The value of always-on ad creative testing

The value of always-on ad creative testing

December 13, 2021

Best Practices

TikTok is a one-of-a-kind platform that’s been shown to offer deeper connections to audiences, leading to increased brand awareness, loyalty and engagement. Many of the benefits come from the fact that TikTok is a platform known for authenticity. Because of that, brands and businesses see the biggest benefits when they have the courage to be themselves on the platform.

Still, brands have to be smart about how to structure, time and plan their campaigns to maximize benefits. Digital campaigns are never "set-it and forget it"; there's always room for improvement. A comprehensive “always-on" testing strategy is a great way to make sure you're always learning more about your audiences, your creative, and what kind of messaging resonates with your community.

Everything starts with knowing what’s really available in terms of testing. You can’t create a solid testing strategy without knowing what you can test and what you can measure. TikTok’s platform allows for quick and easy split testing for your campaigns.

What is split testing? Split testing, frequently called A/B testing, is a form of testing available through TikTok Ads Manager that allows advertisers to test the impact of different ad campaign elements by comparing the performance of two groups simultaneously.

TikTok Ads Manager allows you to test one variable in each split test. By limiting tests to a single variable, you’re able to measure the direct impact of each piece element of the ad campaign. Variables available for split testing on TikTok are:

> Targeting
> Creative
> Bidding & Optimization

For example, you can test different versions of ads to see how different ad creatives resonate with your target audience. At the same time, you could run another test that uses the same creative across different audiences to measure the impact of specific creative with different demographics or at different stages of the purchase funnel.

They key to an effective “always on” testing approach is to always be working to improve incrementally. It’s not about solving the problem, it’s about getting a little better each time. Your testing strategy should be designed to build on the lessons and learnings of previous tests.

Start by understanding which metrics will serve as your key performance indicators for your tests. Is your long-term goal to improve conversion? Expand your target audience? Drive engagement at the top of the purchase funnel? By knowing what you want to measure, you can start to work your way through the process of understanding how each variable could impact the performance of that KPI.

Successfully starting your testing program does involve some initial setup, including the managing of control groups, test balancing, budgets, and confirming your testing windows.

The careful management of control groups is always a critical part of any successful testing system. In addition to limiting the number of variables in a single split test, it’s also important to include a control group where variables have not been adjusted. This is to help identify, account for, and remove any influence of outside factors on performance.

Without a proper control group set up, you could quickly find yourself in a position where you are falsely attributing a performance shift to a variable change you assigned in the split test, when the shift was actually driven by seasonality or a difference in audience targeting, or anything that wasn’t your assigned variable, but was different between your comparison groups nonetheless.

This is also connected to the idea of balancing in a testing scenario. Every effort should be made to make the test and control groups as similar to each other as possible outside of the variables you’ve assigned for that test. Testing windows should be equal, potentially even aligning in terms of seasonality, days of the week, etc. Unless your audience demographics are the variable, audiences should be similar in terms of demographics, including age breakout, geographic region, gender or socio-economic status.

It’s important to have an idea for your target budget for a test before getting started. Daily spend and balancing budget across the testing and control groups will be critical for the test’s success. Plan your budgets to be large enough to make sure you get enough high-quality data to deliver results on the test, but small enough that you’re not pouring money unnecessarily into the group that loses the test.

Selecting the right testing window can make or break a split test. You want to make sure and select windows that are long enough in duration to reliably measure the impact of your variables. For most TikTok advertisers, that will be a minimum of 2 weeks, but could go as long as 6 weeks.

You also want to guarantee that time windows across your test and control groups align, as mentioned earlier. This includes looking at day of week alignment. Community behavior on TikTok can vary wildly between different days of the week, so a test group run over two weeks from Wednesday to Tuesday could deliver very different results from a test run from Sunday through Saturday, independent of the variables you’ve selected for your test.

The whole purpose of testing, and especially for an “always on” testing approach is to learn and improve over time. You’re not testing just to test—successful testing strategy is always with purpose. By carefully selecting your variables, testing windows, testing and control groups, you’ll put your business in a position to come away with insights that can lead to more efficient campaigns in the future.

Get started today by creating a TikTok For Business account,or dive deeper into the options and available metrics for split tests on TikTok Ads Manager.