Long Tail of Search Statistics
Long Tail of Search
Posted by Adam Hayes
Unique search strings accounted for more traffic than all search strings that more than 100 people searched for combined (including keyword sets that had thousands of searches).
To help explain this topic in more detail we will use this sample data.
|Keyword Set||Number of Times Keyword Set was Searched For|
|Red Widgets that Fly||20|
|Black Widgets near Boston, MA||2|
|Black flying widgets that don't float||1|
|I tried to buy a widget the other day that should fly but it couldn't. I'm not sure why it didn't fly||1|
|This is a random question about widgets in Boston||1|
|My widgets sink. Why won't they float or fly||1|
|...(long tail continued for another 300 items...)||1 (X 300 more)
Quantifying the Long Tail of Search. We did some data crunching the other day on all of the Google traffic we received for one site in 2009. Here are some interesting takeways.
- Unique Keyword Sets (or search phrases that where searched for only once during the year as shown with the yellow background above) accounted for 27.8% of all searches
- Keyword sets that sent only two visitors accounted for an additional 15.2% of all searches.
- Combining keyword sets that were searched for only once or twice equaled 43% of all searches received. Almost half of all searches used different keyword sets than anybody else typed in.
- All Keyword sets that sent at least 100 visitors (like widgets and my widgets above) accounted for 23.4% of
all searches. This means that more people performed unique searches than all of the search strings combined that had over 100 people search for them. People are performing longer and more specific keyword searches than before.
- The longest keyword set we saw was 144 words long. That is a long query string to type into Google.
- The average keyword set length was 5.7 words.
- The mode keyword set length was 4 words.
- Not surprisingly, unique keywords sets tended to be longer than searches that occurred more often.
Accept Credit Cards Online »