1. Breaking down creative into three elements
Before we begin, let's break down the TikTok video into three parts: By examining these separately, we can pinpoint what made a video "successful."
- Hook (first 3 seconds): A mechanism to stop scrolling. If the user leaves here, the content that follows becomes meaningless
- Body (contents and benefits): Presents the value of the product and solutions to problems. This is the part that builds trust.
- CTA (call to action): The final push. Give clear instructions such as "Check it out now" or "Learn more."
2. Properly Designing A/B Tests
When performing validation, the golden rule is to only change one variable at a time.
Common validation patterns
- Pattern A: Hook (live action) + same body + same CTA
- Pattern B: Hook (animation) + Same body + Same CTA
In this case, we can conclude that the difference in results is purely due to the "hook format." If you change the body as well, it becomes difficult to determine which factor caused the change in the numbers.
3. Three key metrics to watch
When looking at your Ads Manager numbers, keep these metrics in mind that are specific to TikTok ads:
Hook Rate
25% or more
Number of 2-second/3-second plays ÷ Number of impressions
Click Rate (CTR)
1.5% or more
An index to measure the strength of induction
CV Rate (CVR)
Target Dependency
Final Acquisition Efficiency
Best Practices
If the hook rate is high but the conversion rate is low, it is likely that you are baiting the user too much at the beginning (mismatch in expectations) or that there is a problem with the content of the landing page.
4. Material "wear" and update frequency
TikTok users consume content very quickly, so if you keep posting the same content, it will quickly become unresponsive (this is called material wear).
The key to maintaining a stable CPA is to constantly provide the algorithm with “new signals” by introducing at least 2-3 new hooks and patterns per week.
Mirai accurately determines the winner
Running creative experiments without accurate data is like running blindfolded.
- Calculating true CVR for each creative even in iOS14 environment
- Minimizing data delays and supporting early decision-making on whether to continue or withdraw
- Server-side integration enables material evaluation, including offline contracts
5. The "Golden Cycle" of Creative Validation
- Hypothesis formulation: Created three hook ideas by working backwards from the target audience's concerns.
- Test Run: Distribute for 3-5 days with a low budget and check the hook rate.
- Winners Analysis: Verbalize the hook elements that generated the best numbers.
- Horizontal expansion: Using the hook pattern that worked, change the body contents (benefit) and retest.