Forum Moderators: martinibuster
Obviously, there are a lot of difficulties in trying to measure this, not the least of which is the impossibility of knowing the details of the Google algorithm. But I tend to follow the adage that says, for everything that is difficult to measure, there is some measurement that is better than nothing.
The measurement I've been using for a while now I dub the Google Stability Number. It's a percentage, where bigger is better, and you're extremely unlikely to achieve 100% on this stability scale. I tinkered with more complex measures before settling on this awfully simple one. It's more complicated to explain than it is to actually calculate, since there are basically just 3 numbers involved.
First, you need the total count (call it TP) of pages that got hit by a referral from Google during the time period of interest. You also need to know the total number of referrals (call it TR) during that same time period.
The third number may not just pop out of your webstats package; you may have to get out the old adding machine. This number is the count (call it HP) of the fewest number of pages whose referral counts sum to more than TR / 2. The Google Stability Number is then:
{HP / TP) * 200 An Example
The easiest way to think of this is to generate a report of all hits from Google referrals, grouped by URL, sorted by hit count, with a nice total at the end. It might look something like this (only much longer, I hope!):
3 /woobie.html
3 /woobie-works.html
4 /wickie.html
5 /woot-woot-html
8 /sternum.html
10 /sitemap.html
15 /home.html
-----------------------
48
So we got 48 referrals (TR) from Google in that time period, and now we want to calculate HP, the fewest number of pages that accounted for more than half of that traffic. Well, half of 48 is 24, so we just started with the biggest numbers (15 hits for "home.html") and start adding until we get a number bigger than 24.
Well, that was quick, because you can see that only 2 pages here account for more than half the Google referral traffic. So, HP is equal to 2, and the total pages getting Google referrals (TP) was 7. That's all we need to calculate the Google Stability Number for this website:
(2 / 7) * 200 = 57%
A number I would be very happy to have, on any AdSense website of much depth (that is, a lot more than 7 pages).
What Does it Mean?
This is basically a very simple measure of how "spread out" your Google referral traffic is. The smaller the percentage of pages that accounts for half your Google traffic, the more you can lose if a Google algorithm tweak knocks you out of relatively few of the search terms that you rank for.
If the Google Stability Number was ever 100%, that would mean that (unlike virtually every other site on the web), you have an awfully flat distribution of Google referrals across your URLs. The closer your Google Stability Number gets to 0, the more you know that all your Google referral traffic eggs are pretty much in one basket, and one algorithm change (or improvements by a couple of competitors) can take you out at any time.
As with most stats, the smaller your website and the lower the traffic, the more volatile the number is -- so you have to sum traffic over a larger period to get a stable view of what's happening.
Problems/Limitations
This is a really simple, dumb number, and there are a lot of problems with it.
First, it doesn't take into account your total # of URLs. A website whose number is 5 but has traffic spread across 10,000 pages obviously has less to worry about than another website whose number is also 5 but only gets referrals to 25 pages.
Second, there's no attempt here to incorporate per-page revenue. So, for example, I would be less worried about losing my biggest-traffic page to a competitor or algorithm change if I happen to know that the bulk of my income comes from a set of lower-traffic pages that convert well into high-dollar clicks.
Third, as usual, if you don't know your webstats package intimately, it's real easy to calculate totally bogus numbers. For example, you don't want to sum hits to external references (.jpg, .css, .js, etc.) even though your stats package might be simple-minded to think you do.
Hopefully, it's obvious that this number is really aimed at content-heavy sites. If you're building a new website every day, or scoring the quick buck off of 2 or three keywords, this thread definitely ain't aimed at you :-)
My Experience
Despite the severe limitations, I find this simple number useful. I track this number as I'm building new content for a website. I generally want to see it going up, and that tends to guide content development (e.g., more time spent growing the middle and end of the Long Tail and less time spent trying to rank higher for the "big keywords").
At a certain point, it may stop going up, and I'll begin to suspect that I've dug all the keywords out of the ground I'm going to find for this website (or maybe I need to go back and do some more out-of-the-box thinking about locating new keyword spaces to mine).
I also like to include this number on the report of yesterday's activities that I look at every morning. For mid-sized (not tiny, not huge) websites, this number often provides a quick warning that something has shifted. Getting knocked off page 1 for a couple of top earners, or suddenly getting a lot of new traffic for a previously morbid page will often visibly change the number. There are other, more complex measurements better for looking at how traffic is flowing into your search term bins, but it's nice to have a single number for at-a-glance checking and comparing.
Although it's not necessarily a great number for comparing one website to another, I'll share the numbers for the website I'm currently developing:
TP = 123
HP = 9
(9 / 123) * 200 = 14.6%
Not as stable as I would like, but I'm not even close to running out of relevant search terms to start attracting traffic for.
Questions
What do you use measure your website's sensitive to Google ranking changes? Do you take that into account when building content? And, if you're really bored, calculate and share your Google Stability Number (preferably mentioning the total page count).
Answers to your questions:
What do you use measure your website's sensitive to Google ranking changes?
I don't, because there are too many different rankings on too many pages. I just keep an eye on daily Google referrals and, to a lesser degree, on the ratio between Google referrals and Yahoo/MSN referrals.
For what it's worth, I've had witnessed only two dramatic changes in traffic as the result of Google rankings since I launched my site at its current domain in 2001: Once when my daily traffic jumped about 30% after Google's Allegra update, and again when my Google referrals dropped by 70-90% between late March and late May of last year (probably because of the much-discussed "www vs. non-www" canonical issue that I solved by editing my .htaccess file).
Do you take that into account when building content?
No. I wouldn't know how to do that effectively even if I wanted to, since Google rankings are so unpredictable.
And, if you're really bored, calculate and share your Google Stability Number (preferably mentioning the total page count).
I don't think I have the patience--especially since my stats package tracks different Google TLDs separately, which would require even more time with my pocket calculator.
Here, we're usually too busy agonizing over whether the day's statistics are up or down at 10 a.m.
:-)
Absolutely my motivation for posting -- I just want to see some discussion of non-CTR stats here before I die :-). Next time, I'll make the title "How to make $400/day without creating anything of value whatsoever!"
:-)
This info is good to know. In certain niches it's hard to diversify if you already do as well as you think you can in Yahoo and MSN. Aside from tell a friend and other promotions, there's not much to bump up the traffic levels. As far as the the other engines, their potential for diversifying your traffic seems like it would be limited. Hitting number one in MSN isn't as fun as hitting it on Google.
What do you use measure your website's sensitive to Google ranking changes?
Nothing really, I just try to look at my server logs and see where the referrals are coming from. I know I should invest in a stats program that could show trends in graphs and all. Anyone know a free program that does it? I use analog 5.30 but I don't think it does that.
Do you take that into account when building content?
Not really, I just try to build the most reader friendly pages I can, that offer something of value.
Should have said this sooner - nice post. I'm going to give your calculation a try.
On the other hand, in the sites I manage and SEO for clients -travel related, larger, aimed for large market terms- your measurement makes a lot sense.
Thank you for a very good idea.
Does your model assume that every webpage has the same chance of being hit by an algo change?
Cause I am not sure if that is true.
Perhaps Google implements algo changes that are aimed against specific types of pages (for example pages with no or little content, or pages containing certain words). If that is the case the stability of your earnings might also strongly depend on quality factors. For example; if 90% of my earnings comes from only 10% of my pages, but these specific pages happen to be algo friendly (that's why they make so much money in the first place) your model gets a bit more complicated...
it reports unique vistors from google, pages indexed by google bot and adsense bot among other stats. Quite useful and GPL.
Thanks for sharing that.
Dr
80% of my search-engine traffic comes from one-shot keyphrases...
Yeah, that's the longtail. Those are highly niche phrases sometimes composed of five or six words.
If you want to stabilize your traffic by being less dependent on the Big Phrases, here's a weird one: keyword phrases that are joined together into one word.
Take a popular phrase then turn it into one word. Blue Widgets becomes bluewidgets. Believe it or not, people are typing bluewidgets into the search engines.
Believe it or not, people are typing bluewidgets into the search engines.
I believe it. I've been seeing that sort of search term in my stats for years now.
An easier way to figure out if you're too reliant is just look at the total referers in the site stats and anything greater than 50% coming from Google, Yahoo or MSN is a red warning light IMO.
An easier way to figure out if you're too reliant is just look at the total referers in the site stats and anything greater than 50% coming from Google, Yahoo or MSN is a red warning light IMO.
I'd be worried if more than 50% of my page referrals were from search engines and other external sources.
