Reporting Discrepancies

Overview

🚧

When using any ad serving service, it is common to see around a 10% discrepancy between with impression reports from other platform (such as Quantcast, Google Analytics, or another ad server).

Below are the most common reasons for reporting differences between Kevel's system and other platforms you may work with:

  • Missing Ad Calls: Troubleshoot whether the ad code or server-side requests are breaking to either Kevel or the 3rd-party

  • 3rd-Party Ad Latency: When a visitor views a page, Kevel, being server-to-server and faster, may record an impression before the 3rd-party ad actually loads. If a user abandons the page during this time, Kevel would record an impression, but the 3rd-party would not.

  • Multiple Ad Requests: If your site/app has multiple placements on a given page, Kevel's impression numbers will not match up. For instance, if your homepage has 100K daily impressions, Kevel may report 200K total impressions if you have two ads on the homepage (100K for each).

  • Ad Blocking Programs: This is an increasingly common reason for discrepancies. When website visitors use a service like AdBlock, this prevents ad requests from being fulfilled. Learn more about Ad Blocking here.

  • Reporting Integrity Filters: Kevel tracks only impressions it knows are real viewers. It does not track search engine robots, spiders, crawlers, link checkers, validators of HTML/CSS, RSS readers, scrapers, and more. While other services do the same, often the lists are different and can lead to discrepancies.

  • Time Zone Differences: Kevel may use a different time zone than another reporting tool. Due to this, impressions from one day may count toward another day depending on the zone differences. Kevel uses UTC, so keep that in mind when comparing the two.

  • Blank Ads: A third-party service (or advertiser server) may serve a blank ad, which is then counted as a valid ad request on the publisher ad server. This can take place whether Kevel acts as the advertiser or publisher server.