By Hakan Özçetin, Digital Ads Scientist, Realtime Agency
Traditional brick-and-mortar stores are embracing Ecommerce in unprecedented numbers as they attempt to capitalize on consumers’ dash to digital during the pandemic. It’s no wonder marketers are shifting budgets online: according to econsultancy.com, 45% of consumers said they’d bought a product online since Covid hit that they had only ever bought in-store before.
However, in the race for online revenues, the importance of ad creative is often overlooked. Sure, every marketer knows good creative is important – but what is “good” creative, and how important is it?
Facebook has declared that a staggering 60% of ad performance comes down to creative. If this stat seems high, it’s because it is. And as we enter a new era of customer privacy, the importance of creative continues to skyrocket from here.
Why? Let’s start with the Privacy Age. The Privacy Age is hallmarked by consumers’ reasonable desire to dictate how and when third parties use their online data. This is being propelled by Apple’s decision for iOS 14.5 to require mobile websites and apps to seek user permission to collect their data, and also Google’s promise to remove cookies from online marketing by 2024.
Cookies are crucial for retargeting strategies, lookalikes, first party data, and even second party audience formation.
But, most importantly, cookie data also informs algorithms that seek out consumers in places where marketing is likely to have most impact. Ad platform algorithms make sophisticated guesses about where and when an individual person will respond best to your ad, which ads will garner the most ideal responses, which people in your audience will respond best to which ad types, etc. No matter how much manual optimization your ops teams are doing, so much of your campaign performance relies on machine learning.
So, if cookies are removed from the digital marketer’s arsenal, brands face a snowballing issue of less data being tracked, which in turn makes algorithms less efficient, which in turn makes your campaigns produce less conversions in reality – not just tracked.
What can brands do to counteract this? They need to find new ways to support the algorithm in gaining more positive reinforcement and break this cycle, and creative/messaging optimization is the absolute best place to start. As mentioned, creative holds 60% of the responsibility for your ad performance, so if your creative is bad, you’re going to lose bids in auction – leaving the algorithm to its own devices and perpetuating the downwards cycle.
We asked in the beginning of the article what “good” creative looks like – and although this will vary from brand to brand, we want to give some recommendations of what we test when finding the most ideal creative format for our clients. Things you may want to consider testing are:
- Imagery and Color
- Ad Format
- Logo size and Placement
- In-creative Test/Messaging
- In-creative Call to Action
- Dynamic Overlays
Although it can and will be in many cases, testing needn’t be tedious, especially for smaller brands with limited budget. Because each variable needs to be tested against a constant for proper analysis to be done, we’d recommend testing one factor at a time with enough budget to ensure you can collect sufficient data (for conversion campaigns, this may be enough budget to ensure that creative has enough runway to produce 10 conversions or so).
Designing data-driven dynamic product feeds
Let’s look at this in action. We wanted to test and take advantage of dynamic ad elements for our interior-design client, Andrew Martin, which we knew had worked well for them in the past. So, with the help of our tech partner, Sprinklr, we developed new “dynamic image templates” for the brand – which are simply templates for ad creatives that update dynamically as information on performance per aspect is collected. For instance, if larger logos are performing best, the ads served will begin to contain more of our larger logos vs smaller. The creatives are then pulled through the API and served in Facebook ads.
Now – we mentioned that we run all tests against a constant variable – and this test was no exception. In running these dynamic ads against our standard, non-dynamic ads, we actually saw a 300% ROAS lift in the dynamic ones!
Keep in mind, the non-dynamic ads contained many of the same elements (images, messaging, etc) that the dynamic ones did. The difference is that the dynamic updates helped us gather learnings and optimize the creative much more quickly, and because we were optimizing more quickly, the algorithm was also being fed positive data more quickly, allowing it to do its job of finding purchasers more effectively. It’s a “help me help you” situation, which if you support the algorithm, it will support your campaigns back.
If you’re seeing diminishing returns on your ecommerce campaigns, we recommend investing your time in creative strategy and testing methodology that will spit back clear, concrete learnings. Without this, you risk making optimizations that not only aren’t statistically sound, but ultimately harm the efficiency of the algorithm, and continue to bring down results.