So, Just how Accurate are Your Stats?
Yesterday was one of the better days I’ve had on this blog as far as traffic. The inclusion of my latest theme, Darkwater, on sites such as Weblog Tools Collection, Planet WordPress, and a few others, really drove more than my normal visitors to the site. The spike in traffic was evident no matter which statistics you look at. The experience has left me wondering how accurate these statistics are however. Basically, out of four or five different statistics providers, none of them matched.
Pretty soon after I set this site up in February, I uploaded the WP Shortstat Plug-in, which makes it real easy to glance at your stats whenever your in the WordPress Dashboard. I really like it, and that is what I usually check. Last night it was reporting to me that my site had taken 2,398 hits with 1,202 unique visitors. At the same time, I logged into MyBlogLog, which showed 1,153 page views with 801 unique visitors. hmm. I know I’m not the first person to ever raise the question, but to be off by that much had me scratching my head.
Several other places I decided to look at my stats after realizing this, was Statcounter.com and the statistics provided by my web host, 1 and 1. This is where it gets really interesting. 1 and 1 Hosting reports that I had 3,434 page views with 1,695 unique visitors. That sounds like a lot to me. Well when I checked StatCounter.com, they report that I only had 89 pageviews with 44 unique visitors.
Huh???
Now that’s a big difference! So, I’m wondering, where would I get the most accurate statistics? Regardless of which is right, I at least know that I had a lot more visitors than normal.
Here is my opinion about what is going wacko:
1. 1&1 stats is counting a hit for different images and such and not just loading of pages. They always seem higher than the rest for me too.
2. Statcounter is messed up and only counting your homepage views instead of each and every page. This happened to me for a while when I still used it.
I really have no clue about My Blog Log as I have never used it. However, I think you can trust ShortStat. I have been using the SlimStat plugin (similar.. shares code I believe) for a few days now as opposed to 1&1 stats. It seems to be most accurate to me. Maybe I’m wrong though. According to it I’ve gotten 989 hits since I installed it earlier this week.
That all makes since. I never thought that 1and1 may actually count it as a hit every time an image loads…….interesting.
Yeah unfortunately there is a misconception of what a “Hit” actually represents. Jake is right, if a page has 16 images on it then each time it loads that will be counted as 16 hits on a site. I’ve had to give this same explanation to several people that strictly want to measure a websites “Hits” on whether the site is a success or not.
I don’t even pay attention to anything that refers to hits anymore. Mainly I just pay attention to page views , visits, and unique visitors. I’ve been using Google Analytics for stats, but I recently tried a new analytics option called Reinvigorate and really like it quite a bit.
I just assumed a hit and a page view were the same……..apparently I was wrong.
It could also be that some third party stats scripts are blocked by anti-ad programs.
For me, the only stats I’m confident in would be those from my server log.
Yes, Statcounter is very common to block nowadays. That is probably why it is lower… I didn’t even think of that. Also, visitors without Javascript can’t be counted by a lot of sites.
Well, at least I know not to use Statcounter. I’m going to keep using WP-Shortstat, and assume it’s correct…….at least for now.
What was finally suggested to me was just to track one, and watch its traffic. While it might not be “the truth,” you can at least compare increases, decreases, etc.
HI Anthony,
Sorry to hear about the trouble you have been having with StatCounter. We work really hard to make StatCounter into the best service possible.
I can assure you StatCounter is an extremely reliable way to track the real number of visitors you are actually getting to your website.
I think I may be able to help solve some of the problems you have been having.
a) StatCounter only tracks the actual visitors to your website, whereas ShortStat will also track the search engine robots to your website. This will inflate the ShortStat count.
b) StatCounter will only track the visitors to the pages you install the code on. Whereas ShortStat will work from the server logs so will record all server activity. Currently you only have the StatCounter code installed on your homepage. This will have severely decreased the StatCounter count i.e. only visits to your homepage are counted and not visits to your entire blog. We highly recommend to our users to install the StatCounter code on all pages of their blogs.
To do this you can use this plugin.
http://blog.lucanos.com/2006/04/17/wordpress-statcounter-plugin/
Or alternatively in your wordpress directory you need to open the file wp-content/themes/default/footer.php (or whatever theme you are using) and add the StatCounter code there.
Then don’t forget to remove your old StatCounter code from your homepage or you’ll have it installed twice there!
c) StatCounter will track page reloads whereas a service like Shortstat which works from server logs won’t record this activity. This is because the page will be cached in the visitor’s browser.
I hope this helps clear up the confusion.
All the best!
Aodhan
StatCounter Team
Aodhan,
I appreciate that explanation. I’ve downloaded the plug-in, and will try it out.
@Aodan: Unfortunately, Statcounter code is still blocked by a lot of ad blocking scripts and plug-ins rendering the stats inaccurate for people writing blogs that appeal to advanced computer users, which I would guess is higher on this blog than a lot of sites. I believe this is part of it.
[...] wrote a really interesting post detailing his experiences with different stats services. I can understand why they all give different numbers, but when your traffic goes up and Alexa goes [...]
I’ve been playing around with Google Analytics, Feedburner stats, and Performancing’s pMetrics, and yes–I get completely different results with all of them as well. Does anyone know how much the different stats packages rely on cookies to do their magic?
Congratulations to Aodhan, who just one BusinessWeeks Young Entrepreneur of the Year!.
STAT COUNTER DOESN’T COUNT MANY REAL HITS
I’ve used Stat Counter on about 40 sites for years now, a very long time. I’m gearing up to sell a few of them and I find that relying on Stat Counter for an accurate estimation of your stats is a bad idea. I was certain that Stat Counter wasn’t counting all my hits so I googled that exactly and found your site.
The representative from StatCounter commented that “StatCounter tracks real visits… not robots etc. which results in higher results for Awstats” If you are looking at Awstats, on a site like mine, I get thousands of visits from Google Robots every other day! But in Unique Hits it will only be counted once this entire month, so how is that not accurate more accurate that Stat Counter that allows a user to count the same IP as soon as an hour after the last visit? If that wasn’t enough, in AWSTATS, the pageviews are shown next to Unique and under that is shown something called NOT VIEWED TRAFFIC, which is totally separate from the total of Real views made by real people. Right from the smart people at AWSTATS, * Not viewed traffic includes traffic generated by robots, worms, or replies with special HTTP status codes. what other type of fake traffic is there? Proxies? I know for a fact StatCounter doesn’t filter them out. So where’s the problem and where’s the benefit of adding extra JS that slows your page down over a stat program that is embedded in your system?
After years of using Stat Counter, the answer is NONE.
What we have here is StatCounter protecting their product (yes is a product)by downplaying the importance of Awstats and Raw log files which to the untrained eye seem “dinky” and cheap… but why else would it be there consuming tons of resources if it didn’t have some benefit? Other than Godaddy, What host would provide you with garbage to use that ate up tons of memory? Very few.
Inevitably, it’s cool to see where exactly the latest visitors came from and the latest keywords and paths…. but to be honest, stat counter is so overused they’ve become the slowest site I visit. I can’t remember a time when I wasn’t logged out after a minute or two of use. And to say that their stat counting isn’t affected by millions of people going to STATCOUNTER.com every minute is to say that you can drive on sand without slowing down; there’s going to be some adverse effects.
Once upon a time, StatCounter tracked incoming Google Adwords clicks, but even after setting up a campaign with thousands of incoming clicks, Stat Counter ceased to count them. Which is sad because they were the only thing I had to try and keep Google honest, now I’m in the dark.
This month just started so its easier to run a comparison on incoming SE traffic (if i started at the end of the month the data from the beginning would be lost on StatCounter, my log file is only 5000). They say I’ve received 991 Different Search Terms while in my AWSTATs the number is 2200. (this is not very scientific, but the discrepancy shouldn’t be that high almost two thirds uncounted).
The final prognosis is reached with help of comparing total hits from Awstats, Analog Stats, and Stat Counter. On statcounter I received 113000 uniques and in Awstats 195000; the same discrepancies are repeated on a monthly basis. So what’s going on? Did I receive 80,000 visits from robots all with different IP’s? Of course not, like I suspected, Stat Counter counts your stats but with a plus minus 20-30 percent (always minus), so its not accurate at all.
Keep in mind, stat counter is embedded by default on every single one of my pages, so there’s no excuse really, especially with regard to the statement presented by Aodhan before. I’m sorry I know you’re probably a nice guy, but Stat Counter is just totally overused and it shows dude.
I think the problem is people are so inclined to distrust Awstats when they don’t see the traffic results they hoped for. Which is ludicrous in the sense that by default Awstats is in a better position to track more accurately than a (slow as hell) third party website that has a snippet of code on some of your pages. And sure enough Adohan’s response to your blog post was an attempt to do just that, make you trust in his product by reaffirming your worst fears in your server’s stats. I can see how he became the young entrepreneur of the year, its easy to sell stuff based on misunderstanding of your competitor, I guess BusinessWeek doesn’t take into account his inability to buy enough servers to make StatCounter.com run smoothly when they made their pick… I’m serious.
What alternatives are there? I’ve been looking around for a solution that can be installed via SSH and be part of the back end without having to install any extra JS or other type of includes. So far, they all are just glorified Awstats’, nothing too special that warrants hundreds being spent.
With respect to prettiness of stats I found the best site. I’m trying out VisiStat which is actually less money than Stat Counter (if you pay for it like I do), but I get way more than 25,000 impressions a month and would be forced to pay 50 cents per thousand I go over, which would cost me $300-500 monthly. The one cool thing about them is the live page view feature where you see users come to your site and they pop up on the screen in TRUE real time (Ajax rules). But this isn’t an endorsement, I won’t be buying their service after 7 days free.
Hope this helped, don’t feel bad there are more of us out there like you!