How We Increased Fashion eCommerce's ROAS by +24% With Smarter Product Testing!
Today I would like to tell you about a recent experiment we did for an online fashion retail client. It will also be relevant for any other businesses that have a very visual product and rely on having a lot of SKU’s.
It’s quite exciting as this Facebook ads experiment we did, increased ROAS (return on ad spent) from 2.2 to 3.17 for product testing campaigns — a 44% increase. While at the same time decreasing ad management work hours and increasing overall store returns by 24% from 1.98 to 2.46, just by having more winning products running at any given time.
With a clothing store, you often have the issue where there are hundreds of products - anything from socks to jackets.
Deciding which products to put on an ad that might be the “winners” is difficult, but will make or break your campaign.
Typically, what people do is select a few products that they like, create separate ads for them and test them out. Often this process in and of itself can take a good 5-6 hours.
What we’ve experimented with is to upload their entire store's products catalog on Facebook. Then use this catalog to test out products.
This is known as a Display Product Ad, otherwise known as a DPA, and it’s a powerful tool as it uses Facebook’s knowledge of their users to match your products with the needs of your potential clients.
Yet, as good as Facebook’s algorithms are, and as much as it understands its users, it’s still whittling a list of hundreds of items.
One day, we began wondering whether we could optimize this algorithm. So, we looked at the data — specifically, the top ten best selling items within a week in our Shopify store.
Then, we created a product catalog subsegment with only these top-performing items and used this list as the basis for that week’s ads. This resulted in two different types of ads that used this information to its maximum potential:
- Automatic DPA: Here, we used Facebook’s algorithms to filter through the smaller product catalog list and present the users with the items that Facebook’s data on them implies that they would like.
- Manual DPA Catalog Ads: Given that we already have a shortened list of the products that we know are the likeliest to sell, we can customize the ads to be more bespoke.
Obviously, it’s more time consuming as you can create carousel type ads, group them in specific patterns, and create ad copy that is likelier to emotionally appeal to the audience. This approach is more involved but it has a high ROI.
It bears saying that the primary benefit of reducing the product catalog list and updating it weekly, is that you are using all available information.
In essence, you are using what in economics is known as “the wisdom of the crowds,” meaning the unconscious decisions of thousands of people help you make smarter decisions than had you done it by yourself.
If you randomly pick ten products, chances are that you wouldn’t have selected the best-performing ones, as you would’ve let your own biases make the decision for you.
In the same way, if you allow Facebook to algorithmically go through your entire product catalog and pair items with customers, it might make the wrong decisions. Not to say that it doesn’t understand its users.
However, Facebook might match a jacket with a potential customer, who might be interested in buying it, but get the timing completely wrong.
It can serve this ad to a user in the middle of Summer when nobody is buying jackets.
Plus you would need to spend a lot of money on Facebook to test through hundreds of products and find the best solutions.
By using the top ten best selling items within a seven-day window to make your Facebook ads, you are ensuring that the product list available is relevant and desirable to the users.
And there you have it!
This is how you can supercharge your ROAS if you have multiple SKU’s in the store and which require finding a few winning products to capitalize on them.
What experiment is working for you lately? Would love to hear about it!