Test & Learn: 5 steps to scaling performance on TikTok

As the home of entertainment and creativity, TikTok is a platform that's built to perform –61% of TikTok users have made a purchase either directly on TikTok or online after seeing an ad on the platform [1]. With TikTok, the opportunity to connect with a hyper-engaged audience and merge data-driven insights with creativity leads to performance marketing on overdrive.

If you're reading this blog post, you've probably already got the fundamentals of running a campaign on TikTok down. You've set your goals and found the campaign solutions to make them happen, and you're now starting to measure the results. But how do you use those results to keep growing?

It's time to test & learn! Having a test and learn agenda is an important performance lever to improve KPIs through which you can increase results or improve on your KPIs because it can help you understand what's working, or not, and how to make optimisation and investment decisions based on this.

Here are the five steps you can take to create a successful test & learn strategy:

1. Identify and prioritise variables to test

Create a hypothesis where your assumed variable will help you achieve your defined goals. If you think that there are a few variables that can drive better performance, prioritise those based on historical performance or your TikTok strategy.

For example, if you've struggled with understanding the best creative direction to go in, test this first. Make sure there's a distinctive difference between your variables – i.e. test different types of creative – and run no more than three tests simultaneously (unless you have an advanced measurement strategy in place).

2. Use Split Testing to reduce errors and overlaps

Split Test is a tool available on TikTok Ads Manager. It enables advertisers to test two different ad groups and identify the best-performing one, helping to optimally scale spend. It's the best way to run scientific A/B tests to determine which variables like creative, audience and optimisation work the best.

3. Stick to timelines and minimise adjustments

When running a split test with an ad group that requires a learning phase (e.g. Conversion or Product Sales), you should run the test for at least seven days (i.e. long enough to pass the learning phase and have a week of results at optimum algorithm performance). Even if an ad group doesn't require a learning phase, you should still run one test for at least two to three weeks before you can start to gain useful insights. Make sure to include a one-week reporting period post-campaign and to not make changes to your ad group after the test has launched.

4. Compare results against benchmarks

You need to have benchmarks to compare against to identify whether your chosen variable has a positive or negative impact on performance. Using the split test tool will automatically allow you to see the difference in performance and if there is an uplift. If you're conducting manual A/B testing, you can identify KPIs and compare the test results to your benchmarks after the test is over.

5. Scale the winning variable

Now it's time to evaluate your test results and adjust your strategy accordingly. As an example, you may want to understand which campaign objective drives the best volume of conversions and the best conversion rate, Web Conversion with View Content optimisation or Web Traffic optimised for LPV (Landing Page View). After running the tests, the results might show that web conversion produces the best results, so you can increase your ad spend on web conversion.

Pro tip: Use measurement tools such as Conversion Lift Studies to complement your attribution model for a comprehensive view of media mix efficacy.

Accelerate your performance today

Ready to get started? Head to TikTok Ads Manager and start scaling performance today. If you'd like to run more complex Test & Learn agendas, please speak to your account manager.