Facebook ECommerce Case Study – 9X ROAS & $2M in Sales

Problem Statement

  • To stand out in the highly competitive market, mostly due to the rise of the drop-shipping market.
  • The client was unable to scale up the campaign performance, revenue, as well as the budget, spend on Facebook


Pixels & Audience

The first step started with a proper pixel Setup so that we can catch all the events properly. On top of that, we created multiple versions of the custom and lookalike audience based on these events. Some points that we took care of are

  • Making sure all the Facebook events are firing at the right pages and on right page events
  • Advanced FB pixel matching for a better data-driven approach. Here is the Facebook official link if you are interested in learning more about it.
  • Setting up extra micro-conversion signals so that they will help us in creating the future custom audience

Delayed custom event

What you see in the image above is just an example of micro-conversion events that we implemented and used as micro-conversion signals to create a custom audience for remarketing campaigns and AdSets. Here are some examples of such events:

  • page view
  • category view
  • product view
  • search
  • Wishlist
  • Delayed page load by x amount of seconds
  • Contact etc.


Based on the above pixel, we divided our custom, lookalike audience and campaigns into Four stages of the funnel while setting them up

  • Prospecting
  • Consideration
  • Conversion
  • Retention

Prospecting Funnel Case Study

For targeting the prospecting part of the funnel, we just made one campaign with the objective of the landing page view. Within the campaign, we made several AdSets based on different interests and behavior. Because our budget was high, it enabled us to create multiple AdSets.

Prospecting AdSets

We brainstormed ideas on targeting different types of interests and clubbing the interest types in the same AdSet.

  • Consumer problems
  • Followers of top brands in the same niche
  • Generic eCommerce interests such as engaged shoppers, shoppers, discounts, etc.
  • Behavior-based AdSet such as in relationship status, dating, job, etc.

We made sure to stick to 4-5 interests per AdSet. If you do not have enough budget and still want to test, go with the generic interests as these interests group will have the highest number of audience pool and they also performed the best for us. Having separate AdSets with separate interests allowed us to test different interests.

Because we only wanted the fresh eye views on your website landing pages, we excluded most of the remarketing audience at the top of funnel campaigns. Facebook audience exclusion could be tricky. You can read more about this here on Facebook Audience Exclusion.

Excluding remarketing audience in prospecting campaigns

Excluding remarketing audience in prospecting campaigns

Prospecting Ads Type

For ads type under each AdSets, we created a theme of the ads. These themes were an offer, consultation, top seller, and category. These ads type further allowed to test the best performing ads and then pause the non-performing ones.

Prospecting ads type

All of the above AdSets and Ads went through the initial learning phase. If the results were too obvious to even before the initial learning phase completed, we will pause the non-performing assets. Learn more about the Initial learning phase on Facebook.

Facebook Consideration Level Funnel Case Study

For targeting the consideration part of the funnel or also the warm leads, we just made two campaigns with the objective of Add To Cart. Our idea was to bring users gradually down the funnel and also Facebook would have sooner enough data on Add to cart compared to purchase conversion, so optimization around add to cart will be quicker and more reliable. Here are the campaigns:

  • Static campaign and,
  • DPA campaign (Dynamic Product Ads Campaign)

Static Campaign & AdSets

For the static campaign, we found out that the optimal & average number of minimum product view required for conversion was 3. We used Google Analytics and Facebook Analytics to arrive at this conclusion from the historic data. So, we targeted the following audience per AdSet in the static campaign

  • Audience with product view less than 3 in the last 30 days
  • The audience which engaged with our Facebook or Instagram profile in any form (Warm leads)
  • Lookalike audience of those who had added to cart in last 7, 14, and 28 days

Again for the Ads type, you need to brainstorm ideas based on the audience persona. We decided to have different creatives and added the following ads type. It might be different for your industry.

Ads type for warm leads

Dynamic Campaign & AdSets

For Dynamic campaign, we created three AdSets based on the following conditions:

  • Viewed a product but did not add to cart in
    • last 7 days
    • 14 days
    • 28 days
  • Each AdSet had two types of ads for A/B testing
    • Dynamic Carousel and,
    • Dynamic Collection ads

The dynamic campaign does not need audience exclusion but we made sure to exclude all the audience from the bottom of funnel events such as converters, Add to cart, initiate checkout, etc.

Facebook Conversion Level Funnel Case Study

For targeting the conversion part of the funnel or also the hot leads, we just made one dynamic campaign with the objective of Purchase.  It was also because we had already targeted all the audience until Add to Cart step so, the target audience size was also going to be lower and we wanted to optimize our campaign super fast.

Dynamic Campaign & AdSets

For Dynamic campaign, we created three AdSets based on the following conditions:

  • Added a Product to the Cart but did not purchase in the
    • last 7 days
    • 14 days, and,
    • 28 days. Simple.
  • Each AdSet had two types of ads for A/B testing
    • Dynamic Carousel and,
    • Dynamic Collection ads

The dynamic campaign does not need audience exclusion so it was a smooth sailing ship.


Retention Campaigns

Retention campaign was introduced into the game after 30 days when we had already scaled up the campaign and we had started seeing a lot of traffic coming onto our website. For user retention, we just made one dynamic campaign with the objective of Purchase and another static campaign with purchase objective

Dynamic Campaign & AdSets

For Dynamic campaign, we created just one AdSet based on the following conditions that the audience performed a Purchase Event in the last 180 days but not in the last 30 days.

Static campaign – Buy Back

Since the product lifetime was around 30 days, we started retargeting the audience from different product categories as a reminder to buy the product back. One advantage of the retention campaign was that the cost per purchase was the lowest and also average order value became higher.

Optimization Techniques Used

I used a lot of optimization techniques at different levels in the funnel to scale up the campaign and optimize conversions. I have already covered them in some blogs. I will just list out the gist of the optimization techniques and link them back to the blogs where they have been covered in more detail.

Finding the Targeting Sweet spot

Finding the targeting sweet spot was mostly around finding out the right performing audience attributes. The above campaign structure already allowed to do A/B test at a different level in the funnel, however, from time to time, we also kept digging data to unearthed some missing opportunities. I have already covered here about finding the audience sweet spot on Facebook eCommerce.

Diagonizing the funnel

Diagonizing the funnel is important. We made sure that the audience was not leaking at any stage and also the audience was not facing the ad fatigue. But the overall idea of diagonalizing the funnel was to make sure that audience flow was not being bottlenecked at any stage. Some of the metrics that we always kept a watch on were clicks to view to add-to-cart to checkout to purchase. Again, I have covered this topic here on how to diagnose the Facebook funnel.

Average Order Value

A slight increase in Average Order Value can increase the ROI across the board. Some of the methods that we followed to increase the AoV were the upsell and bundled product. You can read here more about optimizing average order value.

Case Study Outcome

  • Amount Spent: ~$230,000
  • Conversion Value:  ~$2,000,000

Recent Last 90 days performance 


Resources for you:

Have a Question? Get in Touch!

One Reply to “Facebook ECommerce Case Study – 9X ROAS & $2M in Sales”

Leave a Reply

Your email address will not be published. Required fields are marked *