Nov 05

Google Analytics is scrubbing your data

Blogs, Doodling, Google, Heavy construction, Internet Marketing, Mind over Pixels, RGBhat, WebWriMo challenge 5 Comments »

I had a recent talk with an advertiser, who was interested in buying banner space on this blog. The discussion reminded me why I didn’t want to bother with individual advertisers in the first place. 

I was trying to tell him that my blog gets 400 uniques/day, and his argument against that was that my Alexa rank was not high enough, and that I was not getting that many visitors in reality. Here is my Alexa traffic rank, it is all public information anyway.

As I was saying, he claimed that a blog with a rank of 330.000 could not get so many visitors, and I was trying to explain to him, that despite the skewed Alexa data, the number you see there is the overall traffic rank. If you look at the bottom of the page, you will see a little graph that shows 65% of my visitors are from the US. And the chart right below that, shows my alexa traffic rank for the US, 73.000.

EDIT: Just to prove how irrelevant Alexa is, look at what Alexa shows as a relevant site to mine.

An analysis changes a lot depending on how you look at it. The pitfall here is when you look at the wrong kind of data. For example, people want to buy links from high PR posts omitting the new posts on authority blogs that are bound to get PR in a few weeks. Dumb, yes, but people do tend to think like that.

It is a good thing to check the numbers, but make sure you are looking at the right ones.

And now a fact I have wanted to point out for a long time: Google Analytics is scrubbing your data.

It really is. The analysis for glowleaf.net shows 1/10th of the visitors! That is a huge gap, it goes far from a simple discrepancy. Just to make sure, I used another 3rd party analytics script, and the data it showed were pretty much the same as my hosting tracking report. Losing 90% of my visitors is a crappy way to report my traffic…

Despite the crappy tracking, the Google Analytics tabs have a lot to tell me about my blog.

 

Traffic sources, all traffic sources:

 

Source/Medium Visits Pages/Visit Avg. Time on Site % New Visits Bounce Rate
1.
(direct) / (none)
 468 1.77 00:05:15 49.57% 71.37%
2.
wickedfire.com / referral
 91 1.32 00:01:15 52.75% 79.12%
3.
google / organic
 70 1.34 00:01:17 64.29% 75.71%
4.
blindapeseo.com / referral
 14 1.57 00:02:30 35.71% 57.14%
5.
forums.digitalpoint.com / referral
 14 1.07 00:00:12 85.71% 92.86%
6.
cull.gr / referral
 11 1.09 00:00:35 100.00% 90.91%
7.
bobit.gr / referral
 8 1.12 00:03:29 75.00% 87.50%
8.
bluehatseo.com / referral
 3 10.67 00:51:53 66.67% 33.33%
9.
yahoo / organic
 3 1.00 00:00:00 100.00% 100.00%
10.
adbrite.com / referral
 2 1.00 00:00:00 100.00% 100.00%
11.
contempt.me / referral
 2 2.50 00:02:44 50.00% 50.00%
12.
insomnia.gr / referral
 2 1.00 00:00:00 100.00% 100.00%

The avg time on site column is what I see as the most important, and it is worth noting that a simple comment on bluehatseo gave me a loyal reader. 5 minutes on average for the majority of my visitors are enough for me.

Visitors, depth of visit:

 

Depth of Visit Visits Percentage of all visitors
1 pages 520.00 73.97%
2 pages 94.00 13.37%
3 pages 33.00 4.69%
4 pages 22.00 3.13%
5 pages 13.00 1.85%
6 pages 2.00 0.28%
7 pages 9.00 1.28%
8 pages 2.00 0.28%

Here we see that my readers are shallow diggers, with hardly more than one page visited on average.

Visitors, Loyalty:

 

Number of Visits Visits Percentage of all visitors
1 times 384.00 54.62%
2 times 44.00 6.26%
3 times 20.00 2.84%
4 times 13.00 1.85%
5 times 12.00 1.71%
6 times 9.00 1.28%
7 times 8.00 1.14%
8 times 7.00 1.00%
9-14 times 39.00 5.55%
15-25 times 45.00 6.40%
26-50 times 40.00 5.69%
101-200 times 66.00 9.39%
201+ times 16.00 2.28%

This table on the contrary, shows us that half the visitors are repeat readers, with a nice percentage of ~20% coming back for more than 20 times. The information we derive from this table and the previous is that the shallow diggers we saw earlier are basically repeat visitors who stay updated on every post.

A thing we can do to fix that problem of shallow visitors, is to use plugins that show relevant posts etc.

Visitors, length of visit:

 

Length of Visit Visits Percentage of all visitors
0-10 seconds 522.00 74.25%
11-30 seconds 9.00 1.28%
31-60 seconds 10.00 1.42%
61-180 seconds 39.00 5.55%
181-600 seconds 49.00 6.97%
601-1,800 seconds 44.00 6.26%
1,801+ seconds 30.00 4.27%

This table shows us something interesting. While 75% of the visitors barely skim the site (maybe its the big fat smileys on top that deter them, but who cares…) we have a loyal percentage of ~20% who are definately reading the posts, and when we factor in the repeat visits from the other tables, we see that they do so often.

Traffic sources, keywords:

 

4.
wickedfire skittles
 5 2.20 00:08:19 40.00% 40.00%

I just had to write this, it is so funny that I cannot bear it! Average time on site 8+ minutes from the “wickedfire skittles” google search, making that keyword the biggest avg time on site on the keywords page…

All that data, gives me the information that 20% of my readers, that is 80 out of the 400 daily visitors, are my loyal readers, who keep coming back for more (no idea why…)

Sounds good enough for me. Now, where can I find an advertiser that will rent adspace based on 80 loyal uniques/day?

Screw that, I am just waiting for these guys to go live… (crappy anchor text courtesy of their non inclusion in their beta phase of my blog)

Related Blogs

Popularity: 20% [?]

written by Glowleaf \\ tags: , , ,

Sep 18

Statistical chaos

Mind over Pixels No Comments »

Web analytics are essential in the modern business. It’s real dumb not to use them, 10 years ago people sent out questionaries and gathered data and feedback, spending months analyzing them to find out what was happening in the real world that related to their subject.

Now, you can practically do the same with Google Analytics for example, getting more accurate data, without wasting your time or your client’s, and most importantly, you can do it for free.

Web analytics generate tons of data for your online business. But, there is a reason that there are courses in universities for Statistics. Statistical analysis can make you or brake you, it can steer you off course, or show you the hidden gem. And it all depends on who does the analysis.

A statistical analyst cannot be average. He either sucks, or he is brilliant. And Statistical Science by itself is a pile of crap. It’s the only science that can prove that if I have a chicken for lunch, and you are starving, then we both ate half a chicken, and everyone is happy.

But, if you pair the same science to someone with a 3 digit IQ, one who has experience on the subject, and takes a holistic approach to data analysis, then you are doing great.

You see, in computer science, we separate the term “data” from the term “information“.

  • Data, is the raw data, the numbers, the stats. By itself its meaningless.

“324 visitors, 1240 pageviews, 728 incoming links”. It doesn’t really mean anything.

  • Information, on the other hand, is what we extrapolate from the data.

“Devoted audience, multiple article reads per visit, popularity in fora and other blogs”. Now that is information.

Information, shows you the lay of the land, where you are doing well, where you are not, and what you should do to improve rankings/conversions/traffic/loyalty depending on your goal. You need to extract information from data. And if you have a hosting account, you already have data. Gather enough of them so they are statistically important as a sample, and start seeing the patterns.

Chances are that you cannot hire an analyst for your online business, but I am going to give you an example analysis of this blog, so that you can go along the same lines.

Coming soon, at the next post. Hit my big bad RSS button.

Popularity: 10% [?]

written by Glowleaf \\ tags: , ,