In the meantime... do any of you have a formula to estimate the number of times a given phrase was searched on Google over a specified time frame (one month)? I am in a consultant in for an industry that has been SLOW to embrace the Internet as a legitimate advertising medium (I typically work with car dealerships who normally spend thousands of dollars each month to advertise in the newspaper or on TV, but it's difficult to get them to spend much money on Internet promotions). From a sales standpoint it's a very good resource to know the actual number of searches conducted over a given month, so I can present the Internet as a medium to put more advertising dollars into.
So for the last few months I've been using a simple formula to "BALLPARK" the number of times an item is searched. Here is what I've come up with... and I would like some feedback (keeping in mind that I am just looking for a ball park figure).
FORMULA # 1 (Overture X 3 Method)
This formula is based on the theory that the Google Network (including AOL.com, Ask.com, and other search engines that use Googles adwords technology) comprise about 60% of the share of searches (see Nielson ratings [searchenginewatch.com...] ) The Yahoo Network, which includes MSN & Yahoo, compromises about 33% of the share of searches. Because the Overture KEYWORD SELECTOR TOOL gives us the number of times an item was searched on the YAHOO network... we can just plug in the numbers and get a ROUGH estimate of how many people searched for the same thing on the Google network over the same amount of time. Therefore... if you want a very rough estimate, you can simple double the number on Overture and you have the GOOGLE BALLPARK estimate. Triple the number, and you have an aggregate estimate of the number of times that term or phrase was searched on the entire Internet (over one months time).
Example: MARYLAND LEXUS
KEYWORD SELECTOR = 189
Google Ball Park Estimate = 378
TOTAL INTERNET SEARCHES = 567
FORMULA # 2 (A FACTOR OF 1500)
I don't use this formula as much because it's not as convenient to compute. But I once read that the Traffic Estimator calculates the clicks per day based on a CTR of 2%. So in theory, you could multiply the Clicks Per Day by 50 and get the number of times a given item was searched on over a specific day. Multiply that number by 30 and you have the monthly search impressions. Or, to make it easy, just multiple the daily clicks by 1500 to get a search total for the month.
Example: MARYLAND LEXUS
Traffic Estimator Daily Clicks = .4
Estimated Daily Searches on Google = 20
Monthly searches on Google = 600
KEYWORD SELECTOR = 189
TOTAL INTERNET SEARCHES = 789
There obviously some problems with these estimates... and I am not looking for an EXACT number. Quite frankly, Google would sure make my job easier by releasing the actual data... by doing I could more easily sell advertising on their site that would otherwise end up in a local newspaper or on television. Until they do, I've found these formula's to be the best way to estimate traffic. Your input would be greatly appreciated. Thanks.
My feeling is that Formula 2 is more accurate. Just ensure that you use a high enough Max CPC when you do the estimation.
If you want to be even more accurate, you might think of using the API. When you use the estimator in the API, they do show the estimated CTR also. Hence the number of impressions would be more accurate.
However, the system seems to behave a bit better if you add several keywords at once and don't do them all individually.
Since there are several matching options, you also need to run them at once. Consider these keywords:
"widget"
"online widget"
"cheap online widgets"
If you ran them individually, on the query for widget, you'll be counting the other KWs clicks. However, if you run them all at once, G will sort out how many each KW receives.
I recently conducted a study for one of my clients that used a very similar line of thought to the second method you describe.
The goal was to estimate the total search volume for a given market, "widgets". I picked the top X number of keywords based on impressions ("blue widgets", "red widgets", etc), and compared the client's historical results against projected search volume (using your method), then assumed that these results would hold true campaign-wide. The results were surprisingly accurate, I believe, even when compared to the numbers generated by other systems (like your first equation).
Better yet is when you factor in conversion rate and project ROI across an entire campaign.
Very fun!
Then you get the exact impressions.
Of course, you'd have to pay for any clicks :D
EXACTYLY MY POINT, you'd have to pay $$$ to determine the exact amount of impressions. Which, in my case, would make NO sense because I am a consultant for several companies and I do not have anything to sell myself.
Still, I would love more input as to the effectiveness of the forumulas I have used to determine the approximate number of searches done on a given term over the course of a month.
BAYZING
