Google (graphic) How Google Works

Google (graphic) How google works I'm Feeling Lucky Google Search The official word "The sofiware behind our search technology conducts a series of simultaneous calculations requiring only a fraction of a second. Tra- ditional search engines rely heavily on how offen a word appéars on a web page. We Use more than L00 signals, incliding our patented PageRank algorithm, to examine the entire link structure of the web and deter- mine which pages are most important. We then conduct hupertext-matching analysis to determine which pages are relevant to the Specific search being conducted. By combin- ing overall importañce and gquery-specific relevance, we're able to put the mast rel- evant and reliable results first." -Google You write a blog post, tweet, update a site, or otherwise add content to the web. Google bots follow links. If there are no links to your site, typically it will not get crawled deeply or regularly. Page titles are link data Google bots crawling the web come across your post. Google bots won't crawl your site if you tell them not to with a robots.txt are stored in one index use for broad and com- petitive searches. On page content are stored in another index used for obscure and long tail searches. Page content is If links to your site have a nofollow tag, the google bots won't arrive from those links. stored in a reverse index. Once crawled, the page is indexed within seconds. Google may also find your site by being pinged by blog software or xml sitemaps. When you search google you are not searching the active web, but Google's cache of it, which is constantly being updated. Google estimates the domain and page's overall authority based on links. The more links you have from higher authority pages, the greater your own pages' authority. Google is very secretive about its data centers. This is what we know so far. Google's Search Quality team and Webspam Team review and refine algorithms. 10,000+ remote testers rate the quality of their searches. Pages are checked against editorial policies. As long as they are not tagged 'nofollow'. Google solicits spam reports from users. Google gets DMCA notifications to take down pirated work. Penalties are applied and each page now has many pieces of data attached to it that help in user searches. There are over 19 data centers in the US, with another 17 around the world. They can be up to 500,000 square feet and $600 million each. "On most Google queries, you're actually in multiple control or experimental groups simultaneously. Essentially, all the queries are involved in some test." -Patrick Riley, Google search quality engineer User queries Google Google Servers Goor Google suggests keywords based on what has been typed so far. Google servers are housed in standard shipping containers that hold 1,160 servers each Typical Servers Google uses synonyms to look for similar words to include in the search query. Google data centers are some of the most efficient in the world and are carbon neutral. Search Query: 1003043560211 Data centers use between 50-100 megawatts of power and often found near water (cooling). Google may claim millions of results but only 1,000 or less are ever displayed. Result localization: Local websites are promoted in the search results. Initial result set is created. Result set is sorted by authority Google finds relevant ads based on keywords, ad match type and user location. and PageRank and duplicate Advertisers operating outside the guidelines may have their ad accounts banned. pages are removed. With universal search, if Google thinks news results, shopping results, video results, book results, local results, or any other form of vertical search are relevant then they may mix these directly into the search results. Ads are subject to editorial policy If keyword has low search volume or ad generates too few clicks, ads might be automatically disabled. User Personalization: Past websites the user visits are often promoted. Relevant ads are ordered based on earnings potential. (bid x ad quality score) Excessive anchor text manipulation can cause websites to be removed from the results. Favored businesses, per- haps like Amazon.com, maybe be given a discount. Filters applied. Local interconnectivity of the result set: If pages are well linked among other high ranking sites, then their ranking is boosted. For most advertisers the content is already created but sometimes dynamic keyword content is used to make the ad appear more relevant. Some ads also have exten- sions available, like site links, phone numbers, product links, location, etc. Trending: If the search term has a huge burst in search volume and/or lots of recent news results, google might place additional weighting on fresh results. If the ads generate a high enough clickthrough rate, some may be shown above the search results. Multiple pages from the same domain may be clustered together if they all have a high rank. Organic results displayed The rest go to the right rail where they are displayed. and this is all done in less than a second, $00 millions times a day generatmg over $10 bilion a year for Google! PPCBLOG