Beta
Before Marpipe
After Marpipe
+32%
ROAS increase
3.52
Avg. ROAS
5
Designs tested
-49%
Lower CPA

Latico: Finding winning DPA designs that scale

The Enriched Catalog Designs

The first thing we did was sit down with Ben Schreiber, Head of E-commerce at Latico (and Founder of Brand Caffeine) to come up with 5 different enriched catalog designs to pit against the plain white catalog ads that they had been using.

Here’s what we ended up creating for our test:

Everyday Bag

Previous creative testing from Latico found that users gravitated towards the idea of the “Everyday Bag”. With this design we are experimenting to see if this previous learnings will result in performance wins for catalog ads as well.

Afterpay

Services like Afterpay and Klarna are a huge cheat code for DPA advertising. You have the ability to essentially advertise a reduced price without actually reducing the price. To show the amount per Afterpay payment all we had to do is add a divide modifier to dynamically display 1/4 of the current price.

20% Off

Historically we have seen that that strike through prices are very effective in Enriched Catalog designs. Luckily Latico has evergreen discount codes that are always available. To advertise the discounted price we used a divide modifier to dynamically display the price after the 20% discount.

Title By

Sometimes simple and elegant is the best course. In this design we’re making the product the main character by positioning it with a sense of importance.

Publication Logos

In this template we are highlighting the social proof that Latico has collected through placements in publications like Refinery29 and Glamour. We’ve placed the image to fit exactly in between the publication quote and the block of logos.

The Control

The control design is solely the plain image that appears in your feed as this represents exactly which creative is showing within the existing BAU catalog/DPA.

After publishing the 5 enriched designs and the control it was then time to take the new feeds and bring them into Meta for testing.

Creating catalogs for testing

Each of the 6 feeds that we made in Marpipe has it’s own unique URL that we can use to upload to a catalog within Commerce Manager. Before setting up the testing campaigns we need to create a new catalog for each of the designs we are testing.

Setting up 6 new catalogs really doesn’t take as long as you’d think. There isn’t actually a limit to the number of catalogs that you can make within commerce manager. All you really have to do is click “New Catalog”, enter a name, and then copy/paste the Marpipe URL into the data sources section.

To make things simple, I set up the new catalogs with a generic naming convention numbered 1-5 + a control. I made the names generic because I planned on re-using these same catalogs with different creative in the future. With this setup I can easily assign any given enriched design to any one of my 5 catalogs whenever I want.

The curse of the BAU catalog

You may have already been wondering - “why not just use this existing BAU catalog instead of creating a new control?”

Something we learned pretty early into enriching catalogs is that the power of the existing BAU catalog data makes it extremely difficult to compare creative from one catalog to another. Catalogs don’t work like standard ads, they save audience data outside of campaigns, ad sets, and ads. The data collected by a catalog works in such a way that you could create two entirely new ads using two different catalogs that are visually identical but the one that has been used previously will always dominate performance and eat a majority of the ad spend.

That is why for this test we are including a separate - brand new - control that is identical to the BAU catalog creative. That being said 🤪 — We are still going to temporarily include the BAU catalog in the test to prove how much of an impact it has (stay tuned).

Setting up the test in ads manager

To experiment with the best setup we tried out two different campaign types in this test. The first is a manual sales campaign with multiple ad sets to test different audiences. the second is an ASC+ sales campaign where we simply create an ad for each of the catalogs we are testing. Lets walk through exactly how each of these are setup.

Manual Sales ABO Campaign

To build out the first of our two campaigns we created a new manual sales campaign. To allow ourselves the opportunity to advertise multiple different catalogs we made sure not to check off “Use a catalog” at the campaign level.

Within the campaign we included 3 separate ad sets. Each ad set is targeted to a different audience.

  • Broad
  • Re-targeting
  • Customer list exclusion

Each ad set has the exact same set of 7 ads. The only difference between the ads are the catalogs that each are using as a source of creative. Since we did not toggle catalogs on in the campaign settings, we’re able to easily select the catalogs at the creative level. The 7 catalogs used in each ad set are as follows:

  • BAU Catalog
  • Control Catalog (identical to BAU)
  • Everyday Bag Catalog
  • Publication Logos Catalog
  • Afterpay Catalog
  • Title By Catalog
  • 20% Off Strikethrough Catalog

We plan on running this campaign for 14 days with a total budget of $120 a day. The daily budget for the re-targeting ad set was placed at $20 while the other 2 were set at $50.

Advantage+ Shopping Campaign

To setup the second of our two campaigns we start by making a new sales campaign and choosing Advantage+ Shopping. This campaign is going to be much simpler than the first campaign since Advantage+ shopping campaigns are locked to a single ad set.

In this campaign we did not include the BAU catalog or the Afterpay catalog but we did include the other 4 enriched catalogs + the control catalog.

The ad setting are identical to the ads within the first campaign but we also included a duplicate of each ad that featured an additional border frame. The border was added using Meta’s built in creative tool. The only reason we added this was to test out if there was any artificial benefit of using the built in Meta creative tool (spoiler alert: there was not).

We then went on to publish this test with an initial daily budget of $120.

Early on in the test

After the first couple days of the test,  our suspicions that the BAU catalog would destroy the test were entirely validated. Almost no spend was given to any ads other than the BAU in all 3 ad sets in campaign 1. Even the control catalog which is visually identical to the BAU catalog was given no love by the algorithm. This essentially makes testing new catalogs impossible.

So we did what any great experimenters would do and we turned off the BAU catalogs so that we could give our new catalogs a fair shake.

The test continues

After turning off the BAU catalog ads we were now able to see the winners begin to emerge. Across the board every ad using the 20% Off Strikethrough design seems to pulling away with the performance. Additionally we are seeing positive signs from both the Title By and Publication Logo designs.

Manual ABO Sales Campaign Week 1 Results:

In our first campaign we saw that the broad audience was far and away outperforming our other 2 ad sets so we made the decision to cut them and divert the campaign budget solely to the broad.

ASC+ Campaign Week 1 Results:

It does appear that the ad spend within this campaign was spread more liberally among the different catalog designs. Just as in the other campaign, the 20% Strikethrough design is dominating but the Publication Logo design is also establishing itself as a strong performer.

Last week of the test

After turning off bottom performers in each campaign, we see that the 20% Off Strikethrough design has clearly emerged as our winner across both campaigns. In the second week of testing we also saw that the ASC+ campaign performed much better than the Manual ABO campaign, receiving more than double the amount of purchases.

Manual ABO Campaign Week 2 Results:

ASC+ Campaign Week 2 Results:

Scaling the winner

By the end of the test it was extremely validating that the 20% Off Strikethrough design showed such strong performance in both campaigns. We decided to keep the control active the entire time just to give it a chance to find performance but throughout the 14 days the control was never able to match the top performer.

Since we had now established that our top design was guaranteed to outperform the plain untreated catalog images, it was time to bring the enriched design into the BAU catalog and evergreen campaigns.

Memorial day promo

Luckily the end of the test coincided right in line with a Memorial Day promo that was about to go live. Since the 20% Off Strikethrough design was perfectly situated to advertise this promo we decided to duplicate the design in Marpipe and make some subtle edits to adapt it for the new promo.

After publishing the MDW promo design as a new feed, we took the URL and pasted into the BAU catalog as a supplementary feed. By doing this we replaced the plain images in the BAU with the enriched images from our winning design.

Since the BAU catalog is already optimized for higher performance we expected that we should see even higher returns than the testing period from our winning design. To put that idea to the test we created a new ad within Latico’s main evergreen ASC+ campaign that utilized the BAU catalog augmented with our new design as the creative.

Here is the performance that we saw during the memorial day promo timeframe:

Almost immediately we saw the enriched catalog take off. By bringing the winning design into the BAU we saw almost double the ROAS compared to the average of the testing period. The Meta algorithm continued to funnel most of the spend towards this ad throughout the promo period establishing it as a cornerstone of the campaign.

When we look back at the past month of data for this evergreen campaign we see just how much performance our enriched catalog is bringing to the table.

Results

By testing multiple enriched catalog designs we were able to determine what Latico’s Meta audience was most influenced by. While the brand new catalogs in our testing campaign weren’t going to rival the performance of our existing BAU catalog, we were able to unlock even bigger returns from the BAU catalog by bringing in our winning design as a supplementary feed.

After introducing enriched catalogs into the Latico ad account we saw 50-80% ROAS growth compared to the previous period.

During the month of May the average ROAS across all other ads was 2.66 while the average ROAS of our winning enriched catalog ads was 3.52. That’s a 32% increase in performance. Additionally the enriched catalogs tested had a 49% lower cost per result compared to the account average.

“From 2019-2022 the wave was ads with User-Generated Content. Now that the effectiveness of UGC ads has slowed, the new wave will be enriched DPA’s. Normal DPA’s are so bland, the Marpipe solution of dynamic creative automation of these carousels is game changing. Potential customers want to see optimized product options. The more products they see there’s a higher change they find what they’re looking for. TLDR: Screw your white background DPA’s. Enhance DPA creative and skyrocket ROAS & efficiency."
- Ben Schreiber (Head of Ecom, Latico / Founder, Brand Caffeine)

Continue to Demo ➜
SolutionsMultivariate Testing
InformationPricingRelease NotesBlogPressHelpContact UsNewsletter Signup
CompanyAbout UsPrivacy PolicyTerms of Service
Careers
Hiring
BlogHow to Write Compelling Ad Headlines for Creative TestingRelease NotesWhat a Losing Ad Tells Us About How to Make a WinnerNew Research Shows Marketers are Really Bad at Predicting Winning Ad CreativeWhy Multivariate Testing is the New Standard in AdvertisingMore
Follow Us