Data discrepancies
If you are seeing data discrepancies, compare the data at pageview-level, not at the session-level. Why? The definition of 'session' varies significantly between Google Analytics and Contentsquare whereas the 'pageview' metric definition is closer across analytics tools (the main differences will come from artificial page views that may be tracked on one side and not on the other).
If I am seeing data discrepancies when should I file a ticket?
You can open a Support ticket to have the discrepancies investigated if:
-
The data gap observed at the pageview-level is > 10% and/or
-
The data gap observed at the session-level is > 20%
What is the difference between Contentsquare 'sessions' and Google Analytics 'sessions'?
-
CS considers a session to have ended after 30 minutes of inactivity, after the last event/user action. GA allows users to change this parameter (from a few seconds to several hours).
-
GA starts a new session at midnight (in the Google Property timezone), irrespective of the 30-minute rule mentioned just above. So let’s say a real session starts at 11:00 PM and ends at 01:00 AM the next day: it will be split into 2 sessions in GA, while it’s not the case in CS.
-
GA starts a new session when the campaign source changes: if a user lands on a page via a first campaign, leaves the page, and comes back to it via another campaign, then the visit will be split into 2 sessions in GA, while it’s not the case in CS.
Why do data discrepancies happen in Contentsquare?
Discrepancies can happen for a number of reasons:
- CS tag issues in data collection
- Ex: Adblocker extensions, network issues, the tag not being applied to subdomains or 404 pages while GA is applied to the entire site, or a cookie policy that relies on the site's end-user to accept.
- Google Analytics sampling
- If the traffic of your target GA view is > 1 million sessions a day, then Google samples your traffic before running the segmentation query. Sampling will prevent you from getting precise data for your GA segments in Contentsquare, but even with sampling, the integration can be very useful for understanding trends.
- You can reduce the traffic associated with the GA view associated with your integration to bypass this issue.
- Differences in tag implementation
- The Contentsquare tag is called after or before the GA tag (each second can add a 3 point gap).
- The nature of the tag can be different (synchronous/asynchronous).
- One tag can fire before the cookie acceptance while the other fires after (ideally they should both fire once the cookie is accepted).
- Differences in perimeter
- The compatible browsers are different. Ex: CS is only guaranteeing tag execution on Internet Explorer 11+
- The property you are using in GA is not on the same perimeter as CS (different domains, subdomains, inclusion/exclusion of staging environments...)
- filters applied in both tools could be different (especially on the IP addresses).
- The bot exclusion rules could be different and not shared.
- The sampling methods differ
- The timezone could be different and impact daily data, especially on transactions. CS data is saved on UTC time.
- Both tools use the browser’s user agent for device definition (desktop/tablet/mobile) but they are not using the same library, which can lead to gaps only on specific devices (such as mobile).
- Differences in metric calculation
- Session definition: CS considers a session ends after 30 minutes of inactivity after the last event / user action. GA allows its customers to change this parameter. The sampling method and IP exclusion could also impact the number of sessions measured on a given period.
- Session duration and time spent on exit page: CS collects all the interactions on a page and the end of the session corresponds to the timestamp of the last packet of events sent. It is not the case in GA and time spent depends on the custom events set up on the exit page.
- Bounce rate definition: CS defines the bounce rate as the percentage of users with only one pageview. In GA, only visitors who do not trigger a tracked event on the landing page are considered as bounces.
- Load time: CS calculates load time from when the DOM first becomes interactive. The GA definition of page load time takes into account the entire timeline for page load, which includes a couple of steps beyond DOM ready. This means our load time would be less than that of GA.
- Transactions: CS counts the number of visits with a transaction whereas GA counts the number of transactions.
- Revenue: A transaction made in a different currency than the one set up in CS (€) is not converted. Revenue can therefore be higher or lower in GA.