Showing posts with label google algorithm. Show all posts
Showing posts with label google algorithm. Show all posts

Sunday, October 7, 2007

SERPs taking a bumpy ride on the Good Ship Google - could an update be forthcoming?

Anyone who read my previous blog post about my prediction of a major algorithm change (which was way off) knows that I track keywords using a little tool that I created. It gives me a variance figure that gives me an idea of how volatile the current SERP positions are (read the article to get more information about how the tool works). The higher the variance, the more change there has been to my keywords that tend to stay in fairly stable SERP positions. When the variance is high, I can make a pretty logical guess that something is going on over at the Googleplex.

A variance of 5-10 is fairly normal, and figures as high as 20 usually show some minor change in the SERPs, possibly because of a small algo change by the engineers over at Google. In the last week, I've run the test on my little application three times, and all three results showed a variance of at least 20. This morning's test showed a variance of 61, a number I've never seen before, and 26 points higher than any previous test. There are a couple of possibilities:

  1. Google engineers are testing a small/medium/large algorithm change
  2. Some seriously heavy hitters in the industries for which my keywords are targeted (about six of them) are pushing the SERPs around with some brute force SEO tactics.
  3. The datacenter I use is being updated, and the SERPs are bouncing around during the update.
I analyzed the keywords I use to run my tests, and about 25% of them are completely non-existent in the SERPs. These are keywords and websites that have had stable results for quite a while, so I'm leaning towards option 3 as the reason for the large variance. If I see another large variance in the next day or so, and my keywords that dropped out of the SERPs are back in, then option 3 is definitely the reason.

I don't take option 2 seriously, for the simple fact that no one should/could/would have the power to manipulate the SERPs to that degree.

If option 1 is the reason, dare I say that we'll be seeing a toolbar update by the end of the month? I do see some harbingers of that:
  • The large variance in my SERP tool - this indicates that SOMETHING is happening, I'm just not sure what
  • In the last week, I've chatted with several webmasters that have had their PR dropped by a point or 2. These changes are showing up on the toolbar, and have propagated to all the data centers.
  • Multiple backlink updates at a frequency we haven't had in the past. Until July of this year, I expected a backlink update about once per month, but I've counted at least 5 in the last two months. Possibly this is how Google will be working BL updates moving forward, or it may be a sign of often updated data - a signal that the enginerds at Google are running real world tests.
  • A lack of blog posts from Matt Cutts - if he's busy working on real projects instead of posting to his blog, then something is happening. Caveat: whatever he's working on could be completely unrelated to Google SE stuff. For all I know he's managing and working on a project completely unrelated to PR, the SE, etc.
NOTE: I personally don't care when the next toolbar PR update occurs other than the fact that it will quiet the clamor I hear on a daily basis about the update and when it is going to happen. I just want webmasters to get back to their jobs of building good websites with great content instead of the obsession I'm seeing over Pagerank. For those so worried about PR, I have a PR1 site that ranks in the top 5 for a keyword that has a difficulty of 89 using this Keyword Difficulty Tool.

Friday, September 21, 2007

I'm predicting a large Google algorithm change this weekend or next

I created a little tool for myself that helps me track keywords that I'm optimizing on, and the websites associated with those optimization campaigns. This is a tool used offline, and I have to enter data into it manually. It works well when gauging the progress of some of my optimization campaigns. I wrote it myself using Visual Basic 6, and it has some nifty features.

One thing I use it for is to track keywords that I no longer perform optimization for. The SERPs for these keywords are fairly stable, moving up and down through the various Google dance steps. In general, they trend down, simply because I no longer focus on them. I can generally tell when Google is testing an algo tweak because these keywords bounce around in the SERPs then settle back to roughly their same previous positions.

I created a formula based on these keywords (there are about 100 of them) that shows the variance in SERPs over time. The formula produces a number that gives me a basic idea of the current volatility of the Google search results. The number can be anywhere between -100 and 100, with zero being absolutely no change from the previous week (I normally manually add the updated SERP data once per week, but the formula will work equally well on a daily, hourly, monthly, or yearly basis. The shorter the time period, however, the more pronounced are any small changes.)

Normally, the variance figure is between 5-10 (or -5 and -10), which means a variance of 5-10. (*Note-I'm not a math whiz, so if my terminology is incorrect, you know where you can take yourself). During some of the obvious algo tweaks, the variance goes as high as 15 or 20, but usually drops after one week. Since the formula uses about 100 keywords, a severe change for one keyword doesn't alter the end result as severely as it would were I only to use 5 or 6 keywords. When the variance hits 15 or 20, that means that I've had some fairly significant change in SERPs for my keywords. The change may be up or down because the direction of the SERP change isn't as important to me as the actual amount of increase or decrease.

Two weeks ago, I started entering SERP values in as often as possible (often once per day). I've been paying close attention to the variance looking for signs that Google is doing algo testing. Twice in the last two weeks, I've seen major SERP changes that showed variances of up to 35. Even factoring in the changes introduced by doing it nearly daily, that shows at least a minor algo change has taken place. My opinion is that those were instances of Google nerds testing algo changes in preparation for a larger algorithm change.

I may just be reading the tea leaves incorrectly, but I'm expecting the algorithm changes that have been tested over the last month or so to be implemented either this weekend or next. If it doesn't happen this weekend, and next week is fairly quiet regarding SERP movements, watch out - we may be in for a major algorithm change.

(For those who are going to ask the obvious question: no I don't see a toolbar PR update happening for quite a while still.)