There are 3 primary criteria to finding success with custom apparel e-commerce; niche, design, and targeting. We can illustrate the relationship of these three criteria with our custom apparel success matrix. When all three aspects align, the probability of a successful campaign greatly increases.
Most sellers will focus on a niche that can be easily targeted. But what can you do if your niche is difficult or impossible to define with the targeting available in Facebook advertising? In this case study we investigate ways to build your own community on Facebook to market your products to through the use of Custom Audiences.
Custom Audiences can be made in a number of ways and Facebook is constantly adding new ways to create them. We are able to use our own pixel data or user engagement on our pages. The most common ways of creating a Custom Audience for selling online is by segmenting people who have interacted with your page, viewed a specific campaign URL, completed a specific conversion event, or by manually uploading a list of customer emails. We do all of this within the Audiences section of your Facebook Business Manager account:
Once our Custom Audiences are created we can use them to make a Lookalike Audience. When creating a Lookalike Audience, the engine will analyze people within a Custom Audience and find the top 1% of users in a selected country who are most similar to them. To create a Custom Audience, you must have at least 20 people who fit the audience criteria. To create a Lookalike Audience from a Custom Audience, the Custom Audience must have at least 100 people in it. The more users you have for your Custom Audience, the more accurate the Lookalike Audience will be.
The niche used in this case study was difficult to define with targeting. This happens often when we want to sell a product for situations like a national holiday, a recent news event or a pop culture phenomenon. There was no existing community of people within our niche that we could just plug in and run with so we had to get creative with our targeting. We started by creating a new page and launched with 3 different designs which were marketed in their own ad campaigns. Both Page Post Engagement and Website Conversion ads were used and we tested over 5 different ad sets with unique targeting criteria for each design. The designs all had equal appeal to people within our niche and used the same design motif with similar graphics and humor. Here are the results for the first attempts with our original designs and targeting:
The results in our initial test were not ideal. On the surface these campaigns would not be considered successful and many sellers may decide to move on and try a new niche. But, with the use of custom audiences, we can take advantage of the pixel data and page engagement that we generated to find more people like the ones who are engaging or converting.
In our initial test, we learned a lot about our potential customers. We do this by looking at various statistical indicators like demographics, cost per thousand, click and conversion rates as well as other feedback we are able to analyze from comments and shares of our ads.
With the information we learned we were able to narrow down to the demographics that work best as well as preferences like product types, colors, and humor. We made subtle improvements to our original 3 designs and added 2 more for a total of 5. We kept the look and feel of our designs similar which established a motif that was able to make all 5 of our designs feel like a series.
Next we began to build various custom audiences. Ideally you would use conversion events that are down-funnel such as purchase or add to cart to segment your Custom Audiences but if your campaigns are new you may not have enough purchase or add to cart data to build audiences.
In this case we did not have enough purchase or add to cart data to create a Lookalike Audience from but we had plenty of page visits. We created Lookalike Audiences from Custom Audiences based on the following criteria:
With all 5 of our new designs, we created 4 unique ad sets with the Custom Audiences we created plus an additional ad set for remarketing. For each ad set we created a carousel ad featuring all 5 designs. We set all campaigns to end at the same time which allowed us to easily compare data and offer upsells on each. Campaign length was set to 7 days, at the end of the first 14 days using Custom Audiences, the results are as follows:
Although we didn’t see record breaking numbers, at the end of our first two relaunches of these campaigns we were able to create a very profitable campaign. 3 out of 4 custom audiences were profitable but the Lookalike Audience created from visitors to all of the campaigns was by far the best performing and most scalable. This is likely because the Custom Audience that this Lookalike was based on was large in size. Since the traffic continued to flow to the pages used in our Custom Audience criteria, the Lookalike Audience was receiving live data as the campaign was running allowing it to keep learning more about page visitors and adjust the audience. A Lookalike Audience will update every 3-7 days based on changes in the source audience.
Lookalike Audiences like this don’t make sense for every situation so it’s important to know when and how to use them. It will take time and experience for you to make the right judgment so we only recommend custom audiences for intermediate or advanced marketers. Keep this strategy in your playbook and you will find ways to make a turnaround on campaigns that aren’t quite becoming profitable on their own or if targeting for your niche is difficult to define.
