Recently, I was asked by a company to run some test Facebook ads and run analysis on the amount of traffic coming from the paid social media channel and find the conversion percentage. What I found was very interesting - and somewhat alarming.
Facebook is a great source of traffic
Let me start by saying a majority of companies use Facebook as a main source of generating traffic, conversions and revenue. Large corporations have made billions of dollars by both organic social media and paid social media and these are legitimate channels to focus on. That being said - after I ran a test ppc campaign (pay per click) for a local company, I found that there were some discrepancies between how many clicks to the site Facebook was saying had occurred, and what was ACTUALLY showing on Google Analytics. Weird right? Believe me? Let's find out.
How did I find this?
It started by the amount of traffic and the lack of Google events occurring when these users were visiting the site. Google events fire at specific times during the user journey and can be configured to fire and send data to GA whenever we want. Well, this company had a popup window that shows after 5 seconds of the user being on the site - which at that time - an event fires and tells GA that the user viewed the popup.
The strange part is...none of the users that had visited the site on the same days the campaign had run, triggered this event. Hmmm...
So, the first thought was "are users leaving before 5 seconds? Who does that?"
The answer to that question is - drumroll please!
Enter the Bot Analysis
It's very simple to run a Bot Spotter through Google Analytics and filter out traffic that is actually these little creepy crawlies.
The first step I took was looking at the screen resolution and service provider for all of these users.
Notice anything strange here?
1. Facebook Ireland is where most of the traffic is coming from. This is a server farm.
2. When is the last time you saw a computer monitor that was 2000x2000? This would be a perfectly square monitor.
More signs pointing to a bot.
After discovering this, I built a custom report that shows all of these things side-by-side:
So, we are getting some interesting findings here for the dates of our Ad Campaign. Notice that all of these users have a session duration of 0 seconds? This means that user wasn't even on the site for more than a second. These users are also almost ALL from facebook. The total users is at 11.
Let's look at the Facebook Data
Now that we have our total users - which is 11 - our source and our dates, let's look at Facebook Ads Manager.
Not the same.
Typically this ok and you will see discrepancies between platforms. However, Facebook is saying that for the same dates, they sent over 30 users to the site. I thought we had 11?
So who is right?
Well, Google is the one that you can test yourself to ensure that these events are firing. When you do, you should see that the screen res, the network and session duration should all reflect accurately.
Plus, Google isn't making money off you.
So, that just excluded 80% of the traffic from the last week.
What if they had made business decisions based off that data? The would have made a decision based off bad data .
1. Facebook ads work, but just keep an eye on things.
2. Run small tests to see if things are working before you dump money in it.
3. If you aren't sure if they are legitimate users, have someone check for you.
I hope this article helped!
Please leave a comment if you have any questions!