eCommerce Case study: Scaling a Facebook account by 352.81% compared to the previous year
Next we will show how we scaled the sales from a Facebook account by 352.81% compared to the same period of the previous year.
Our client had a Facebook Ads account with all settings done, customer acquisition and retargeting campaigns, multiple ads under test, Pixel installed and product feed set.
However, the account barely converts and the client feels that there is more potential on the market for their product. He asked us to conduct an audit and evaluate whether any improvements could be made to his ongoing campaigns.
If everything is set correctly and the account does not convert, where can the problem be? With this question in mind, we started analyzing the campaigns one by one in detail.
First we look at the figures and see the following:
The account has 235 conversions in the previous month and a cost / conversion of 10.01 USD.
Conversions are low and the cost is good, so the potential is quite high. In fact, the cost per conversion is less than half the maximum target on this account, of 22 $.
The figures are really good but when the result is so good and the campaigns are not scaled, it means that money is left on the table.for competitors to grab. You can lose money on expensive conversions, but also on potentially unrealized ones.
We first looked at the campaign with 225 conversions out of 235 total, and we see that it is a mixed campaign, with interest based ad sets and remarketing ads:
We do not recommend such a set-up as it may cause problems inside Facebook’s algorithm and seriously impact the delivery. Especially as a result of the changes made to Facebook’s budgets, which will have to be managed at the campaign level and in some ad sets may not be allocated enough budget, though it would be worth it.
Here we divide the analysis into two, the customer acquisition part and the retargeting part.
By “Acquisition” we mean attracting people who have never been to the site before. This is where interest targeting and lookalike come in.
We open the ad with the most conversions …
… and we see that there are a lot of interests added:
There are 10 indirect interests. For this customer we can not build audiences with direct interests for the products sold, so the problem is not here. The problem is that the added indirect interests are too niche, and have few people on them.
The system cannot find new customers if it has no choice because indirect and narrow audience does not have enough people to determine a customer profile.
There are also problems with ads, where we see some like these two, which cost per conversion double against the target of 25 USD, or even no conversion:
There are many such ads in your account this account and have been let to run for months without being stopped. This means that the cost per conversion at the account level of 10 USD could be even lower!
Retargeting across the entire account consists of an ad set with audiences of “all visitors” (the PageView event) for the last 30 and 60 days and excluding customers from the last 30 days:
It’s good that recent customers are excluded, we don’t want to overwhelm anyone with ads. But the campaign is not segmented and because of this the budget cannot be adjusted.
For example, people who add to the basket we would want to bid more than for those who bounce after 5 seconds without even seeing a product.
The current setting does not allow this.
There are still two key components to successfully promote an online store on Facebook, pixel events and a product feed.
These are not set, just the basic code. Without events we don’t know the value of conversions, the products purchased, the number of add to carts, whether or not someone saw the checkout page, etc.
That is why the conversion name is “Purchase CUSTOM”, because it is not tracked through the Purchase event but by the pixel on the order confirmation page.
Without events and without the product feed, dynamic carousels cannot be created. It is possible to make a manual carousels, but not dynamic because the system does not know what products a visitor was interested in and has no way to display the products afterwards. The dynamic carousels convert very well, which is why we’ve identified this as a lost opportunity for the customer.
The event pixel and feed are among the first things that need to be set up for an online store and this client is missing both.
In conclusion, the account is converting but far below it’s potential. It’s good that there are more ads being tested. But the area of interest is under-explored and retargeting is too general. The pixel is incomplete and the product feed is missing.
Also, cost management would have led anyone to turning off several unprofitable ads.
We decided to set up the account from scratch.
Full Pixel setting
One of the most important optimizations made is the Pixel event setting. The most important ones are ViewContent, placed on the product pages, AddToCart, placed on the “Add to cart” button and Purchase, placed on the view of the order confirmation page or on the “Send order” button.
Once the events were implemented on the site and verified, the data on the actions of the visitors to the site started to flow, so we had to create retargeting and then lookalike audiences.
Full product feed setting
Through dynamic remarketing, Facebook records what products a visitor viewed and displays ads with those products.
These ads convert very well and bring in new conversions from retatgeting, which is why an online store stands to lose if it doesn’t have the feed set, or if it has not been set up correctly.
It is important that the feed is updated on Facebook as often as possible, even if the prices do not change a whole year. Why? Because Facebook prefers to display ads with the freshest data.
Although we know we have the same prices, the system does not know this and if it sees that the feed has not been updated for 6 months we will lose impressions. So it is good to update the feed every hour so that the algorithm receives fresh information signals all the time:
Set Purchase Audiences
Since audiences with direct interests to the products sold cannot be set, we must go with the closest option, namely combinations of indirect interests and demographic data.
We’ve made several ad sets with combinations of indirect interests and demographic data that correspond to the established audience:
In the first month after launch almost all had a better conversion cost than the target and the volume of conversions increased considerably.
Set retargeting audiences
Because we now record events we can create ad sets of retargeting segmented by the pixel events recieved.
We segmented visitors into people who saw products but didn’t add to the basket, who added in the basket but didn’t buy and who bought.
We also added a segment on how recently the actions were performed and based on this double segmentation, action + time we created the ad sets.
Budgets are higher for people who recently added to cart and smaller for those who have seen products 30 days ago.
You can see how the cost per conversion varies by action + time:
Set up dynamic ads
Once the events were implemented and the feed we set dynamic ads for retargeting.
We worked on the dynamic carousels initially. The text of the carousels being with the benefits that customers experience if they buy from our store.
Launch your account
Once the setup was complete, we started the campaigns and started optimizing them daily. The results are below, viewed over several periods of time.
In the first four months after taking over the account conversions increased by 352.81% compared to the same period last year, from 604 to 2,735:
The cost per conversion, in the same period, decreased by 9.50%, from $18.16 to $16.43:
It’s worth setting up your account as detailed and segmented as possible, and testing more audiences and ads.
If we have an online store, dynamic remarketing is essential. We need to set both the Event Pixel and the Product Feed to convert campaigns.
Budgets should be tracked daily and adjusted to performance based on a simple principle: more on what goes and less on what doesn’t.