Refining site keyword relevance

As I mentioned in my post of two weeks ago, Google Webmaster Tools displays, in order of frequency, the keywords that Google search has identified in the collection of indexed pages from a website. For this microEnterprise blog site Google considered the keywords August and July to be significant, listing them each in the top thirty results; 28th and 30th in rank respectively. These months appeared so frequently because I had set the post date format and the archive list format to show date in a verbose format. Nearly all the posts on this blog (89 of 104) so far were posted in those two summer months.

I decided I did not want the dates for posts to affect keyword significance or search results, because of their irrelevance to the content of the posts. I changed the date format displays from verbose to numeric.  After nearly two weeks the offending names have dropped from 28th to 31st, and 30th to 38th.  I expect that they will continue to drop further as Google proceeds to re-index the various individual post pages over time.

Webmaster Tools - Keyword view

Google Webmaster Tools shows the significant keywords of a site, as seen by the Google search index. This will depend in large part upon which pages or posts that Google has actually crawled and indexed. Rarely is it every page.

Apart from the question if Google sees as much of your site as you wish it to, there is the issue of how Google sees a site.  The Google Webmaster Tools keyword data is a useful way to discover keywords you do not want to be given Google search significance. The screenshot below shows the resent keyword results for this blog after seven weeks, eighty posts, and 80 percent of the posts being indexed.

As you can see the red arrows point to "august" and "july" being counted as keywords in the top thirty of the most significant. This is not what I want, and it comes as a surprise. The month in which a post was written has little is any relevance to the content of the posts.


Now that I am aware of the problem, I can investigate it and devise a remedy.

To me it looks like the cause of this problem was my blog setting choices for date formats.  I opted for a verbose display, with the names of the months spelled out for each blog post and for the blog archive display. It is a simple matter to change these settings so that dates are displayed in a purely numeric format. Actually doing that immediately changes the date format on the blog posts, but so far it seems to have had no effect upon the blog archive date format.

It makes sense to me that part of keyword optimization is to eliminate the keywords you do not want.

Google webmaster tools - Search queries

Below is a screen-shot of today's available data for this microenterprise blog on Google Webmaster Tools, Search Queries page. The red numbers one through seven have been added by me for ease of understanding. This page on Google Webmaster Tools provides search data for one day only, July 9, 2010.

The reason for this limited amount of data is the newness of the site registration with Google Webmaster Tools on July 6th. See prior post. Since there is only one day of data available, the date range pointed to by number 1 is fixed in this example.

Presumably, once more data is available, the date range can be user selected. The most important parts of this data set are:

1. Date range
2. Total number of search strings that Google considered relevant enough to return, during the date range.
3. Total number of times the blog appeared in Google search results, during the date range. This is called "impressions" because the Google user would have actually had the opportunity to see the site's title and blurb whether or not the site link was clicked and visited.
4. Total number of times the site link was clicked and visited from Google search results page.
5. The specific search string(s) that Google responded to.  In this example, there was only one, "visitor management."
6. The high-low search position range that Google served as a result for this blog.  Number one is the highest search position, appearing at the top of the list on a Google search results page.  The are generally never more than ten organic search results (not including paid search results) per page with Google, and sometimes less.
7. The specific page that was considered relevant enough to serve up as a search result.

Google served this blog page, /2010/07/website-configuration-management.html on the first-page results at least twice, but certainly less than ten times, on July 9, 2010. But it was delivered in response to the search query "visitor management," which is fairly irrelevant in the page itself. Google adaptive and self-correcting heuristics will soon eliminate this page for that search inquiry entirely. Indeed, less than four days later, as of today, the blog page is nowhere in sight when that search is repeated.

---

Screenshot 1 - 20100715-01

Tom Fox
Louisville, Kentucky
Tom Fox on Twitter

Google webmaster tools - data time lag

I went through the process of verifying ownership of this microenterprise blog of mine with Google Webmaster Tools late evening on July 6, 2010. The first date was returned by Google overnight to be available first thing Monday July 12, 2010. There was one day of data available, and that was for July 9, 2010. It seems, therefore:

1. There is a 24 to 48 hour delay after registering a new website with Google Webmaster Tools before data begins to be collected.
2. Once data collection starts, it is prospective only, and accumulates with the passage of time. Historical data prior to the start of data collection is simply not available.
3. It takes a while before a significant amount of data is collected, for most low-traffic sites. Delaying the set-up is not an advantage here.

Tom Fox
Louisville, Kentucky
Tom Fox on Twitter