Google Adwords Keyword Search Statistics - Strange happenings

Search Engine Optimization people use the keyword data to select keywords to use in optimizing their Web pages. Adwords advertisers use the data to determine how to word their ads. Should we use "Vitamin" or "Vitamins?" "Legal" or "Law" etc. People who make advertising based Web sites may use those keywords to decide what topic to push on their Web sites. Should I write about country music or Christian Music? So the numbers had better be right. A lot of people depend on them. But here is the problem - The numbers for February show a great big crash relative to the average of previous 12 months! It is systematic across a large number of keywords I tested (63) regardless of topic or previous popularity. The sole exception was the keyword Facebook, which got 83,100,000 searches in February versus 45,500,000 for year round average.  The extent of the crash is not systematic. for different keywords. The keyword Sex, which led the pack, got 124,000,000 searches on average over each of the last 12 months, but only 90,500 searches in February, a huge difference (no there are no zeros missing or added - you can check the adwords tool for yourself.

The keyword girls likewise crashed from 55,600,000 year round monthly average to 1,500,000  searches in February. The keyword Sears however, had a less precipitous drop from 6,120,000 to 2,740,000 searches in February, while the keyword Panama supposedly went from 3,350,000 average searches per month over 12 months to 165,000 searches in February. The phrase Panama Canal dropped from 201,000 to 110,000. Overall, there was a correlation of only 0.42 between overall search frequency results and February search frequencies as reported by the Google tool, but most of the February results are much lower that the year round averages. A correlation of 0.42 means that the results for the 12 month average predict only about 16% of the variance in the February results, which means, if the numbers are correct, that using the overall results for complex calculations like KEI - Keyword Effectiveness Index is not going to help you much in planning an ad campaign or a search engine optimization strategy. Search frequency varies from month to month, but that much variability would be really exceptional.

And why the big drop? It appears to be consistent over almost every keyword, though the size varies. Is Google losing their search share? Not according to the rating companies. Is sex going out of style? Maybe, words related to sex seem to all have a lot fewer searches in February, but so do many others.

Is February a bad month for Web visitors? Not particularly. See Weekend and Summer Web traffic depression.

Did Google change their reporting criteria? Could be. Are the numbers for last month calculated in a different way than the numbers for the 12 month average? Could be. Is Google showing us year totals instead of 12 month averages? Not likely, since in some cases the numbers for February are close to the 12 month average. Google's help specifically says these are average search frequencies per month. But anything is possible.

Are the numbers just wrong? Could be.  Here are numbers given for two different keywords (keep in mind that the numbers are rounded to the nearest hundred):

Maybe people just are not using search engines.

Can you see something odd there?

Check it out for yourself.

Ami Isseroff

March 16, 2009