The best way to continually improve paid media performance is by testing.
Without tests, you’re relying on guesswork, assumptions and other unreliable sources of truth. But with an intelligent framework, you can test, learn, iterate and optimize your paid media channels into high performing powerhouses for your brand.
Our agency maintains a wide scope of testing across our client base. From custom audiences to target segments to bidding styles to ad creative to video formats, there are very few things that we, as performance marketers, don’t test. And after working with dozens of the top DTC ecommerce brands around the world, we believe:
THAT your brand tests is important; but HOW your brand tests will make the difference between average results and significant growth.
In this new series, we’re going to be answering key questions about the testing mindset, frameworks for structuring tests and other curiosity driven digital marketing topics.
# # #
Our testing framework forces us to make data driven decisions for our clients. Even if we, as account managers, are not the target customer for our client, that doesn’t mean an ad won’t work. The data will tell us if it works. We’re forced to focus less on what we think and more on what we know. We trust the algorithm. Our teams are never scrambling for answers because we have a reliable source of truth to use when reporting to our clients.
In fact, many times we’ll run tests for our clients and the results will be unremarkable or inconclusive. And that’s okay. It’s an opportunity for us to engage our grit muscles and keep trying. Besides, the algorithms change so frequently. What worked last Summer during a sale might not work this Winter. Six months is a long time in digital marketing, and that’s why it’s typically worth a test anyway. There’s no harm in a redundant test, because it’s better to have data validation behind our guesses than assumptions without proof.
When testing, naming conventions are critical. Keep the file names the same for all of the creative ads in your set, with the main difference being a number as the suffix. This makes organizing and reporting much easier and cleaner. Think of naming conventions as a way to over communicate to yourself now to save time later. It helps you keep track of the specific assets you need, rather than having to search all of them every time.
For example, here are a few naming conventions from several of my client accounts (company or financial data NOT included for privacy purposes):
Building on the structure question, let’s talk about organization. Keep track of everything in a way where you don’t have to remember fancy naming structures makes your job easier. Complete standardization assures UTM effectiveness, which is essential for tracking via Google analytics as your source of truth. It’s also a benefit when you have multiple team members or an external agency working on the account. Also, from a time management standpoint, rather than stressing over structure, you can free up your time and energy on building and managing the ads themselves.
As for quantity, it depends on the size of your business. If you’re a large, established brand with tens of millions in revenue and a robust product line, then four to eight ads is ideal. While your newer, smaller and medium sized companies might only use two ad versions during creative testing.
For the ads themselves, there will be slightly different executions, in terms of the copy and the order of assets, but overall the creative changes will be minor. When we help our clients to create dedicated campaigns within their accounts, solely for testing, here’s how it works: Launch that ad set with a fairly low spend, and see which one or two wins. Typically you should be able to tell within a couple of days what the winner is. Then, take the champions and relocate them into the main campaign of the account.
Once you know what ad is your winner, serve only that ad to broader audiences. You want to make sure you’re finding the best ads and optimizing content before you put a lot of spend behind it. And be careful not to test too many, because having too many ads running at once means each ad will get so little ad spend, it makes it hard to get learnings.
For example, you might learn from your reports that all video ad creative under twenty seconds performs well. Or you might discover that your eight second videos have a two hundred percent ROAS. Or you might learn that your carousel ads with multiple products versus a single product have a higher conversion rate.
Awesome. Once you garner those insights, communicate to your creative team or agency what specific types of ads you’d like to replicate and optimize. If you know gifs work better than static images, don’t be afraid to ask for those assets. The whole point of data is to inform the next tactic.
# # #
Remember, guesses are what you think, but tests are what you know.
THAT your brand tests is important; but HOW your brand tests will make the difference between unremarkable results and significant growth.
Tune in next week for part two of our testing series!