search engines will use the data accumulated with mobiles.

Conclusion:

1. Identify new Signal - careful not to chase things too wildly, careful about how you change things
2. How to address the new signal
3. Establish cross-team measure of success.

Example of loading time and why it should be considered a factor: more page views for ad inventory, increase visibility in SERPs, improves usability reduce bounce rates, raise adwords quality score. Its getting harder and harder to get results.

Textbook SEo wont be enough, Social search will take more room, don’t wait… initiate. marius’s Blog: keyworddriven.com

Dan Zarella’s presentation (hubSpot): Social Search Quality Signals

Dan Zarrella

Websearch is broken. Page to page links are too slow. and news can take days to get indexed. More and more news is breaking on Twitter, not Google, because its not real time. Digg, Reddit, etc is sometimes faster than Google. Especially since each Social news site has its own flavor or audience.

What is the Solution?

69.05% of reTweets contain a link. This shows that for a search engines, retweeting is a perfect pool to draw data from. Dan then shows keywords that appear often in retweets such as: ”new blog post”. People are effectively marketing their content real time on twitter.

What are some flaws in this method?

1. The more links you already have, the more you are going to get. the model is “the rich get richer.”

2. Sock puppet accounts that were opened and not used “coming soon” etc. 52% of twitterers have 0 followers. 54% have never tweeted. 75% have not entered a bio, etc.

How can we gauge a user’s influence in Twitter?

1. Number of tweets divided by number of followers of those who retweeted it: RT/Followers

2. Twitter Authority: Mentions-Per-Day (the amount of times your user name appears in other tweets)

3. 1 & 2 Multiplied

Basically write a lot of tweets and get retweeted often.