Not every deviation is an error

von Katrin Nebermann

Not every deviation is an error

Valid data is the basis for meaningful analyses and well-founded decisions. If data is compared between different web analytics solutions, between web analytics and backend systems or between web analytics and ad server evaluations, the following should be noted:

1. ignore minimal deviations

Deviations of less than 5% should be considered normal. Even with identical processing operations and key figure calculations, the position of the tag in the HTML of the page or the distances to the different data centers alone can cause slight deviations.

2. understand minor deviations

Even deviations of 5-10% are not unusual and do not necessarily indicate faulty implementations or reporting problems. The following aspects can cause justified deviations:

• Advertising marketers usually measure clicks directly via the publisher page or the advertising medium, while web analytics is only used on the target page (except when using redirect links). If the visitor closes the browser window or tab very quickly, no recording takes place on the target page.
• Clicks, visits and visitors are different pairs of shoes in different solutions. While the definition of clicks should be the same, the calculation of visits and visitors is often based on different logics and techniques, e.g. end of visit only based on time-outs or visitor recognition using cookies, fingerprints, user IDs or session IDs.
• etracker Analytics observes the DoNotTrack signals of the browsers by default. Other solutions do not offer this option or require special settings.
• Data collection can also be prevented by “strict” data protection settings in browsers or “private” windows as well as additional advertising and tracking blockers.
• With etracker Analytics, traffic from known bots and spiders is automatically filtered out. With other solutions, the exclusion of bot traffic must be activated.
• Deviations due to payment service providers such as PayPal, Amazon Payments and credit cards can be minimized, but not completely avoided. The reason for the deviations is that customers can cancel the return to the store page after the external payment process despite the “Auto-Return” setting.
• With other solutions, deviations can occur due to sampling, especially with high traffic. In the case of tracking requiring consent, an extrapolation is sometimes made from the users who have consented to the total of all visits. Neither is the case with etracker Analytics.

3. investigate major deviations

If the deviations exceed the 10% threshold, possible sources of error beyond the aspects listed under point 2 should be investigated. In most cases, high deviations are due to incorrect implementations. The following questions need to be addressed:

• Is the execution of the etracker code incorrectly blocked by Consent Management?
  
  The type of interaction with consent management (i.e. pre- or post-consent tracking) has a huge impact on data collection. The etracker code contains integrated cookie blocking as standard and should therefore be excluded from blocking by the CMP. See also our instructions for the correct integration of etracker Analytics in CMPs.

• Is the etracker code implemented correctly on all pages and are all pages clearly recorded?
  
  If the page titles are not unique or pages are only differentiated by URL parameters, the default settings must be adjusted or code parameters must be used. The special requirements for AMP pages, single page applications and the like must also be taken into account.

• Does the CSP header of the website possibly not allow parameterization?
  
  The Content Security Policy (CSP) can erroneously block the execution of the etracker code or cause it not to function correctly.

• Is campaign and especially SEA tracking implemented correctly?
  
  SEA traffic is automatically assigned to the SEO medium if the SEA campaigns were not provided with UTM or etracker parameters in the Suffix field of the final URL.

When testing, please make sure that you are not excluded from the measurement by IP blocking or an opt-out cookie!

If you have checked all the aspects mentioned, you can count on your etracker analysis data.

Our consulting experts will also be happy to help you in the form of an individual and thorough implementation check-up. It is best to repeat this regularly – as usual for motor vehicles – every 2-3 years in order to benefit fully from all etracker innovations and possibilities.