Differences in numbers: piwik vs. awstats vs. google

Hi,

I do not know which counter to trust. I am playing around with variuos webstat tools at the moment (in a real world lab). And the resulsts differ - extremly. I did not expect this …

Scenario:

  • Website written in simple PHP, output is plain HTML (no Java, Flash, Javascript, whatever but some pdf files)
  • Piwik in each footer (inserted via CMS)
  • google urchin after piwik code (inserted via CMS)
  • awstats offline (nightly log analysis)

I just have a look at VISITORS and EXPLICIT VISITORS for november (rounded):

  • piwik: 1000 visits, 800 users (80 via adwords)
  • awstats: 8900 visits, 2100 users (110 via adwords)
  • google analytics: 1400 visits, 1000 users (130 via adwords)

how does this happen? which one is right? style_emoticons/<#EMO_DIR#>/smile.gif

I just searched the forum for awstats and found this topic - maybe someone has an answer to this now?
We’re facing the same problem: way less visitors in piwik than awstats shows. Example for a day in June:

Piwik (0.63): 52 visitors
Awstats: 302 visitors

I read the FAQ but even thinking of all the points mentioned in FAQ#50 I think the difference is too high. Robots etc. are exluded in the Awstats numbers.

Also we’re running some ads with adwords and the keywords of some clicks that were generated are not shown in piwik (when looking at the keywords in piwik), but here piwik seems to drop way less visitors (coming from google) than in general.

Any ideas what we could do or what the problem could be here? Or are the 302 visitors of that day just completely wrong and over years we thought we had a lot more visitors than we actually do?

Is that number from AWStats the unique visitors? Or raw visitors?

Oh, good point, I now see I mixed that up. It is the number of visits of that day which is of course higher than the amount of unique visitors. Unique visitors were still more than 200 though. Piwik counted 52 unique visitors and 61 visitors that day. 204 compared to 52 still is a huge difference though… hm.

Have the same with stat numbers, awstats says 6600 visits and piwik says 449 visits isn’t that a very huge difference in numbers.

Hi there will always be a discrepancy. What i recommend you do is this.

Pick a complete day from the recent past, do a complete pages report for that day in Piwik, awstats or GA

Look at the lists and compare, you will find missing pages from one or another or find odd pages that dont seem to be expected. Awstats is very good at finding internal systems pages that you haven’t expectedly built yet contribute to stats.

Goof luck

Thanks for replying.

The stats of awstats 6600 and piwik was for Sept 26th which was a full day ago. In fact awstats are consistant between 5000 - 6000 per day but piwik is consistant with 300 - 400 per day.

So you are saying that awstats is not the true reflection of your visits, how about this post: Google Analytics Accuracy » The SEO Guy

Thanks again.

did you pull a list of all pages visited in each of the 2 analytics programs? Do it and compare the 2 lists beside each other in excel for example. Once you have done that I would recommend you look at the discrepancies of actual pages contributing to stats you will be surprised!

Awstats was great when there was no great 3rd party analytics programs but in the industry today there are 2 main analytics tools. GA and Omniture. Piwik is in my mind a great up and coming contender as it has many advantages and each version just gets better and better. Plus its free! and allows you to keep your own data!

regards

Ok will do a comparison and see what I get.

Thanks and appreciate your help.

Hi, we are also experiencing a big difference in numbers between Piwik and another log analyser, Webabacus.
We want to adopt a new web stats tool and have chosen Piwik so we’ve been running it on two of our high volume sites for almost a year now and are really impressed with it. However the figures generated by Piwik differ considerably to Webabacus. Piwik consistently tracks a third less unique visitors than Webabacus. We’ve identified certain factors which would account for some of the differences e.g 301 redirects, RSS feeds, direct hits to PDFs etc, Javascript disabled, pages not fully loading, time zone differences etc. However on every page, Piwik reports at least a third less than Webabacus. Webabacus uses the Spider and Robot Exclusion list from ABC to filter out the main bots so this difference should be minimal. I’ve done a line by line comparison, comparing IP addressees tracked in Piwik and logged by Abacus for individual pages and there is nothing obvious that points to why Piwik hasn’t tracked a third of the IPs reported by Webabacus. I know there will be inevitably be a 5-10% difference between a page tagging solution and a log analyser but I wasn’t expecting a consistent 30% difference. I’ve looked at previous posts on the forum to help but still haven’t identified the issue. I would be very grateful if you could offer any further suggestions?
Thanks!

not having seen a data pull from each I can only speculate…
Where did you have the code for Piwik VS Webacus?

any way you can PM me a list of IPs tracked by each and a list of actual pages contributing to PVs for a 24hr period? I am curious to see what is making such a disparity.

When it comes to ips would be interesting to see the referres for the IPs as we could see what exactly is driving traffic from the ips to contribute.

regards