Personalization of search results is very important to the future success of any search engine, and its an effort well underway with many of the big 3 search engines as we speak.
If its true that the key success factors of search are:
a. reliability
b. ease/speed of use
Then it stands to reason that the more often a search engine optimization provides searchers with the exact information they are looking for, the more often they are to use that search engine in the future, and the less likely they are to switch to another search engine.
Personalization is not however, a magical phenomenon or crystal ball interpretation. It is purely based on observed patterns, and resulting probabilities. To put this into perspective, here are a number of patterns that search engines are likely to track, and that will permit search engines to calculate probabilities - namely the probability of increasing either reliability or speed of use.
1. Location Based Personalization - serve you personalized results based on your current location (eg. pizza). Eg. search engines have a very good idea where you are (based on IP) when you perform a search query. So someone searching for "pizza" from a location in London, is very likely looking for a pizza location within 10km (6 miles) of their inferred location. This works well for some types of keywords, and not as well for others (eg. paper manufacturers).
2. Interface Based Personalization - serve you personalized results based on the interface you are using.
Eg. if a search engine knows you are performing a search for "hospitals" from a BlackBerry, they can typically surmize you are looking for a hospital within a given radius of your current location, and could reorganize the results with the closest hospitals first. Keep in mind, small screens aren't great for general surfing or researching. Large screens are much more efficient.
3. Query History Based Personalization - serve you personalized results based on a logical sequence of keyword queries.
Eg. if you performed a search for "crayons" 10 minutes prior, and now search for "pens", personalized results should show gel pens results above Mont Blanc pens.
4. Input/Output Mode Based - personalize the mode through which you receive your results.
Eg. if you use Goog411 and input your search query by voice command, then you're more likely to want to receive the results via voice also (or at least a combination of voice and visual output). This will become more of an issue going forward.
5. Time Based Personalization - serve you personalized results based on times it perceives you are typically working.
Eg. searching for "Trains" has a different meaning at the office (think - book train tickets to Montreal), then when you're at home with the kids (think Thomas the Train). Perhaps Google has identified the pattern that most of your queries are business oriented between 9:00am and 6:00 pm, and more child oriented in all other hours.
6. Individual User Behaviour Based Personalization - serve you personalized results based on your actions in the past. Eg. Did you click on this site in the past in response to a similar query? Did you abandon it immediately, or stay on it and interact with it? Your previous interaction with the site offers insight into whether its a better match this time or not.
7. Group Based Behaviour Based Personalization - serve you personalized results based on the actions of all who have made similar queries in the past. Eg. If 90% of people who click on a certain search result in response to a specific search query abandoned it after 5 seconds (while others hold user attention much longer), then there's a high probability that the listing in question is not highly relevant to the search query and the search results should be reorganized as a result.
8. Social Search Based Personalization - imagine Google or Yahoo knowing who your friends are (perhaps through Facebook, gmail, or any number of social media sites).
Eg. Then imagine the search engine can either:
a. aggregate all friends positive and negative reviews of certain sites, and reorganize the search results with those sites with the most positive reviews first.
b. aggregate all friends user behaviour patterns associated with sites appearing in the search results, and then reorganize your specific search results accordingly
9. Virtual Personas/Assistants Based Personalization - serve you search results through a virtual persona you created.
Eg. Think of Star Trek or Terminator if you will. When you talk to a computer, it feels uncomfortable. We often feel the need to assign a personality and persona to it. Now imagine we can customize the appearance and sound of that persona. You can choose the accent, the sex, the age, and much more.
Conclusion:
While it is possible to improve the efficiency of search through each of the personalization methods above, they infact work best when operated in conjunction with one another, acting as a checks and balances mechanism. When used in conjunction, the inferences truly become more probable, and lead to dramatically better search results.
If you think this type of personalization is a long way off - think again. Google, Yahoo, and MSN all are trialling various means of personalization. See for yourself - perform a few search queries, then ask a friend in another part of the country to perform the same searches and compare results.
If its true that the key success factors of search are:
a. reliability
b. ease/speed of use
Then it stands to reason that the more often a search engine optimization provides searchers with the exact information they are looking for, the more often they are to use that search engine in the future, and the less likely they are to switch to another search engine.
Personalization is not however, a magical phenomenon or crystal ball interpretation. It is purely based on observed patterns, and resulting probabilities. To put this into perspective, here are a number of patterns that search engines are likely to track, and that will permit search engines to calculate probabilities - namely the probability of increasing either reliability or speed of use.
1. Location Based Personalization - serve you personalized results based on your current location (eg. pizza). Eg. search engines have a very good idea where you are (based on IP) when you perform a search query. So someone searching for "pizza" from a location in London, is very likely looking for a pizza location within 10km (6 miles) of their inferred location. This works well for some types of keywords, and not as well for others (eg. paper manufacturers).
2. Interface Based Personalization - serve you personalized results based on the interface you are using.
Eg. if a search engine knows you are performing a search for "hospitals" from a BlackBerry, they can typically surmize you are looking for a hospital within a given radius of your current location, and could reorganize the results with the closest hospitals first. Keep in mind, small screens aren't great for general surfing or researching. Large screens are much more efficient.
3. Query History Based Personalization - serve you personalized results based on a logical sequence of keyword queries.
Eg. if you performed a search for "crayons" 10 minutes prior, and now search for "pens", personalized results should show gel pens results above Mont Blanc pens.
4. Input/Output Mode Based - personalize the mode through which you receive your results.
Eg. if you use Goog411 and input your search query by voice command, then you're more likely to want to receive the results via voice also (or at least a combination of voice and visual output). This will become more of an issue going forward.
5. Time Based Personalization - serve you personalized results based on times it perceives you are typically working.
Eg. searching for "Trains" has a different meaning at the office (think - book train tickets to Montreal), then when you're at home with the kids (think Thomas the Train). Perhaps Google has identified the pattern that most of your queries are business oriented between 9:00am and 6:00 pm, and more child oriented in all other hours.
6. Individual User Behaviour Based Personalization - serve you personalized results based on your actions in the past. Eg. Did you click on this site in the past in response to a similar query? Did you abandon it immediately, or stay on it and interact with it? Your previous interaction with the site offers insight into whether its a better match this time or not.
7. Group Based Behaviour Based Personalization - serve you personalized results based on the actions of all who have made similar queries in the past. Eg. If 90% of people who click on a certain search result in response to a specific search query abandoned it after 5 seconds (while others hold user attention much longer), then there's a high probability that the listing in question is not highly relevant to the search query and the search results should be reorganized as a result.
8. Social Search Based Personalization - imagine Google or Yahoo knowing who your friends are (perhaps through Facebook, gmail, or any number of social media sites).
Eg. Then imagine the search engine can either:
a. aggregate all friends positive and negative reviews of certain sites, and reorganize the search results with those sites with the most positive reviews first.
b. aggregate all friends user behaviour patterns associated with sites appearing in the search results, and then reorganize your specific search results accordingly
9. Virtual Personas/Assistants Based Personalization - serve you search results through a virtual persona you created.
Eg. Think of Star Trek or Terminator if you will. When you talk to a computer, it feels uncomfortable. We often feel the need to assign a personality and persona to it. Now imagine we can customize the appearance and sound of that persona. You can choose the accent, the sex, the age, and much more.
Conclusion:
While it is possible to improve the efficiency of search through each of the personalization methods above, they infact work best when operated in conjunction with one another, acting as a checks and balances mechanism. When used in conjunction, the inferences truly become more probable, and lead to dramatically better search results.
If you think this type of personalization is a long way off - think again. Google, Yahoo, and MSN all are trialling various means of personalization. See for yourself - perform a few search queries, then ask a friend in another part of the country to perform the same searches and compare results.
About the Author:
Toronto SEO Firm and internet marketing company offers search engine optimization, search engine marketing, social media marketing and reputation management. When looking for SEO representation in Canada, consider Search Engine People.
No comments:
Post a Comment