search

matomo-org / matomo

Empowering People Ethically with the leading open source alternative to Google Analytics that gives you full control over your data. Matomo lets you easily collect data from websites & apps and visualise this data and extract insights. Privacy is built-in. Liberating Web Analytics. Star us on Github? +1. And we love Pull Requests!
https://matomo.org/
GNU General Public License v3.0
19.71k stars 2.62k forks source link

Multiple visits logged when conflicting campaign URL parameters are used #18511

Open bx80 opened 2 years ago

bx80 commented 2 years ago

If a URL containing campaign parameters with different values is tracked when the Referrers and MarketingCampaignsReporting plugins are both enabled then multiple visits will be logged for the same visitor.

For example:

https://exampledomain.org?utm_source=bing&utm_medium=sem&utm_campaign=&matchtype=b&utm_term=test1&utm_content=&utm_medium=ppc&utm_term=test2&utm_campaign=test3&utm_source=bing&urlref=https://se.search.yahoo.com/&_id=uuidabc123&_idn=1&_rcn=bing&_rck=test4&_refts=1638204380&_ref=https://se.search.yahoo.com/&send_image=%27

It looks like first plugins\MarketingCampaignsReporting\Tracker\RequestProcessor() chooses test3 as the campaign name and also sets it as the referrer name and the referrer type as a campaign.

Then plugins\Referrers\Columns\Campaign::shouldForceNewVisit() (via plugins\Referrers\Columns\Base::detectReferrerCampaignFromTrackerParams()) uses the _rcn URL parameter to detect bing as the referrer type. Since the values don’t match a new visit is forced.

This is more obvious when there are media elements on the page sending multiple media tracking requests, where 4+ new records are created in the log_visit table for the same idvisitor withing a few seconds.

See L3-187

Expected Behavior

Only one visit should be recorded.

Current Behavior

Multiple visits are recorded.

Possible Solution

MarketingCampaignsReporting and the Referrers plugin should prioritize the campaign / referrer URL parameters in the same way so that the referrer is seen as the same and no new visit is forced.

Steps to Reproduce (for Bugs)

  1. Enable the Referrers plugin
  2. Enable the MarketingCampaignsReporting plugin
  3. On a tracked website visit a page with these URL parameters: https://exampledomain.org?utm_source=bing&utm_medium=sem&utm_campaign=&matchtype=b&utm_term=test1&utm_content=&utm_medium=ppc&utm_term=test2&utm_campaign=test3&utm_source=bing&urlref=https://se.search.yahoo.com/&_id=uuidabc123&_idn=1&_rcn=bing&_rck=test4&_refts=1638204380&_ref=https://se.search.yahoo.com/&send_image=%27
  4. Check the log_visit table, there will be multiple visits for the same visitor id.

(optionally for more duplicate visits: Enable the MediaAnalytics plugin, add the MediaAnalytics tracking code, then visit a tracked page using the above URL parameters where the tracked page has a few autoplaying videos)

Context

Your Environment

sgiehl commented 2 years ago

@bx80 We can't really prioritize the parameters the same way in core and in the plugin as they are using different config flags and classes. A quick suggestions that comes to my mind when thinking about it would be to kind of move the CampaignDetection class from plugin to core and use it with DI in core only. That way we could replace the used class in the plugin to use other / additional detection.

samjf commented 10 months ago

Another instance of this occurred in L3-608.