Social networking is stickier than search, to use some marketing terminology. You have to sign up and commit serious time to it, which is why Google is ramping up all sorts of side businesses in the hopes that something does stick. Google’s Android operating system is about the stickiest thing it offers
Google created Gmail, which is somewhat sticky, and why it’s working on GoogleTV, which may ultimately prove even stickier once the dust settles on this market. But even Facebook will be facing new competition. Social networking and corporate collaboration are just getting going.
With as many services as Google has, Adwords accounts for almost all of their revenue-a whopping $23 billion in 2009. Their flagship advertising platform is showing no signs of slowing down, but Google sees the writing on the wall and is being proactive preparing for “life after Adwords.”
The numbers and trends point to continued success of Adwords as the largest Internet ad platform in history. Those numbers are, however, short term, and Google is not waiting for the trends to turn down before doing something about it.
Many of their actions over the last few months (and possibly longer) would be considered by most to be harmful to their most profitable product, but that’s the point. Hit it hard now while it can withstand the punishment.
Move Adwords down on the search results page and add revenue-free distractions while it’s still generating billions of dollars.
Move one of the most successful technology company executives in history to another department so she can work her magic there.
Move your focus from something that has worked for a decade to something that will likely work for the next decade.
A new style of data management is emerging, where all the data available is stored without enforcing a pre-defined data structure, and kept online for any question to be asked. This approach has been dubbed schema on read, since the data is interpreted at query (read) time, rather than when the data is loaded into the data store (schema on write). The advantages are that loading data into the database is cheap, since there is no clean/transform/index step and that any question can be asked of the data. But this only works if the underlying data storage and compute engine is powerful enough to operate on a large dataset in a time-efficient manner.
This is where the open source Apache Hadoop project comes in, with its distributed file system (HDFS) and compute engine (MapReduce), both based on systems that Google invented. HDFS provides the raw underlying storage on commodity hardware for massive datasets (which may be petabytes in size), and MapReduce provides the programming model to run computations over significant portions of the dataset in parallel.
Ebay used Hadooop to dramatically improve the buyer’s search experience and better match them to sellers, thereby improving sales and profits.
Using Hadoop, Yahoo is able to do research for advertisement systems, optimize the content that users see, and prevent spam. Or consider Facebook, which uses Hadoop for reporting and analytics for understanding user behavior. Facebook was able to test the effect of introducing the “Like” button, for example, by testing it on large subsets of users. Facebook’s analysis showed it increased participation compared to those who didn’t have it, so it rolled the Like button out to all users