Four ways how to track the real effect of Facebook Ads

At this time of big changes, businesses and marketers need to understand the real values of paid channels and their effect (incrementality).

This crisis creates lots of opportunities. Especially because of the competition’s growth, budgets decreasing and declining in customers income that makes them pickier. That’s why it is really important to understand how your marketing works.

Also, today, a person’s daily content consumption occurs across multiple devices, different flows, which makes cross-device conversion measurement more comprehensive and important than ever. Moreover, right now, it is almost impossible to track the full amount of marketing touches (clicks and impressions), because of data-safety from advertising platforms and regulations. By Facebook statistics, 37% of conversions are missed using cookie-based data, and most often users visit social networks from mobile devices.

Traditional types of marketing attributions (such as first, last click, time decay) are sometimes effective but have some limitations, especially in tracking the incrementality of every touch. Moreover, in my experience, some touches may harm the conversion journey probability (journey lengthening, journey mismatching, etc). But most popular attributions do not take count of this.

Few facts about Facebook ads:

1. Facebook Ads is a black box. We can’t understand the real impact of ads and their incrementality. Event credit attribution (7–28 window) is usually unrealistic and does not show us the real impact
2. The big picture and small text format. That is more useful for games segments, such as translation of demo of the game
3. Facebook does not share any raw data via API

Talking about incrementality in marketing, we need to take into count the classical brand awareness theory.

There are three types of awareness:

1. Aided. List of brands. Pick a list from which respondents can choose the brands they are aware of.
2. Unaided. All the brands that people know in a particular segment
3. Top-of-mind. The number one (first) brand in the customer’s mindset. Usually, the leader in top-of-mind = the leader in the market

The more awareness your brand has on the market (especially on a top-of-mind level), the harder it takes for you to achieve incrementality in marketing.

What can we do with incrementality? See this article.

So, what can we do to analyze the real incrementality of these ads?

1. Use data-driven Facebook attribution

Facebook has a built-in function to assess the contribution of each campaign or channel for various types of conversions (form sendings, registrations, trials, payments, and other metrics) by using the first, last touch, even credit, time decay, and data-driven attribution, including view and click touches.

Attribution makes it possible to assess the contribution of each touch to the conversion. The data-driven option shows the most accurate value for the campaign’s conversion weight, calculating values based on ML.

You can compare between different attributions in a different account level to catch the real Facebook performance results — overall and by the campaign.

Also, the biggest plus of Facebook attribution is that it includes Facebook impressions touches, that is not accessible via API, and we can’t track them by pixels in any other way.

To make full use of Facebook attribution, you need to configure connections between ad platforms to import their data to your Facebook ad account.

2. Push your traffic on the single landing page, that is not used in other channels of traffic

Usually, it is hard to understand incrementality because of the large number of traffic sources on every landing page.

To understand the real incrementality of Facebook, you can push all (or one of the segments) to the specific landing page, and use it only for running Facebook ads.

Further, it will be enough to calculate the number of traffic, conversions, from a specific page and campaign and their CAC.

For this purpose, you can create a new landing page or use the active one. But be careful with active ones — they must not acquire a huge amount of organic traffic, because it can affect the results of the incrementality test.

3. Stop Facebook Ads. Causal impact

You can just stop all the Facebook ads and track the differences in the overall number of conversions for some period.

To understand the Facebook impact, you can compare the number of conversions between the current and previous periods (be careful with seasonality and other external effects). But if your company has some analytical resources, the Causal impact approach seems to be one of the solutions. It is the Google algorithm that forecasts the number of conversions via correlated metrics. For example, you can forecast the number of overall payments in your company by building causal impact based on the real number of trials at this period. Of course, these two metrics need to be correlated with each other.

But, be careful, this can give you some negative effects:

1. Decreasing awareness in the channel. Marketing is not only about instant performance, but awareness. To get conversions, your potential clients need to know you, especially your brand and job stories, that you are trying to solve. It is well described in Phil Barden and Nir Eyal’s books.
2. Decreasing reach in acquisition stage users. Facebook has a different auditory than Google platforms and other social networks. Because of this, the remarketing audience will decrease too.

4. Create Facebook Conversion Lift research for your account

One of the best ways to track the Facebook Ads effect is to run Conversion Lift and do it as often as possible.

Conversion Lift algorithm is really simple and effective:

1. It divides your audience into two groups: test and control. And ads are only delivered to the test group. While control will not see any communications from your brands’ side. The audience is split 50% (test) to 50% (control). But you can request for resizing the share of groups.
2. Runs for your preferred period (weeks-months), so it makes it easier to achieve the statistical significance
3. Tracks different metrics at the same time
4. With various types of attribution windows. the iOS 14 update made this harder, but it is still possible in short windows.

You can run Conversion Lifts as fast as you need, one after another, in case of staying in touch with every change in your marketing and product performance.

Key pluses of this approach contain full cross-device tracking (Facebook is a social network with billions of users, they are really good in cross-device matching), full touches tracking (especially, including Facebook impressions; it is really important because of users type of social media content consumption (scrolling coma)).

Key limitations: the potential of losing awareness in the channel, because of big control group size (mostly for medium and big-size companies), no raw data, sometimes it is hard to achieve statistical significance in rare conversions.