Data discrepancies across different analytics platforms

When migrating from one analytics platform to another it might be the case that metrics don't match because of different measurement approaches, instrumentation mistakes or other reasons.

Below you can find a list of different reasons why metrics might not match across different analytics platforms.

Concurrent users discrepancies

You might see a different number of concurrent users in Compass real time compared to other analytics systems. Marfeel SDK sends presence events every 5 seconds to provide high accuracy on the real time view. Other platforms only track one page view event when the user lands on the page and track the user in the realtime for 5 minutes.

By looking at the historical data, the number of page views and unique users should match while time on page might differ due to the same reason.

Unique users discrepancies

More users and page views tracked

There are situations where Marfeel reports more unique users than other analytics platforms.

One of the reasons might be related to privacy and GDPR compliance. Marfeel is committed to maintaining user privacy and is a certified IAB TCF 2.0 vendor. Marfeel integrates out-of-the-box with any TCF-compatible CMP.

Other analytics platforms rely on ad-hoc publisher’s implementation. It’s quite common to see that tracking pixels don’t wait for user consent or rejection to fully block the tracking, which normally means more or less users tracked; data inaccuracy in front of privacy.

Marfeel waits 1s for the CMP to initialize, if consent has been rejected or there’s no consent available Marfeel will track the session anonymously and GDPR compliant. Marfeel won’t be able to identify the same user in a future session but will track in-session events. If the user provides the consent after the first event has been triggered, Marfeel will reconcile it. Marfeel always gives publishers rightfully, ethically, and legally data for their business goals.

You can debug this situation by opening a Network developer tool panel in an incognito window. Don’t accept the consent. If you see a collect request going to a google analytics domain it might mean your implementation is not waiting for the consent.

This factor might have an impact on the measurement of the RFV, where more new users will be accounted and might differ from the new vs returning visitor reports on platforms like Google Analytics.

Consent issues should only affect unique users, page views must be accurate and similar.

Less users tracked due to AMP sessions recirculating to web

On sites with AMP traffic that recirculates to the Web version other analytics platforms might consider 2 different unique users and sessionss. Marfeel, if the user has 3rd party cookies enabled, automatically reconciles the two users into one.

Usage of utm_sources and utm_campaigns for internal traffic

On platforms like Google Analytics each session can be attributed to only one campaign, traffic source or traffic medium. Thus, if new campaign or traffic source values are sent to Analytics at collection time in the middle of an existing session, it causes the current session to end and a new session to start. In practice you get 2 sessions instead of one.

Page views discrepancies

There are several aspects that can contribute to page views mismatch.

CDN and different cache layers

When you add Marfeel SDK to your pages, depending on your infrastructure, it might take several days for the changes to propagate.

The Google AMP cache implements a stale-while-revalidate approach to pick up any change on amp pages. Unless you have a mechanism to evict all its cache it will take several days to see accurate data on AMP traffic.

To assess if the implementation is right we recommend comparing page views on a newly published article which is not cached by the AMP Cache

Triggering events at different loading times

Loading times have a big impact on bounce rates. As Google disclosed, 1 to 3s in loading times can increase bounce rate up to 32%.

Adding the instrumentation javascript in the same exact point as the other analytics platforms is important to prevent data discrepancies. The later an analytics pixel is added the less accurate it will be and the less page views and users it will track.

If you are using Google Tag Manager or any other tag manager make sure to add the instrumentation pixel in the same place. If not results wont’ be comparable.

Reading time discrepancies

Marfeel SDK sends presence events every 5 seconds to provide high accuracy on the real time view and reading time.

Google Analytics tracks time on page and time on site by measuring the difference between the timestamps of page views. If the visit only views one page, no time is recorded. Google Analytics counts visitors’ time on a page regardless of whether the browser window is in a hidden or visible tab.