If you are reading this case study for the first time, we guarantee you that this is the best decision you have ever made for your business. We will reveal one of the most amazing stories that made us famous.
The case study examines how we took a popular brand that spent $1,527,036.15 in ads in 6 months to turning over $7,493,495.75 with an ROI of 2.89X. We will explain how we improved their ROI by 233% from 1.24X to 2.89X in 180 days.
Get ready, as the ideas and hacks we included in this post are worth millions to some brands.
A little about the client.
The client is a men’s fashion store. It is one of the first companies to take advantage of dropshipping from major marketplaces like Amazon, Alibaba, and AliExpress, and has around
4,600,000 followers. It was the fastest growing online store between 2012 and 2013 and many success stories were written about this brand.
When we started working with the store, it owned a private label clothing. It also had its own fulfillment center that catered to all of its customers worldwide.
The Client’s Goal. This client had its own media buying team and was struggling to create profitable Facebook ads since the items it sold were priced so low. As a result, there was not enough room for advertising spend.
The client’s marketing team consisted of two media buyers, one designer, a copywriter, and an editor. Implementing profitable paid traffic was hard for them and their greatest asset was their Facebook page.
With 4.6M followers, every post they created got massive amounts of traffic and conversions completely FREE. They understood that Facebook is moving toward the pay to play model and soon organic reach will be limited even more.
Their problem was that they couldn’t make their campaigns break even. They lost a lot of money from paid traffic. It may not be a lot to them, but as an experienced agency, our philosophy is that every dollar counts.
Here are the results they got for the previous 6 months before hiring our agency.
1.24X ROAS. $1.52M spent on ads. $1.89M in sales generated.
We got to work, deploying three frameworks:
Discovery mode, strategy, and execution.
Step One: Discovery Mode.
Our goal as a marketing company is to gain as much past knowledge as possible and dig into the entire brand’s presence online. In order to market something, you need to know the product or service like the back of your hand. The strengths, weaknesses, challenges, questions, and everything that can help us to craft the most efficient campaign. In fact, we don’t publish or create anything during this phase; we only research, examine, and discover. Nothing else is done during this phase. In this case, since we are talking about massive amounts of data and integrations, it took us over two weeks.
Below were our most significant findings:
- Extreme organic reach on social media platforms. Perhaps, it was the number of followers that led to massive organic reach or perhaps, it was the items. But even though the organic reach on Facebook was very limited, posts were still reaching 8K likes organically before we even promoted them.
See some examples below.
As you can see, the reach was insane. That’s without being promoted, and those posts were at the end of 2018. We found this to be very rare. We knew that something about these followers was different; they engaged more and are top followers. We needed to find a way to take advantage of this.
- Unbeatable low-priced items. We researched everywhere and noticed something very strange. Even though the client started as a pure dropshipper, his prices were the same as the major marketplaces like Alibaba, AliExpress, and others.
His buying power became so powerful, so you couldn’t find better prices anywhere. The shocking news? He never mentioned prices in his ads or posts. They were nowhere to be found. His strongest selling point was hidden from his ads for no obvious reason. What else was missing? Low price guarantee.
- Our analyst took a dip dive into the complex analytics and report and found another interesting fact. Although he had more than 4000 items in his store, only 41 products accounted for 69% of his total sales in one year. How insane is that?.
The top 20% of items usually account for 80% of total store sales or more, but we are not talking about 20% here; we are talking about 41 products out of 4000, which is 1%.
1% of inventory accounting for 69% of the sales is something extreme. An even stranger thing was that this pattern repeated itself in similar years when we took a look at the previous years. We presented it to the client in a way that he is able to grab the information.
We explained how important these items were in terms of the campaigns and how important it is that they remain in stock at all times. Every time these items were out of stock, the learning system of the campaigns took a major hit that would be hard to recover from.
Why? Imagine turning from 2 months of the conversion rate of this product from 3.5% to 0% since the product is out of stock. It had a dramatic impact on the algorithm optimization of Facebook ads and hurts the performance of the store.
Our immediate recommendation was to use the predictive inventory model to better prepare so he doesn’t run out of stock. In addition, instead of closing your door for business and noting that the product is out of stock, just add a backorder feature that allows you to continue to accept sales and ship later.
The conversion rate on the page won’t be as great, but it was better than nothing.
- Zero branding and poor copy and storytelling. The creatives were great, the prices were amazing, but the branding was horrible. Imagine going into a product page without having any copy to set the tone — only technical details.
The art of marketing is selling a story, not a product.
Customers might want a shirt, not because they need it but because they like how it looks in the picture. They like the model, the fitting, the background, and the story.
Without branding, your website becomes unmemorable and generic. Instead of being sold on the creative and copy, a customer might just compare prices instead and search online to see if they can find anything similar for a better price.
When you sell a story, people won’t rush to look for alternatives; they will want the real thing.
- The most interesting insight we found during the discovery mode was how popular this brand was in developing countries. While the US holds 30-40% of the global market in online sales, only 20% of his customers were from the US.
The strange thing was that the US was his lowest-performing country, but 50% of his ad spend was targeting the US. Countries from Asia, Eastern Europe, Middle East, and South America were booming countries for his business and the price per impression in those countries was a steal.
It took us a lot longer than we planned, but it’s always better to be very prepared and triple-check your homework. The Discovery stage was completed and we were ready to move on to the next step — strategy.
Step Two: Strategy Talk.
- Quality over quantity. We understand that it’s a numbers game. So if 41 items accounted for 69% of total sales, we knew that our entire focus had to be on these bestsellers. Their add to cart and buy to detail rate were 2-3X better than any other product. That means that if the buy to detail rate is 2X better than the average, the CPA of those ads will be 50% cheaper and the ROAS will be 2X greater.
Here are some examples of the top-performing items based on a 14-day time frame.
If the products convert at a rate of 4.5% while their competitor’s products convert at a 1.5% conversion rate (industry standard for prospecting traffic), it would mean that the cost per purchase will be 66% lower than their competitors.
We selected all the products in the category of monthly best sellers that they recorded more than 4% buy to detail rate for and got a list of 16 products based on a 30-day time frame. Those products will serve as the core products for the first month’s campaigns.
We created simple slideshows using these items and the strongest selling points (item price and lowest price guarantee) on all creatives. In addition, we created high converting short stories using best practices to take advantage of the low cost per impression on this placement.
In addition, we created a product set on the Facebook catalog to be able to dynamically showcase products from this 16 product list we created. This allows us to use the optimization algorithm to find a better match between potential audience and the exact products that they might be interested in.
- Organic to paid ads. We identified organic reach and massive social proof on posts, which are the most valuable assets of the brand. We then created the page posts conversion objective hack to achieve thunder-fast results.
Instead of running hundreds of different ads for the same items in different variations, we focused on the 16 posts for the 16 products we selected.
Think about this mind-blowing hack for a second…
Instead of having a new ad that has zero engagement, we ran the ad with a post that has 10K likes to start with. Facebook ranked the engagement high and the ad got an exceptional relevance score, which led to insane results. You’ll see the results in the end, but here’s a spoiler.
This is from one ad set with one page post.
ROI: 13.56X. 1456 Purchases. $4.73 cost per purchase. WOW.
- Capitalizing on page followers. With the death of free organic reach on Facebook, having a page with 4.6M followers is rare, especially with a strong engagement rate. Even huge brands that are known worldwide do not have as many followers. He was operating strictly online
We immediately decided to focus on this core audience in order to take full advantage. It meant more focus on recent followers, people who engaged with the page posts and all the ads focused on the entire followers. So instead of just reaching 5-10% of the people who follow the page, our goal was to reach 20-30%.
- Optimize by value, not just for conversions. Most advertisers are not familiar with this feature. Value optimization lets you optimize not only for purchases but for the purchasers who are likely to spend the most money. One of the issues we noticed was the extremely low average order value the store had.
Customers were mostly purchasing 1 item and leaving the store. We knew the issue wasn’t the website but was more with the customers.
If we could drive more valuable customers that spend more money, we can dramatically increase the average order value.
More info on value optimization.
- CBO Remarketing on STEROIDS. One mistake advertisers make is that they don’t invest enough in retargeting campaigns. They spend massive amounts of money on getting people to check out but don’t put great efforts on remarketing campaigns for people who forgot to check out — those who did not make up their minds.
We created the following remarketing campaigns:
- People Who added or viewed the products but didn’t purchase within 3 days (hot leads – 10% coupon codes in each ad).
- People who viewed or added items in the cart within the last 7 days but didn’t purchase (15% & FREE shipping coupon).
- People who viewed or added items in the cart within the last 14 days but didn’t purchase (ads with testimonials and social proof to demonstrate brand credibility and authority).
- Top 25% of users, based on the time spent on our website in the last 14 days, who haven’t purchased in the last 14 days. (Dynamic ads with catalog sales with a 15% coupon code, as these users were spending a lot of time on the website but for some reason didn’t find what they were looking for.)
- People who viewed or added items in the cart within the last 30 days but didn’t purchase (a mixture of testimonials ads, social proof, and dynamic ads with coupon codes).
We used campaign budget optimization and started with a 50% lower bid cap than cold traffic. This ensures that remarketing campaigns are super profitable and our daily budget was $750.
Using campaign budget optimization allowed us to scale faster and let Facebook optimization algorithm to shift budget across these audiences when it saw more opportunities.
When we had strong days with massive amounts of traffic, the 3-day audience performed better, as there were more volume and opportunities.
On slower days, more budget was shifted toward the larger groups of 14-30 days audience for more volume at a better price.
Now that we are done with strategy, it was time to move to execution.
Step Three Execution.
After the discovery and strategy phases were completed, it was time to execute. Here is how we structured our entire marketing campaigns.
Ads Structure: In order to keep the number of campaigns at a minimum, we created the following campaigns.
A) Prospecting Campaigns.
1- Campaigns with existing posts, targeting worldwide audiences, optimizing for conversions. In the campaigns, we had 7 different ad sets with completely broad targeting and selected only the relevant demographics and geo-targeting.
The rest was left blank to allow Facebook complete freedom to deliver the most effective results. In each ad set, we had three page posts that would compete against each other.
We could have used one page post for each ad set as we knew they were all top-performing posts. Instead, we picked three to accelerate the ad set learning phase and have 3X the amount of data in each ad set.
Then, the algorithm will have the option to alter delivery between the ads based on demand and opportunities. Rather than forcing it to continuously push on a specific ad, we gave it three options to allow for flexibility and increased opportunities.
2- The next campaign was for the custom creatives we created for the top 21 performing products based on the research we did above in the strategy section. In this case, we did create 21 different ad sets, as the goal here was not for products to compete with each other but for the creative styles and ad formats to compete within the same ad set.
For each product, we created 6-7 different creative variations ranging from stories, slideshows, creative videos, static image ad and carousel and tested which works the best.
Each format type was labeled in the ads manager with a specific name so we can later see across the board which format works the best for this brand’s audience. This allowed us to better prepare creatives for the following month.
Here is an example:
3- The next campaign was for catalog sales ads. In this case, since we wanted to take full advantage of the DABA feature (Dynamic ads for broad audience), we used 1 ad set with 3 different ad copy variations. No targeting, no restrictions, and all the products from our catalog.
We would be foolish not to take advantage of the best of what the algorithm optimization can offer, so we left that option on the table.
For every campaign we created above, we then duplicated and labeled the duplication Value and changed the optimization method to value campaigns.
We ensured that every campaign targets value, as regular conversion optimization would allow us to ensure that we get the best from both worlds — optimizing to get the lowest cost per purchase and optimizing toward the highest spenders and getting the greatest ROI.
The tradeoff is more learning systems and less condensed data groups, but we didn’t want to rely solely on one optimization method. We wanted to have both options available.
B) Existing Followers campaigns.
We duplicated the campaigns for prospecting above with a few changes. Instead of running them for prospecting traffic, we included only people who like and follow our brand.
Instead of duplicating all three different types of campaigns, we only included the promoted posts’ campaigns and the catalog sales’ campaigns.
We didn’t see the need to duplicate the custom creatives campaign we created (#2) until we see which one works best and is most effective.
There’s no point running the same test twice on two different campaigns. Like the first campaign, we duplicated these campaigns.
We created a catalog sales campaign for the remarketing audiences we created with every audience group, which is separated by different ad sets with CBO. Each ad set included 3 different variations of copies and the complete product catalog.
This allows Facebook to dynamically show personalized product recommendations the user can view and might be interested in.
We didn’t create another duplicate for value since we are retargeting our existing audiences. We wanted to include anyone relevant and not segment further by only targeting the top spenders.
Budgets & Bidding
We didn’t use automated bids at all because we wanted full control over delivery and results. Ad accounts that have more than 50 conversions per week are able to leverage cost controls more effectively.
Facebook doesn’t just have access to the 50 conversions per week; it has access to 180 days of data. 50 conversions per week can be likened to 25 weeks’ worth of data. In our case, it was thousands of conversions per week, so the data is massive. Bid caps allow you to tell Facebook what your goals are.
If it doesn’t it doesn’t predict that it will be able to deliver the desired results based on your bid caps, delivery will be limited. Instead, the algorithm will look for other auction opportunities that are likely to provide results that align with your goal.
We used bid caps and not target costs, as bid caps are a way to control the maximum you are willing to spend per conversion while picking up a lot of cheaper conversion opportunities at a low cost.
If we had used target bids, we would have had solid and steady results, but bidding would be more aggressive, so we would end up missing opportunities on low-cost conversions as well.
- Our initial bid cap was 20% higher than our target CPA for optimize-by-conversion campaigns.
- For value campaigns, the minimum ROI bid cap was 1.8X, as our goal was a minimum of 2X. We bid slightly lower to ensure we don’t miss relevant audience because of the under-reporting by Facebook pixel or wrong predictions of the algorithm.
One of the greatest mistakes advertisers tend to make is to optimize results from the same day. One day is 24 hours and 24 hours mean nothing when it comes to your campaigns. Facebook is not designed to work well for instant results since you are not targeting people based on what they are searching, but based on the interests they might have.
We only analyzed our campaigns based on a 7-day window. If the campaigns were performing well, under our target CPA, we increased the campaign’s daily budget and bids by 15-25%. We only edited ads once every 48 hours to ensure we let the optimization algorithm work without interference.
We didn’t pause any ad sets and there were no ads; only campaigns. If you understand the dynamics of campaign budget optimization, you know that there is a difference when it comes to evaluating the campaigns on a campaign basis and not based on the ad set level. For campaigns that performed badly, we lowered the bids.
If lowering the bids didn’t improve performance, we paused the campaign once it reached a minimum of $1000 spend and the actual ROI was 30% or lower than our target.
Our take on optimization is to have as little optimization and interference as possible. Let the machine do the work and re-optimize using our data and feedback.
Every week, we created new creatives and analyzed new best sellers and new data. We launched new campaigns weekly with the new best sellers or products that we saw have potential via analytics and insights but didn’t get enough exposure. This framework continued and the cycle was repeated.
Fast forward to 180 days of work, more than 1100 hours of work from our end, and here are the results..
Total ad spend in 6 months: $2,592,881.339
Total revenue generated: $7,493,495.75
Total ROAS: 2.89X ROI.
Average cost per purchase: $18.13
Total impressions: 60,197,953…
Total Reach: 56,699,829. 56 million people were reached.
Link Clicks: 6,715,332
Post Engagement: 30,272,957
We saw incredible results on some of the ads. ROI of 8.82% and cost per purchase of $8.11.
Here is an example of our lowest cost per purchase campaign with the lowest CPA.
$4.90 average CPA and an ROI of 10.21. Wow.
But ROI is not the whole story here. Besides the $7.4M generated in sales, think about the bigger picture here. Is 2.89X the real ROI?
What about all the customers who will come back and make another purchase, and another purchase?
That ROI will skyrocket to 4X or 5X or even higher. That would depend on how you measure it. 6.7M people clicked on the ad and visited the website. Imagine having a real store and driving 6.7M people in.
Take note of all the followers that were gained in the process, the email list that grew to massive numbers, and the people who are now familiar with the brand.
There are brands that are willing to break even on their paid advertising and use it strictly as an engine for growth, as they understand the value of a strong repeat customer base.
And they understand the real value in the long term more than just evaluating the dry metrics.
You can’t put a price tag on that, can you?