I spent some time last night collecting stats and putting them together. The result is a huge table with a lot of numbers which by itself is pretty useless unless you want to use it as a reference.
My goal however was not to impress you with the amount of information but to find if Alexa and Technorati can be valuable tools in estimating a blog (or any website for that matter) popularity.
My assumption was that if blogs belong to the same category and the visitors demographics is similar then Alexa and Technorati ranks can be very precise when compared in a group relative to each other.
How the data was analyzed
The sample set included 62 Personal Finance blogs and you can browse the complete table with raw stats here. I fed this data into a statistical analysis tool and ran several correlation tests between the data sets.
Correlation in statistics is a number ranging from -1.0 to +1.0. The closer it is to +1 or -1, the more closely the two variables are related. A value close to 0 means little relationship. If you want to find more about correlation, here is a good article with some real life examples.
In addition I threw in another statistical coefficient, significance level. This is a number from 0 to 1 which shows how likely a result is due to a pure chance. For better readability, I will show (1 - significance level) so values close to 1 will mean the correlation is more significant (is likely to be true).
Sample size shows how many blogs are included in a particular calculation. Since some blogs lack certain stats, the calculations will exclude these blogs from the original set. Bigger sample set means better calculation precision.
Results: high correlation
Here are the most interesting findings. The correlation of 0.9 or more indicates that the values are highly related to each other. Please note however that correlation highlights overall statistical trend and doesn’t exclude exceptions. A good example is Not Made of Money which lags by both Alexa and Technorati but still scores impressive daily visitors counts.
Correlation | Sample size | Significance | |
---|---|---|---|
Technorati to Visitors | 0.95 | 29 | ~1 |
Alexa to Visitors | 0.86 | 33 | ~1 |
Technorati to RSS | 0.78 | 33 | ~1 |
Del.icio.us to Visitors | 0.77 | 33 | ~1 |
Del.icio.us to RSS | 0.72 | 43 | ~1 |
Below is a scatter plot matrix which shows how the dots distribute around X/Y coordinates where Y is 1/Technorati and X is daily visitors.
Results: some correlation
Here are some results that show less correlation but are still very interesting to see. P/V is the number of pageviews divided by number of visits. It indicates how long the same visitor stays at the website before they leave or how often they return back during the day.
Correlation | Sample size | Significance | |
---|---|---|---|
Alexa to RSS | 0.46 | 43 | 0.99 |
RSS to Visitors | 0.45 | 24 | 0.95 |
RSS to P/V | 0.38 | 24 | 0.95 |
Google Rank to Del.icio.us | 0.31 | 62 | 0.98 |
Google Rank to P/V | 0.25 | 33 | 0.83 |
Google Rank to Visitors | 0.21 | 33 | 0.75 |
P/V to Visitors | 0.20 | 33 | 0.73 |
Blogs with most devoted readers
Here is another interesting list. This is a list of blogs sorted by the pageviews to visitors ratio. The high P/V can indicate two things:
- The blog has good design and convenient navigation which encourages new visitors to stay longer and browse around;
- The blog has devoted readers, the kind who come back several times a day to comment or check on updates.
In either case I encourage you to check out these blogs. Whatever the reason for high P/V may be, they must be doing something right.
Note, not all blogs from the original list are included since some bloggers do not publish their SiteMeter stats.
# | Blog | P/V |
---|---|---|
1 | Blogging Away Debt | 2.80 |
2 | Mighty Bargain Hunter | 2.59 |
3 | Personal Finance Journey | 2.52 |
4 | Free Money Finance | 2.01 |
5 | The Simple Dollar | 2.00 |
6 | Experiments in Finance | 1.89 |
7 | My Two Dollars | 1.85 |
8 | Not Made of Money | 1.84 |
9 | Get Rich Slowly | 1.81 |
10 | Consumerism Commentary | 1.78 |
Conclusion
The correlation numbers above have exceeded my expectations. Alexa and Technorati turn out to be very good indicators of a blog/website popularity if used against a group of blogs/websites from the same category. Technorati seems to be better than Alexa which goes along well with this other study done at SEOmoz.
The last list can also be of some interest, mostly for bloggers. It looks like these blogs are doing a great job with their content and/or design and encourage readers to stick around for longer. Check them out and let me know in the comments what you think about it.
Nice job Yan. You obviously didn’t fall asleep in statistics like I did.
So it looks like Alexa and Technorati do a fairly decent job.
What do you think the major factors are that cause Alexa to SKU?
One example I can think of is, if I put my site up for sale on a forum, and webmaster went in to check alexa rankings…
Yan,
I will make my sitemeter public for this sake. :) Can I get my P/V ratio? :)
Oh yeah..I think I confused it with the earlier correlation results.
On this, I have a few thoughts. A simple P/V is a cursory pointer towards returning visitors. Earlier, my my Google search traffic was dismal and as a result, the percentage of returning visitors was very high…that resulted in a P/V of more than 2.5 on certain weeks. However, with the gradual increase in search engine traffic, the relative percentage of returning visitors has gradually dropped and that results in a lower P/V ratio.
I wish there was some way to track P/V history over time.
You can do it using Google Analytics, a free tool from Google. I agree that search traffic may lower your P/V to some extend. The ads you put around your pages do the job as well since people click on them and navigate out of your blog.
A spike of popularity related to getting to Digg front page could skew Alexa for quite a long time. Most Digg users are technically advanced and a high percentage uses Alexa toolbar. Once on the Digg front page, the Alexa rank will be elevated for the next 3 months since the value is a 3 months average.
P/V is just your number of page views divided by your unique visits, and you can calculate it easily yourself. Looking at your stats, the current value seems to be 1.63, which is 1559 divided by 959.
I guess you must have published this post while I was on my blog hiatus, but thanks for putting it together.
My blog has a lot of posts related to how-to-use/do-something-in-Excel/finance, which might explain my P/V ratio. Most visitors are new (87% during the last month according to Analytics), but they tend to be people looking for guides on how to do things, so that combined with my multi-page tutorials are probably what drive my ratio. Again, thanks for your analysis :)
Your web page does not correctly work in safari browser