Literally all our customers see that Facebook figures on ads performance show a really nice ROI (Return On Investment), while Google Analytics, which is often used for comparison between advertising channels, leaves no reason for such optimism.
Here’s a live example from one of our customers that shows the inconsistency between Facebook and Google Analytics figures. According to Google Analytics, only 7.63% (342) of all 4400 purchases come from ads on Facebook.
Facebook conversion figures show as many as 784 conversions for the same period, which would make as much as 17% of all sales, in case properly attributed.
Why does this happen? Should you blame incorrect settings of your analytical tools or believe that Facebook figures are wrong? Neither of the above! Google and Facebook simply use different attribution mechanics: one focusing on actions like clicks, and the other focusing on the person performing the action. Let’s try to find out what is the best way to estimate the effectiveness of Facebook ads and to manage advertising budget.
Each sale has a path and a story. Marketing attribution reveals this story. Simply speaking, attribution is a way to determine which marketing channel or effort actually contributed to sales or conversions, what exactly happened between the first time a person saw your ads and the time they ended up purchasing your product. In reality, it is never that simple with numbers as far as marketing goes.
Marketing attribution is the practice of evaluating each and every touchpoint a customer meets on their way to purchase. The goal of attribution is to clarify what channels or messages were crucial for a customer’s decision to make purchase.
Let’s consider a person that sees a Facebook ad about an event. Just think, how often people purchase tickets to a concert seeing an ad for the first time. One is sure to consider if the date and time are convenient, if they are OK with getting to bed late and getting up for work early the next day, who to go with and what section of the venue to choose. This takes several days, and provided the event is no sooner that in a couple of months, no one is eager to run and buy tickets today.
For large business conferences the delay between seeing the ad and converting to purchase is even greater: one has to fit the conference into their business schedule and approve the ticket costs with the management. We are also accustomed to discounts and sell-outs before the event, and many people will simply wait for a better price to come up.
So it will be a standard path for a customer to see an ad on Facebook or in their Instagram feed, discuss the idea with friends or approve it with their boss, then 2 weeks later to search for the event on Google form a desktop, open the ticketing website and complete the purchase 2 hours later.
Types of attribution
Nowadays there are several popular attribution models used by marketers to get insights into how, where and when people interact with their ad messages, and enable marketers to tune campaigns to improve ROI. They can be referred to differently in different sources, but the most commonly used ones are:
- First click attribution model gives 100% of the credit to the first touch point on the path to purchase.
Considering our example with the indecisive customer, that ended up searching for tickets on Google, this model will likely attribute the sale to Social channel.
- Last click attribution model, on the contrary, attributes all results to the final step in the conversion path.
In our example, the purchase will be credited to Organic Search channel.
- Linear attribution model divides credit equally between all stages on the path to conversion.
All touchpoints on the path of our customer will be granted equal appraise, so Facebook and Google will get 50% of the credit each.
- Position-based attribution model enables to distribute the credit in a certain percentage between the touchpoints of the conversion path.
Depending on the percentage settings for the model, both Facebook and Google will get some (unequal) part of the credit in this purchase.
- Time decay attribution model assigns more conversion credit to touchpoints that are closer to the date of conversion.
Here it is clear that our indecisive customer took a really long time to finally get their concert tickets, so as a result the purchase is credited mostly to Organic Search, being the last stop on this long way to conversion.
User intent to rule the day
The cornerstone of this difference is the user intent. People use Facebook to relax, chill out and read the latest updates from their friends and popular accounts. They are not searching anything intentionally, so seeing ads in their newsfeed they are unlikely to rush for purchasing your products. You are rather building the idea or implanting interest to your product into the person’s mind than driving direct sales with your Facebook ads. Therefore we introduced our methodology of promotion on Facebook: creating Events to gather people, increasing their interest to the event with relevant posts, pushing them to purchase with ads and reminders. Most sales drawn by social networks will need several contacts with the potential customer and the actual sale can come from a last-click from another channel, often organic search.
As far as Google goes, people usually use the search engine to find something in particular. They have a particular clear-cut intention: to find some information, to purchase something. Thus, they are much closer to purchase than those who saw an ad on Facebook for the first time. In many respects, organic search in Google can help complete a purchase that actually started with Facebook ads.
Facebook attribution vs. Google Analytics attribution
For example, a user clicks on a Facebook ad for an interesting concert but decides not to buy tickets. The next day he changes his mind, googles the concert and clicks on a search result; ending up on the same website to make the purchase. Facebook takes full credit for the sale, but by default for Google Analytics only the last click matters (the sale is attributed to organic search channel), so Facebook gets nothing.
A user clicks on an interesting Facebook ad about a concert, but decides not to buy tickets.
The next day he changes his mind, googles the concert, clicks on a search result and completes the purchase.
Facebook attributes the sale to itself, Google Analytics attributes it to organic search. For Google Analytics only the last click matters, and Facebook gets no credit in the sale.
There are three major differences in the way Facebook and Google Analytics attribute sales.
1. By default, Google Analytics gives full credit to the last non-direct click that lead to purchase, while Facebook attributes itself all sales however matched with people who saw ads on Facebook.
This is the prime reason for numbers incompatibility for our customers: the default attribution model in Google Analytics is last click, which is rarely the right strategy for ticket sales.
2. Google Analytics is not able to track Facebook’s view-through conversions.
There is no way to track Facebook impressions in Google Analytics. Facebook’s default attribution window is 28 days post click and 1 day post view, but both are adjustable. This means that anyone who saw an ad but did not click on it during the timeframe will count as a conversion on Facebook but not on Google Analytics. To get the numbers closer to each other, you can choose to look at only post click conversions in Facebook reporting.
3. Google Analytics is not able to track cross-device conversions, while Facebook tracks activities of users across different devices.
One of Facebook’s biggest advantages is the ability to link different actions that lead to conversion with particular Facebook users. In practice this means you can track and target the same user across all their browsers and devices as long as they are signed in to Facebook. Google Analytics, on the other hand, relies only on cookies which means all tracking happens inside the same browser where the cookie was dropped.
Should I trust Facebook figures at all?
It seems that Facebook is too optimistic about ad results and tends to ascribe a bigger share of the result that it actually should. So does Facebook simply lie to me? Definitely not. Evaluating ad effectiveness, you simply cannot take ad impressions out of the equation.
Conversion lift tests show that Facebook should be credited more than it actually is, as long as the impact of ads is much more complicated than tracking direct clicks, and the remaining influence upon sales has a long ‘tail’ long after ads were shown. Such tests help to avoid comparison between incomparable entities, and focus on comparing the effectiveness of Facebook ads against one another.
The part of your audience that saw no ads on Facebook bought 100 tickets on your site during the study.
The other part targeted with Facebook ads bought 200 tickets, out of which 80 purchases were directly attributed to your ads.
As a result, you can make a conclusion, that the remaining 20 purchases were also affected by your ads and that the actual value of sales through Facebook is 20% higher than shown.
In simple words, such test means splitting your audience in half and only showing ads to one half while conversions from both are tracked. Facebook relies on cross-device tracking and view-through conversions for a reason.
How to fit a square peg into a round hole
So we have a pretty optimistic Facebook attribution model and a fairly pessimistic Google Analytics approach: can they become friends?
It is clear enough that to adequately estimate how effectively your ads work and to make budgeting decisions, you should consider all interactions of your audience with the ads. Hammering in nails with a microscope is a weak strategy, while much wiser is to use each tool according to the task it was designed for. Facebook provides an extensive toolkit to analyze ad effectiveness and to compare one ad to another, while Google Analytics is in any respect ideal to track Google Ads and organic search results. So we recommend to analyze the effectiveness of your Facebook ads with the reports provided by the social network. You can also take a look at some of the most common attribution models offered by Google Analytics, and consider which one works best with your sales model. Google Analytics offers its Model Comparison Tool, that ensures setting up individual attribution rules and seeing into multi-channel funnels to identify conversion paths.