To start with, a topic that many online marketers have to deal with on a regular basis – deviations in the counting of clicks or impressions. This can have numerous causes, a few of which should be highlighted here.
What it´s all about
Anyone who wants to get visitors to their site or increase the awareness of their brand uses online advertising – whether in search engines, in display networks, on social media or in form of native ads. Mostly you will be charged based on clicks or ad impressions. In order to be able to better understand whether the purchased advertising service really works, you often rely on your own tools for visitor counting – be it your own ad server, Google Analytics or another web analysis tool.
Most of the time, the provider also offers statistics on the clicks and impressions delivered – if these now deviate from the self-measured values, the question arises as to what causes this difference. The current jurisprudence on the subject of cookie consent has brought a new level of complexity to the subject.
Loss on the click chain
Many different providers are often involved in the delivery of advertising materials – starting with the publisher’s ad server and various supply side platforms (SSP) to the provider’s demand side platforms (DSP) and ad servers. In case you use your own ad server or click tracking, there can be several redirects via different systems from the first click on the banner to the successful visit to your website. If one of those systems goes down the clicks will be tracked untill they hit this system – all other systems down the chain will not count the click. Another issue might be the users adblocker, which only blocks a selection of the involved redirect urls. Or maybe the redirections take to long and the user gives up and closes his browser window before he reaches your website.
These are rather the normal scatter losses that can be tolerated in the range between 5-10%. If it goes beyond that, one should investigate the cause.
Javascript switched off or adblocking active
Many analysis tools rely on Javascript – users with deactivated Javascript will not be counted properly. Common analysis tools are now also listed in many ad blockers, so that visits are not counted when those tools are integrated on your website.
These losses also fall within the tolerable range of a maximum of 5-10%.
Insufficient filtering of bots
Adbots are necessary in times of real-time bidding in order to check the advertising material for valid URLs, malware and unauthorized content. To do this, the bots simulate a browser, load the advertising material and execute a click. If a tracking system does not recognize this adbot correctly, it incorrectly counts a click or an impression – other systems ignore the adbot based on their filter lists. This means that there are significantly more clicks in one system than in the other.
Search engine web crawlers should normally not trigger this effect, but here too, incorrect configuration can lead to deviations.
Only the use of current bot lists in all systems and appropriate filtering can prevent this.
Missing Consent
A factor that should not be underestimated since the introduction of the GDPR – analysis tools may only be loaded if the user gives his explicit consent. Many websites use so-called consent layers for this purpose, which are of course optimized so that the user accepts all cookies as possible. However, if the user comes to a website via an advertising campaign and leaves it again without consenting to the analysis, his visit is not counted. This is likely to be responsible for a not inconsiderable part of the count discrepancies between advertising networks and on-site analytics.
The only remedy here is a consent tool optimized for consent or an upstream ad server for counting – but be careful, consent is required here too, the industry relies on the IAB’s TCF framework for this – but this method should generally provide more reliable figures as web analysis tools on the site.
Adfraud
A factor that should not be underestimated when it comes to counting deviations is extensive botnets that generate clicks and impressions in order to earn advertising income. A profitable business for the operators – in the first quarter of 2018 more than 750 million dollars should have been generated. The mechanisms to tackle such networks are of course getting better every day – but the game of cat and mouse continues and impressions and clicks are still being sold without any real person ever seeing them.
Of course, these adbots do not interact with the website or the consent layer and are therefore not recorded in the on-site statistics.
Numerous DSPs and ad service providers offer solutions to avoid buying clicks and impressions from botnets in the first place – the only thing that helps here is comparing several providers and comparing your own figures with those of the advertising network.
Conclusion
Counting discrepancies will continue to accompany us, but there are ways to reduce them.
Of course, this post doesn’t go into all of the possibilities – if you have problems with counting discrepancies, feel free to contact me for assistance.
