How to take Facebook Ad Optimization to the Next Level

Any digital marketers without deep Facebook expertise are bound to lose market shares and sales to more savvy competitors that know how to fully leverage this powerful ad tool. Yet Facebook is still in many ways new ground to a lot of marketers. In this post, we explain how to get to the next level with your Facebook advertising.

True, it is fairly simple to set up a campaign with Facebook’s ad management tool leading many marketers to believe that running Facebook campaigns is within their grasp. And it is. But, running fully optimized campaigns at a larger scale is a very different trade. What do you do if you are not quite hitting those KPI targets? And how do you scale your campaign to deliver five times the amount of sales while maintaining the great conversion rates you have worked so hard to achieve?

This kind of advanced knowledge is not as readily available and it requires a deep understanding of the underlying mechanisms of Facebook as an advertising platform. This article is going to share three advanced ways to optimize and scale your Facebook campaigns.

Perhaps the most underrated tool in Facebook’s ad solution stack is the lookalike audience feature. When it comes to effectively reaching new potential customers with whom a brand has not yet done business, there are few ad competitors who can keep up with Facebook’s lookalike audience solution.

On a basic level, you can let Facebook find other people who resemble your customers by selecting a group of customers, i.e. your ‘seed audience’. Your customers may be identified via the Facebook pixel on the website, the Facebook SDK on the mobile app or a csv upload of your customer emails and phone numbers which are then matched to Facebook users’ login information.

But, it is not as simple as pressing the “create” button. Most marketers are aware of the fact that their customer base can consist of many different types of customers. Or, your product or service is so universal that virtually anyone could be a potential customer for it. The old “crap in = crap out” principle applies to lookalike audiences as well: the more indiscriminate your seed audience is, the worse the results usually are. That is why the success of every lookalike campaign lies in the quality of its seed audience.

Three axes to think about when segmenting your customer base for the purpose of creating a lookalike audience are product, value and behavioral segmentation. Product and value segmentation should be fairly familiar: for a business with multiple products, use the customer base of each product separately to create a lookalike audience. Or, filter your customers based on the lifetime value and use the top 20% of your customers as a seed audience. Behavioral segmentation on the other hand might be a little bit more difficult to implement but can potentially yield the best results. Especially for multichannel retailers, it is important to differentiate whether you are trying to reach people who shop online, in-store or over the phone. It does not make a lot of sense to use in-store shoppers as a seed audience for generating online sales. The meticulous marketer can take this approach even one step further by creating a lookalike of customers who have purchased via a click on search ads as they might look entirely different from a lookalike of people who have purchased after clicking on a display or even a Facebook ad. The higher the overlap between who the seed audience is and what the marketer is trying to achieve the better the results are going to be.

One caveat: do always check for overlaps before taking a lot of different lookalike audiences live! Facebook allows you to analyze the overlap between different lookalike audiences so marketers will know ahead of time if it makes sense to target two different lookalike audiences separately or simply target one of them because of high overlap which almost always negatively impacts results.

The higher the overlap between who the seed audience is and what the marketer is trying to achieve the better the results are going to be.

To put some actual numbers behind these approaches: working with a technology company whose only goal was to directly generate sales from Facebook ads, the initial approach consisted of using all customers as a seed audience. After some testing and inadequate results, the initial customer audience was further distilled into customers whose purchases have been attributed to Facebook ads via the client’s internal attribution system.

The results speak for themselves: a 38% reduction in CPO and a 79% increase in budget spent. This resulted in more than twice as many sales generated through the adapted seed audience in the same 12-week campaign period! After optimization measures were taken, the seed audience was much closer aligned to the desired outcome the client was seeking and thus Facebook was able to deliver more results at a lower cost. This example shows that the choice of seed audience has tremendous impact on the results of a Facebook campaign.

Let’s turn the focus to the micro level. Determining the best settings for your media buy on Facebook can have a tremendous impact on the overall campaign success. Most media platforms used for performance advertising purposes have moved beyond mere CPM or CPC buying options. More and more research shows that there is no correlation between the effectiveness of an ad to generate clicks and its effectiveness to generate sales. So why would your media buy be focused on these metrics? This is why technology is evolving; adapting and enabling marketers to specifically buy impressions that are more likely to result in the desired action (sale, sign up, install etc.). In the Facebook world, this is done via the optimization goal of your campaign.

More and more research shows that there is no correlation between the effectiveness of an ad to generate clicks and its effectiveness to generate sales. So why would your media buy be focused on these metrics?

However, it would be simply too easy to just flip a switch and automatically achieve perfect results every time. So, sorry, but it does not work like that. The power of the optimization goal lies in the data it has available to form predictions and calculate likelihoods of generating the desired results for each impression. If the desired action is to generate sales you need to make sure your campaign is generating enough sales to use as data points for determining which impressions to bid on in the future. As a general rule of thumb, each campaign needs to:

• generate at least 25 actions per day and;
• achieve a view-through conversion rate of 0.05% in order to achieve optimal results.

What do you do if it is simply not possible to meet these requirements above? This is where the power of the “bid funnel” and “micro conversions” comes into play. In order for you to get the most out of your media buying settings, it is imperative to implement the Facebook pixel on a granular level that not only captures the ultimate goal (in this example sales) but also each step towards that final goal. Here is what an ideal pixel implementation would look like for an ecommerce shop:


If your campaign is running with this pixel implementation and it is not meeting the action count and conversion rate requirement for sales, you must move from the right side of the above illustration to the left and select the event furthest down the funnel that meets both requirements. So instead of bidding on a sparse event like sales, you are now letting Facebook use a much more frequent event like “initiate checkout” in order to make much better bidding decisions which will ultimately result in more sales and a higher return on ad spend. And in turn, changing your optimization goal to initiate checkout in this scenario might lead to such an increase in the events down the funnel (towards the right side) that you might be able to move up the optimization goal in a few days to “add payment” info or even “purchase” which will ensure continued improvement of your results. By knowing how to feed Facebook with more and better data, marketers can better leverage the power of the platform and increase return on investment.

The measurability of digital advertising has been touted as the ultimate benefit compared to traditional media, and it is very true that the digital era allows marketers to examine all kinds of metrics and data. It is possible to figure out how many people have seen an ad, clicked on it and then examine what they did on the landing page and if they ended up making a purchase.

All of this is incredibly useful, but, as technology advances, the question marketers ask of their measurement tools and providers needs to evolve as well. Marketers should change their question from “If I invest $10,000 on Facebook, how many purchases can I generate that at some point in the customer journey involved a click on a Facebook ad?” to “If I invest $10,000 on Facebook, how many incremental purchases can I generate that would not have happened if I had not made that investment?” The answer to the second question becomes infinitely more valuable because it allows the marketer to move his focus from optimizing statistics to actually maximizing revenue.

as technology advances, the question marketers ask of their measurement tools and providers needs to evolve as well.

This is where Facebook comes in. Due to the architecture of Facebook, with its required user login and the widespread use across all demographics, Facebook is able to perform true experiments. This experiment – dubbed the ‘Conversion Lift Study’ or ‘Brand Lift Study’ depending on whether you’re measuring sales or brand sentiments – uses a test group that is exposed to ads and a control group that is not exposed to ads. Both groups are filled in a randomized way so we can be sure that the makeup of the test and control group is identical. And since both groups are exposed to the same outside influences, i.e. other media, the difference in sales or brand lift can confidently be attributed to the exposure of ads in the Facebook ecosystem.

Currently, there are obvious limitations to this specific measurement tool, e.g. it only measures Facebook’s incremental results. But the sooner marketers make themselves familiar with this new approach, learn how to read insights from it and take actions based on those insights, the better prepared they will be when this type of measurement can be implemented across multiple channels and platforms. Whether it is Facebook that makes this kind of measurement available or a different provider.

This article is by no means exhaustive. There is a multitude of ways to manage and optimize your Facebook ad campaigns. The addressed topics (seed audience optimization, managing optimization goals and different forms of measurement) are just a sample of interesting approaches to think through when it comes to optimizing your Facebook ad activities. Without the knowledge on how to make all aspects of Facebook ads work towards your goals, it can become quite a costly endeavor, either in terms of unprofitable campaigns or lost opportunities for more profits from sub-par campaign performance. Needless to say, the knowledge of a true expert in this field can more than pay for itself. This is why ultimately the most knowledgeable, experienced and technologically savvy marketers and their agencies will end up winners in their markets.

This blog post is an adapted version of an article that featured in Leverate Media’s company magazine ‘Leverate ONE‘.