can anyone explain why the URL’s of the search engines are displayed under the channel types in the websites?
And why is the number of visits there significantly higher than the visits under the channel type search engines?
Doesn’t Matomo recognize every reference of a search in a search engine?
when I look at the tracking data of the last years in Matomo, it always looks the same as above.
Although the individual search engines websites appear in SearchEngines.yml (e.g. www.bing.com, www.ecosia.org), they are listed as a website.
It is different with Google. Strangely enough I did not find the following Google URL’s in SearchEngines.yml: www.google.com www.google.de
Each configuration needs to contain one ore more urls. Please only define the hostname. You can use {} as a placeholder for country shortcodes in subdomains or tld.
{}.searchengine.com would also match de.searchengine.com or nl.searchengine.com
searchengine.{} would also match searchengine.de or searchengine.nl
Notes:
For TLDs only {} would also match combined TLDs like co.uk . (Full list com.*, org.*, net.*, co.*, it.*, edu.* )
The first URL will be used for the icon, so the most popular/representative URL should be placed there
Okay, I worded this a bit badly:
There is an entry here:
So Matomo should automatically group all referrers that look like google.something into the search engine category. If it doesn’t (as it looks like for you), then this is maybe a bug.
I recognized the same issue with the latest Matomo v3. www.startpage.com and eu.startpage.com are classified as websites and not as a search engine. This happens even General['enable_internet_features'] is not disabled here.
So I tried the console command mentioned in the Github issue:
The task is successfully executed but this does not solve the problem. As an alternative an SQL query was mentioned to delete the SearchEngineDefinitions field.
But I’d like to ask first if this could be really the solution. I don’t want to mess up the Matomo installation completely with executing raw queries on the database. It would feel more comfortable solving this with the console command so I wonder why this does not seem to have any effect?
Matomo uses this list for detecting search engines:
But as specified it only detects startpage if either the referrer contains a ?query=searchterm part of if the referrer is /do/asearch. As startpage does neither anymore, but uses a referrer policy to tell the browser to set the referrer to https://www.startpage.com/ Matomo doesn’t detect it.
I wrote down my thoughts about this in a new issue:
that’s an interesting issue, I just saw startpage.com was in the YML file so it seemed related to this thread but this is indeed a separate problem.
Thanks for analysing it and creating the Github issue. I also thing search engines should just be identified by the domain part in future. This would be also easier to maintain.