The way to prove what ROI Hunter changes on TikTok is an A/B test: two campaigns running side by side with TikTok's native Split Test, one your business as usual, one with ROI Hunter applied.
There are two designs worth running, depending on what you want to prove.
What to agree before you launch
Your main KPI, and which source you judge it by — TikTok Pixel or your analytics tool. They will not agree, so decide up front. See Understanding your TikTok numbers.
The market, and which design you are running.
Budget — enough for at least 50 conversions per side per week. Below that the result is not readable.
Two to three weeks, assuming that pace holds.
Launch both sides as new campaigns, so neither starts with delivery history the other does not have.
Option A — Test the creative: no template against template
Control: your usual creative, no templates applied.
Test: the same ads with ROI Hunter templates applied.
Both sides run Smart+ catalog ads with the same setup, so the creative is the only difference. Several templated ad variants on the test side are fine and encouraged — creative diversity is part of what you are testing.
Keep a creative test inside one ad group. You are testing whether templates should become your default across all your ads, so several templated ads in one ad group puts the largest share of the budget behind what you are actually testing.
Option B — Test the products: All Products plus a segment
Smart+ can run several product sets in one campaign, one per ad group. So:
Control: one ad group on All Products.
Test: one campaign with two ad groups — All Products, plus one on your segment.
Frame it as "plus", not "instead of". A segment ad group boosts a group of products; it does not replace your catalogue. All Products has to stay in, because Smart+ is broad targeting that does not exclude retargeting and All Products is the ad group carrying that delivery. A test designed as All Products against segments is the one that tends to fail.
The segment needs its own ad group because budget optimisation in Smart+ works at ad group level — that is the only way it gets its own budget and its own delivery history.
Let budget optimisation decide the split at first, then raise the segment ad group gradually. If it receives almost no spend, that is a signal to adjust the setup, not the result of your test.
Which segment to test with
New to TikTok, especially with a large catalogue — start from a segment already proven elsewhere. TikTok has little history to learn from yet, and products you know convert are the fastest way out of a low-spend, low-learning loop.
Already running TikTok at scale — build the segment on TikTok's own data: filter on the TikTok channel in Product Insights and save that as a segment.
Reusing a segment from another channel assumes you sell to broadly the same audience there. If your TikTok strategy targets a different audience with a different range, it may not transfer.