Expand my Community achievements bar.

Join us at Adobe Summit 2024 for the Coffee Break Q&A Live series, a unique opportunity to network with and learn from expert users, the Adobe product team, and Adobe partners in a small group, 30 minute AMA conversations.

Discrepancies between Adobe Analytics and Facebook

Avatar

Level 4

Hi all!

We are working on a Facebook campaign started on August 9th. We reviewd data from August 9th to August 29th. During this date range we get around 175.000 clicks to our site from Facebook, but in Adobe Analytics we get around 39.000 visits, coming from the campaing we launched. Also, if we check total visits to site, we have around 41.000 visits and around 43.000 pageviews in total. Site was created and focus on campaing we created, so it's expected not to have organic visits.

The thing is, even if we focus in total pageviews, there are huge discrepancies between clicked ads in Facebook and our data in Analytics. The site is for Latin America, so this is not related to cookies consent, because cookies are enabled for all users.

Does any one with similar issues? Any clue could helps to clarify what is happening?

4 Replies

Avatar

Community Advisor

Hi Oscar,

I believe cases like this is universal whether cookie policy is enabled or not as Facebook and Adobe Analytics data will never match. What I have observed is there is a usual drop off from Facebook traffic to the website, I have seen as high as 70%-80% eg 100 clicked on Facebook Ads, only 30 was recorded on the website. 

 

1. Users close the window even before your site has fully loaded. You can backtrack your company's performance's historical data on what is your drop off benchmark for Facebook.

2. Use FB link clicks rather than clicks. Clicks include all click instances eg click to Facebook page etc.

3. The website effect will also be dependent on the optimisation event for your ad. Landing page views will bring you more website lands vs optimised for Clicks/CPC- Facebook algorithm will serve your ads to as many users who have higher propensity to click than users who are likely to land on the website.

Avatar

Level 4

Hi @Benjienen

 

I know numbers will not match. This is something I've seen many times, using Adobe and other analytics tools. But the thing is, I've never seen such huge discrepancy.

 

I mean, we are comparing 175.000 vs 43.000, it's around 400% more reported clics in Facebook. Take note, we are not talking about ad display, we're talking about clicks to get to our site and we have around 400% discrepancy.

Avatar

Community Advisor

This is not unusual and sadly more common than people realize. I like to call it a sort of click fraud.

 

You mention its not about ads so its important to note are the links clicked truly viewed by user links or links multi scrolls down? Make sure you tag each CTA relevantly to allow you to differentiate referrer source even if its not an ad. Can be tough if it is a shared organic link I know

 

Also you can do a deeper dive analysis like I have done I will tell you my process and what I found:

If we are to rule out that browsers block/drop tagging code so we are in essence missing some of the users from the Analytics side we need to go low level. You will need access to your main webservers logs. These logs look at whats called GET requests, these request will show the actual referral URL(so anyone from Facebook). Whats great about this is that you can isolate by IP address/link and useragent.

 

Bottom line I looked at individual complete days 1 at a time. I isolated all facebook requests and even a specific page where my ad was targetting. I then matched them to IPs and I looked at actual date stamps. 

 

Off the top I noticed my numbers were pretty close to my analytics numbers maybe a few % higher as some users indeed seems like analytics didn't have time to fire but I saw they came to website and based on timestamps was for milliseconds.

 

I then started to notice some odd patterns.  So clicks were not coming as randomly as I would expect in fact most came in bunches. I would have thought every few seconds or minutes we see a new IP hit but I would see anywhere from 3 to 8 all within the same sec(just a few milliseconds difference). What also boggled my mind was some of these FB referrers accessed 3 different referrer links on the same page within a few milliseconds of each other. It looks like we have super fast users clicking 3 links in under 300 millisends.(A bit sarcasm here dare you to try and click that fast)

 

Bottom line seems there is a lot of bot like traffic and bunched traffic getting send from many of the display services. If the traffic I measured on my server side was this suspect imagine what traffic tricks the Ad displayer could be doing.

 

Something I have found to be really useful and allows me to strip out alot of the bot like traffic if you are using ads is just simply use floodlight pixels from FB or Google. Then make sure you create a 1 or 2 sec delay before the pixel fires.(so allow a real person to have the page load have a bit to actually see your page content).

 

Once I started doing this I suddenly saw my ad conversion rates more closely mirror what my analytics all along was showing.

 

Hope this helps you!

 

Pablo