90% discrepancies with site analytics
Many marketers have noticed discrepancies between the clicks reported by their programmatic ad platforms and the clicks that arrived on their landing pages, as recorded by their site analytics. Sometimes these discrepancies are 90% or higher -- that is, the programmatic campaign reported 100 clicks, but Google Analytics only reported 10 clicks that arrived on the site. Google Analytics filters some bots but it does not tell you which bots they were and how much was filtered. So you are left with a discrepancy you cannot explain. The fact that Google Analytics filtered the clicks from bots meant the bots were so simple and obvious that even Google Analytics could filter them out.
The above shows the clicks from 15 different utm_source=programmatic campaigns. Red and orange means bots and dark blue means humans. About 1 - 8% of the clicks from programmatic campaigns are humans (dark blue). It doesn't matter if you got a lot of clicks, because bots don't buy. And there is no brand lift when bots view your ads. I know you knew that.
Clicks from humans come from sites that have human audiences. See the example above. You will recognize these sites. Humans don't visit sites they don't know about; or use mobile apps they've never heard of before.
Also, if a human does click on your ad, and arrive on your site, they typically do something, like look around on your site. So we usually see mouse movement, clicks, page scrolling, touch events, etc. See the data below from the Clicks tab in FouAnalytics - clicks, mousemove, touch events -- grouped by screen resolution. These are things that Google Analytics and Adobe Analytics do not show you.
What do bot clicks look like on-site (see below)? There's hardly any data to show, because bots generally don't need to work that hard (e.g. faking mouse movements, page scrolling, etc.) to get away with it.
Recommended by LinkedIn
Advertisers are using FouAnalytics to measure their ads (where the ad went and whether a bot or human caused it to load). And advertisers are using FouAnalytics on the landing pages to measure the quality of the clicks coming from all channels, paid or otherwise. This includes clicks from Google paid search ads, YouTube ads, LinkedIn, Twitter, Facebook, Instagram, TikTok, etc.
Not only can advertisers optimize away from bots and fraud (dark red), they can also optimize towards humans (dark blue). None of the legacy fraud verification vendors measure for humans. Time to upgrade from fraud verification to proper media analytics.
Freelancer
10moWOW, great article. Does an adblock app causes a big problem about discrepancies for site analytics app?
Data Modeling and Advanced Analytics
1yGreat to see someone talking about this. I've hypothesized this for some publisher traffic and I tend to use path analysis to figure out if they're bots. It's often higher with certain 3rd party data sources. Works for both Adobe and Google properties.
Senior Data Science-Marketing Professional
1yI would not expect GA to show this but Adobe could do more; hopefully their new CJA approach will make this easier to do... 😅
VP Marketing - Consumer, Brand and Digital Growth
1yDo you have any articles that delve into the source and motivation behind these bots?
Global Head of Media and Analytics at Fundamental Group
1yLots of pollution in attribution at present and not enough investigation from advertisers as to who and what is coming to site. Interesting times for sure and an insightful article.