When running a Facebook Ads campaign and analyzing its results both in the Ads Manager panel and in Google Analytics, you may notice smaller or larger differences in the reported data. While small discrepancies usually do not cause much concern, larger ones start to raise suspicion. Does it mean that one of the tools is not working properly? Most often it does not. Below we will try to explain where the differences may come from.
Discrepancies in the analysis of the number of clicks and user sessions
The first differences between Google Analytics and Facebook statistics are already visible in the way clicks and user sessions are counted. Let us start by explaining these terms.
- Click-through is a parameter that adds up all user reactions to a given ad creative. So if you click an ad once and after a while another, Analytics will count it as two interactions with the ad.
- The session, in turn, shows how many users clicked the ad. Two or more clicks by the same user will be considered by the system as one session, provided they occur within a time window of 30 minutes. Only when the time interval between clicks is larger, the actions will be considered as two separate sessions.
Up to this point, everything is fairly straightforward, and it would seem that there is no room for discrepancies in the subcounts. And yet. Because at what point can you start counting users who actually responded to an ad? Google Analytics claims that it is only when the user is on the page to which the ad refers. After all, he could have clicked the ad by accident and stopped loading or closed the browser tab before the landing page fully loaded, and such an action can hardly be considered an ad success. In the Google Analytics system such “incomplete” visits are not counted.
Facebook, on the other hand, will record every click, i.e. user’s interaction with the ad, even if the recipient does not manage to fully see the landing page or does not leave the Facebook platform at all. Admittedly, Facebook’s Ad Manager can measure not only all clicks, but also the clicks of links themselves that lead outside the platform – to the advertiser’s website. Still, Facebook’s system will also count those clicks that did not lead to a full loading of the landing page. What is the conclusion? The number of clicks reported in Facebook Ads Manager may be higher than what we see in Google Analytics.
Tracking users across devices
So Analytics is definitely more accurate when it comes to counting sessions and clicks at the single device level. But what about the situation when a user uses different devices and clicks the ad, for example, first on his phone and then on his computer? Here Facebook copes with the counting much better. This is because in order to use Facebook, and thus click any ad there, the user must be logged in to his account. And most users use the same account on all devices, so Facebook’s algorithm has no problem distinguishing when it is dealing with the same user and when it is not.
In Google Analytics, the default data collection is based on cookies, and these are assigned to a specific device. Hence, any change of device by a user makes him/her qualified as a new person.
What does this mean in practice?
If a recipient clicks on an ad first on a phone and then responds to it again on a computer, Google Analytics will register them as two separate users from different traffic sources. Facebook, on the other hand, if the recipient is logged into the same account on both devices, will recognize them as the same person.
Differences in attribution models
A conversion can be preceded by a variety of user actions. This means that before he performs the goal we have set, such as buying an item in an online store, he may interact with the website in different ways. Let’s see it with an example. A user sees an ad for shoes from your store on Facebook. He clicks it, looks at the shoes, but decides not to buy. After a few days, he decides he needs shoes and types a search query into Google, for which he receives a search ad with a link to your site. He clicks on the link, sees the same shoes again, but decides to ask his wife for an opinion before buying. In the evening he will type the search query again, but he will not click on the sponsored link, but will choose the organic result, and this time he will buy the shoes.
So which interaction with the website can be attributed to this purchase? Was it the first contact with the offer, i.e. the Facebook ad, that decided the user to make the assumed conversion? Or maybe the user didn’t remember it anymore, when he searched for shoes in the search engine? Then it would be the sponsored link that made him come to your store and make a purchase there. But he didn’t buy the shoes through the sponsored link, but through the organic result, so maybe it should be “responsible” for the conversion? On the other hand, there is a chance that the user initially did not plan to buy shoes at all, and it was only the Facebook ad that awakened this need in him. In this situation, the first ad would have played a part in convincing him to buy.
The questions are constantly multiplying and it is not easy to find a clear answer to them. That is why different analytical tools may adopt different methods of calculating the effectiveness of ads, i.e. they have different attribution models. We will now show how this looks like for Facebook Ads Manager and Google Analytics.
Facebook Ads Manager – attribution model
If a user makes a conversion on the page within 7 days of clicking on the ad, Facebook Ads will attribute it to your ad (of course, if we have a well-configured FB pixel). The same will happen if the recipient completes the goal within one day of the ad being shown to them. Granted, the user may have still been exposed to ads in other channels later, but Facebook will still attribute this success to itself. This is now the default attribution model in this channel, which may be objectionable, but it has merit. Facebook is not a strictly sales channel, so the job of ads on this platform is to elicit a purchase intent from the user. So even if the ad did not lead to a purchase, but planted a seed in the user’s mind, which then germinated and bore fruit, this is in a way also due to the Facebook ad.
For example, before making a purchase, the user may have wanted to compare prices with competitors or look for opinions on the presented product, to finally go directly to the company’s website. However, he found it through a Facebook ad, so the system rightly recognizes its participation in such a conversion. Besides, Facebook’s system is unable to control whether the user was later still exposed to ads created by other systems, so assuming that the Facebook Ads campaign influenced the recipient’s decision is the best way to count conversions.
But is it reliable? Unfortunately, no. Clicking an ad is not tantamount to loading a landing page for this system, as we wrote above. It is even more difficult when we count views. Facebook’s algorithms take into account whether an ad has been displayed to a user, but they do not measure how long the ad has been visible on a user’s screen. So it may turn out that yes, the ad appeared on the page, but was scrolled through so quickly that the user didn’t even have a chance to notice it, let alone read its content. Nonetheless, if the user subsequently makes a conversion in the appropriate time frame, Facebook will count it as having been acquired through your ad.
Attribution models in Google Analytics
In Google Analytics another attribution model is used by default – the last indirect input is counted as the one that decided about the conversion. What is indirect input? In the simplest terms it can be any method of accessing the website apart from direct entering of its address in the browser bar. So if, for example, a user clicks on a Google Ads ad but doesn’t buy anything, then enters the site by typing its address into the browser and makes a conversion on the site, the credit for that conversion goes to Google Ads. So when comparing this counting system with Facebook’s, it’s easy to see that Facebook Ads campaign performance in Analytics may be weaker than Facebook’s stats.
However, Google Analytics also offers other attribution models, and changing the default settings can strongly change the results presented by this tool. We can choose from models such as:
- Last interaction – if this is the case, the system will give all the credit for the conversion to the channel the user used immediately before making the conversion. This setting makes sense when the ad is targeted at people who have made a decision to buy, and not at those who are at earlier stages of the purchase funnel.
- Last click Google Ads – 100% of the “responsibility” for the conversion is assigned to the last ad displayed by the Google Ads system, which the recipient clicked before making the conversion. It is worth using this model only when you want to compare the effectiveness of Google Ads. It is not applicable to statistics for Facebook ads.
- First interaction – in this model Analytics considers that all the credit for the conversion belongs to the channel through which the user first encountered our site. It can be used in this case when we want to build brand awareness with ads.
- Linear – recognizes that 1 conversion = 100%, and gives an equal percentage share to all the channels through which the user came into contact with the site. So if the user first saw an ad on Facebook, then clicked a Google Ads sponsored link and made a purchase, then 50% of the value of that one conversion will be attributed to Facebook Ads and the other 50% to Google Ads. It is best to use this model when every user contact with the page is important to us.
- Time distribution – assigns channels a different share of the conversion, depending on how many days before the conversion the user had contact with them. The closer in time the contact is to a particular channel, the more important that channel is.
- Item Consideration – This model also divides the share between the channels that were on the path to purchase, but does not take time into account. Most commonly, this model assigns 40% of the conversion value to the first interaction, 40% to the last interaction, and distributes the remaining 20% equally among the other channels on the purchase path.
Google Analytics or Facebook Ads – which statistics to take into account?
Advertisers often try to unify the results from both tools as much as possible. This can be done by changing the attribution in Facebook Ads, so that conversions are counted only within 1 day of clicking on the ad. However, before you take such steps, you should analyze your business and industry or track the average length of the conversion path of your audience. Sometimes it’s not necessarily about fixing discrepancies, but accepting that both Google Analytics and Facebook are complementary tools that can be used in parallel to better understand your business through web analytics and make better marketing decisions.