Banner Advertising Woes

Does a banner ad click equal a website visit?

To round out our brand awareness campaign for one of our food clients we launched a summer banner advertising campaign in May of this year. We created a few simple standard size ads and had them placed on some very prominent food related sites.

May come and went and we received our first advertising report from our Media Buyer who helped us place our ads. To our surprise we had some staggering statistics for our first month of the campaign. 4.5 million impressions, 19,000 clicks giving our simple banner ads a click through rate of .43%!

Wow! This is fantastic as the team didn’t expect that type of impact or metrics from our banners ads. However, before we popped the corks on the Champagne we wanted to check one more thing. We were curious to see how all of those banner ad clicks corresponded to visits on the website that the banner ads targeted. Beyond verifying the ad clicks we were interested in seeing what people were doing on our site once they clicked the banner ads.

And now for disappointment. As we dove into the analytics of the site we quickly noticed a pretty major problem with what was reported. We didn’t see a major influx of traffic from our banner ads. Out of the 19,000 reported clicks we only saw a meager 3,000 visits to our site. That meant that more than 80% of our banner ad clicks could not be counted as a visit to our site via Google Analytics.

Where did 16,000 clicks go?

Before we went back to our Media Buyer with questions we wanted to do some detective work on our own to make sure we had all of our bases covered. We began to think through possible scenarios and reasons behind the discrepancy in numbers. These are some of the questions and scenarios we thought through:

• Are people leaving the site before it loads?
  The mindset here was that once the ad is clicked a user might leave before the site is fully loaded and Google has a chance to track a visit. Google can track a 0 second visit once the site is fully loaded but if someone leaves as the site is loading that visit would not be tracked. This would be very apparent in slow loading sites but our targeted site’s load time was average. I’m sure we lost some visits along the way but we didn’t feel that the site’s load time would influence 80% of our potential visitors. Plus, when you consider user intent, a user actively clicking the ad to learn more they would be inclined to let the page load to view the content. Granted, no one wants to wait around for a slow site but this is desired content.
• Is Google not tracking the visits?
  Perhaps. Google does some odd stuff but nothing seems clear on why they wouldn’t track visits from our ads. We used custom tracking URLs and we haven’t really had any problems with GA in the past. No real solution behind this one but it’d be too easy to blame Google.
• Bad tracking urls?
  Our ads utilized tagged urls so that we can easily segment traffic in GA. Perhaps we messed these up or they weren’t placed properly. We double checked everything and these were not the culprit.
• JavaScript disabled?
  For Google to track a visit these features need to be enabled in a user’s web browser. If not, they could visit our site undetected. Based on our research we found that only 2% of users have these features disabled. I’m sure we might be missing some traffic from undetected visits but your average internet user will have these features enabled.

Solved!

We had no solid clues on why the banner ad clicks weren’t showing up as visits on our clients site so we got in touch with our Media Buyer. We walked them through our findings and they began to dig on their end to really see why these numbers were so off. After much deliberation and some excellent sleuth work they came back with a pretty solid explanation of what was happening to our ads.

In their investigation they took a look at the advertising server logs that were hosting our ads. With these logs, they had the ability to see visitors and traffic to the ads and quickly noticed 18 IP addresses all originating from the same geographic location hitting our ads very frequently. This was odd and a very unnatural behavior and not reflective of a “real visitor.” In our discussion the conclusion that we came up with was that it appeared like our ads were getting hit with some sort of bot or spider. It looked like someone created a program that frequently hits the site’s we placed our ads and then would somehow generate a click on our ad.

Be on the lookout

It’s pretty safe to assume that our ads got hit with a bot that bloated our stats. The lesson we learned from this experience is that you should always try and verify your clicks and data to really see what’s going on with any form of advertising on the web. It may take some extra work or a few conversations with partners but figuring out this problem reassured us and our client about the campaign’s accuracy. Knowing that our ads got hit with some sort of nonhuman behavior our Media Buyer was able to filter out all of the bunk impressions and clicks. We’re now back down to a very reasonable 7,000 clicks. A little far away from 19,000 but we now know that at least 12,000 of our so-called reported clicks came from an unnatural source.