Comments on: How accurate and reliable are statistics from keyword research tools? http://websitebusinesskit.net/blog/how-accurate-and-reliable-are-statistics-from-keyword-research-tools.html How To Create Your First Profitable Website Questions and Answers Thu, 08 Sep 2011 14:30:10 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Mark Welch http://websitebusinesskit.net/blog/how-accurate-and-reliable-are-statistics-from-keyword-research-tools.html/comment-page-1#comment-226 Mark Welch Mon, 05 Oct 2009 00:21:35 +0000 http://websitebusinesskit.net/blog/how-accurate-and-reliable-are-statistics-from-keyword-research-tools.html#comment-226 I am not aware of any keyword research tools that are "very accurate," and at best they tend to be "somewhat helpful." There are several issues here. First, no tool is more reliable than the data it is "fed," and neither Google nor Yahoo provide their search data to third parties. WordTracker uses data from a pool of smaller search engine sites, and often shows high traffic levels for very bizarre search phrases. (In my experience, WordTracker's data includes "fake traffic" including traffic from other keyword-research bots.) In addition, even your own data is not complete: you may find that your bidding draws 150 impressions per day for searches for "where dubai" but in fact, your ad might not be shown for 1,000 other searches done each day. (Your data may also be "over-inclusive," as you may be including searches through the Search Network, which aren't done at Google, or traffic on the Content Network). Second, search patterns change over time; the number of searches for "Santa Claus' address" rises in December and probably zeroes out from January to October. In addition to predictable seasons like that, there are random fluctuations as well as "surges" (when Jay Leno mentions widgets, there may be a bunch of web searches in the next 24 hours). There are also "cycles" for certain search terms. For example, searches for "porn" peak on weekends (Friday and Saturday night), while searches for "football scores" peak on Sunday and Monday. For some search terms, it's possible that 80% to 90% of the total searches occur on a single day each week ("lotto numbers"), or even a single day each month or each quarter ("Fed interest rate change"). Another crucial factor to consider is the activity of "fraud operators," who use keyword research tools to identify high-bid keywords; when bid amounts rise, those terms attract more fraudulent activity (because fraudsters can make more money from each click than for low-bid terms), increasing volume. The bottom line is that you should absolutely NEVER rely on any tools to be reliable -- not even within an "order of magnitude" -- in predicting search volumes for specific keywords or even "keyword families." Mark Welch Internet Marketing Consultant http://www.MarkWelch.com/ I am not aware of any keyword research tools that are "very accurate," and at best they tend to be "somewhat helpful."

There are several issues here. First, no tool is more reliable than the data it is "fed," and neither Google nor Yahoo provide their search data to third parties. WordTracker uses data from a pool of smaller search engine sites, and often shows high traffic levels for very bizarre search phrases. (In my experience, WordTracker’s data includes "fake traffic" including traffic from other keyword-research bots.)

In addition, even your own data is not complete: you may find that your bidding draws 150 impressions per day for searches for "where dubai" but in fact, your ad might not be shown for 1,000 other searches done each day. (Your data may also be "over-inclusive," as you may be including searches through the Search Network, which aren’t done at Google, or traffic on the Content Network).

Second, search patterns change over time; the number of searches for "Santa Claus’ address" rises in December and probably zeroes out from January to October. In addition to predictable seasons like that, there are random fluctuations as well as "surges" (when Jay Leno mentions widgets, there may be a bunch of web searches in the next 24 hours).

There are also "cycles" for certain search terms. For example, searches for "porn" peak on weekends (Friday and Saturday night), while searches for "football scores" peak on Sunday and Monday. For some search terms, it’s possible that 80% to 90% of the total searches occur on a single day each week ("lotto numbers"), or even a single day each month or each quarter ("Fed interest rate change").

Another crucial factor to consider is the activity of "fraud operators," who use keyword research tools to identify high-bid keywords; when bid amounts rise, those terms attract more fraudulent activity (because fraudsters can make more money from each click than for low-bid terms), increasing volume.

The bottom line is that you should absolutely NEVER rely on any tools to be reliable — not even within an "order of magnitude" — in predicting search volumes for specific keywords or even "keyword families."

Mark Welch
Internet Marketing Consultant
http://www.MarkWelch.com/

]]>