Less transactions tracked than with GA

Hi folks,

I have a problem with piwik conversion tracking which has been bugging me for quite some time now and after countless research I am running out of ideas where the problem might lie.
The thing is: On our site we use Google Analytics and Piwik for conversion tracking. While both platforms succeed in tracking conversions we usually track 20% less conversions with piwik than with GA.

We are running a piwik conversion tracking script which has been constructed on the basis of the guidelines in the piwik docs. It is just a simple javascript, which first parses the conversion data from the page (for simplicities sake from the google-tag-manager data layer) and then sends the ecommerce tracking request.

First I thought the problem might be that the ecommerce-tracking requests sent to piwik are sometimes incomplete (i.e. lack data). However I can rule this out now as I have constructed a side script which logs all request data before it is sent to piwik.

I also do not necessarily think that it has anything to do with https as both our piwik installation and our shop have a SSL-certificate. However when we just started to experiment with piwik we had some problems there as we first did not have a SSL-certificate for piwik. This has been obviously changed now but it might be that we have still an old version of the tracking script hidden in the depths of our sever which still might try to send data to piwik (but to the old “http domain” and not the new “https domain”). Do you think that the problem might be there?

Otherwise I do not really have any other clue.

Therefore I please ask for your help. Do you have any idea what might cause the tracking difference of 20% between Google Analytics and Piwik?

I am grateful for any advice! Thanks!

Is your archiving process running during production hours or some other process that could be causing the Piwik requests to be canceled (E.g. the user navigates away from the page) or time out before a conversion could be recorded? I’m not positive this would be the issue, but I moved my archiving to only very off hours to prevent this from happening during peak hours. During archiving our server is very slow.

To rule out any javascript issues, I would set up a few parallel tests. E.g. create conversions that you can segment via some parameter. Then create the conversions by running through your user process. Segment the views in both GA and Piwik and see if the numbers add up. I would also open Developer tools in Chrome (while also running the GA debugging plugin) and watch the various items get tracked in real time…

Thank you very much for your response!

I will definitively look into it however I have one follow up question to your answer. Could you please elaborate on what you mean with “archiving”? Do you mean the server saving its current status, data etc. or something else? Because the piwik requests are tracked and saved on the server right away once the come in, aren’t they? I just want to make sure in case I miss anything.

Otherwise thanks again for your suggestions!