Guide · Meta and TikTok

How to test UGC ads on Meta and TikTok: a simple A/B test plan

Published 2026-11-10

You ordered two UGC videos. Now you need to know which one to put your budget behind. Turning ads on and off and comparing them by eye feels like testing, but Meta’s own help center advises against it. Both Meta and TikTok have a built-in tool that splits your audience so the comparison is fair. Here is how each one works, what to test first, and a simple plan you can copy. We checked both platforms’ help pages in November 2026.

The short answer

Use the built-in tool. That means Meta’s A/B test or TikTok’s split test. Both make sure nobody sees both versions.
Change one thing. Compare two different videos, or two hooks on the same video, but not both at once.
Run it for at least 7 days. Both platforms recommend a minimum of 7 days and a budget large enough to get real results.
Don’t edit it mid-test. Changes can reset delivery and spoil the comparison.

Why “just turn them on and compare” doesn’t work

Meta’s About A/B testing page says it does not recommend testing informally, such as by turning ad sets or campaigns on and off manually. It warns that this can lead to inefficient delivery, unreliable results and overlapping audiences.

The problem is simple. If both ads run at the same time without a split, they can reach the same people. If they run at different times, the week itself can change the result. A built-in test fixes both: it splits the audience in two and shows each group only one version.

What to test first

For UGC, we suggest starting with the video itself rather than small settings. TikTok’s split test best practices recommend making the two versions differ significantly, so the system can pick a winner. A sensible order:

Two different videos. Different creators, or the same creator using a different angle, such as problem-and-solution versus an unboxing. Our script template describes five formats.
Two hooks on the winning video. Same body, a different first few seconds. This needs alternative hooks or raw footage from the creator.
Length or call to action. For example, a 15-second cutdown versus the full 30 seconds.

Test one of these at a time. Meta’s best practices page says results are more conclusive when the ad sets are identical except for the variable being tested.

How Meta A/B tests work

Setup: duplicate an existing campaign, ad set or ad and change one variable, or compare two existing campaigns or ad sets. You can start from Ads Manager or the Experiments tool.
Audience: each version goes to a separate segment, and Meta ensures nobody sees both.
Budget: Meta recommends the same budget for both versions.
Length: at least 7 days is recommended, and a test can run for up to 30 days. If customers usually take longer than a week to buy, Meta suggests running longer, for example 10 days.
Audience overlap: don’t use the test audience in any other campaign at the same time.
Winner: Meta’s winner page says it compares cost per result, such as cost per purchase. It simulates the outcome tens of thousands of times and reports a confidence percentage.

How TikTok split tests work

Setup: when you create a campaign in TikTok Ads Manager, choose an objective, then turn on “Create a split test.” Creative is one of the variables you can test. TikTok’s setup guide says the key metric you select decides the winner.
Audience: TikTok’s split testing page says it splits the audience into two equal groups, and each group sees only one ad group.
Length: at least 7 days is recommended. A split test can run for up to 60 days.
Budget: TikTok shows an “Estimated Testing Power” for your budget and recommends a budget that gives a power value of at least 80%.
Winner: TikTok says the test uses a 90% confidence rate and picks a winner only if the results are statistically significant.

Meta vs TikTok at a glance

	Meta A/B test	TikTok split test
Audience split	Separate segments, nobody sees both	Two equal groups, each sees one ad group
Recommended minimum	7 days	7 days
Maximum length	30 days	60 days
Budget guidance	Same budget for both, enough to get results	Aim for at least 80% estimated power
How the winner is picked	Lower cost per result, with a confidence percentage	90% confidence, only if significant

Sources: Meta Business Help Center (About A/B testing, A/B test best practices, Determining a winner, About the learning phase) and TikTok Ads Manager help (About Split Testing, January 2026; Split Test Best Practices, November 2026). Checked November 10, 2026.

How much budget does a test need?

Neither platform gives a fixed dollar amount, because it depends on what a result costs you. One useful sense check comes from Meta’s learning phase page. It says ad sets usually leave the learning phase after about 50 results in the week after the last significant edit.

That gives a rough worksheet. It is not a rule for when a test is finished:

Test budget worksheet

Your usual cost per result: [e.g. US$8 per add-to-cart]
About 50 results per version per week: 50 × [US$8] = [US$400] per version
Two versions for 7 days: [US$800] in total

If that is more than you can spend, pick a result that happens more often, such as landing page views instead of purchases. Just remember that the cheaper result may not predict sales. In that case, treat the winner as a lead to confirm, not a final answer.

What it costs to make the test videos

In our November 10, 2026 sample of 30 Fiverr UGC gigs, the median basic package cost US$95 before Fiverr’s buyer fees. So a two-creator test starts at roughly US$190 for the videos, plus ad spend. See what UGC videos cost for the full price breakdown.

For a hook test, ask one creator for alternative openings when you order. In our earlier check, only 6 of 30 gigs listed additional hooks, so it usually needs a custom offer.

Copy-paste test plan

UGC ad test plan

Question: does [video A / hook A] get a lower cost per [result] than [video B / hook B]?
The one thing that changes: [the video / the first 3 seconds / the length].
Everything else stays the same: audience, placement, budget, caption, landing page.
Tool: [Meta A/B test / TikTok split test].
Length: [7–14] days. No edits while it runs.
Budget: [amount] per version, from the worksheet above.
Next step: keep the winner running, then test a new hook against it.

Reading the result

Look at cost per result, not likes. Engagement can rise while sales don’t.
Check the confidence. A low confidence percentage means the two versions may simply be close. Meta notes that a “winner” can still have higher costs if the test was too short or had too few results.
Learn why it won. Write down what was different, such as the opening line or the format, and brief your next creator with it in our brief template.

Drafting hooks before you pay for them

Every test costs money for both videos and ads, so some brands narrow the options first. They draft hook ideas with an AI video tool, then pay a real creator to film the strongest angles for the actual test. An AI actor has never used your product, so its video must not be presented as a customer’s experience, and platforms may require an AI label. Our AI UGC vs real creators guide covers where that line sits.

Affiliate link

The link below is an affiliate link. If you sign up and pay through it, we may earn a commission at no extra cost to you. Plans use monthly credits, so check how many videos your plan covers before you commit.

Try MakeUGC (affiliate link)

Common mistakes

Changing two things at once. If the creator and the hook both change, you won’t know which one mattered.
Stopping after two days. Both platforms say tests shorter than 7 days may be inconclusive.
Editing during the test. It can reset learning and muddy the result.
Testing two near-identical videos. TikTok recommends obvious differences between the versions.
Running the same audience in another campaign. Meta warns that overlap can contaminate results.

Common questions

How do I A/B test UGC ads?

Use Meta’s A/B test or TikTok’s split test, change only one thing, such as the video or the hook, give both versions the same budget, and run it for at least 7 days without edits.

How long should a UGC ad test run?

At least 7 days on both platforms. Meta tests can run for up to 30 days and TikTok split tests for up to 60. Run longer if your customers usually take more than a week to buy.

How many UGC videos do I need to start testing?

Two is enough for one test. Make them clearly different, such as two creators or two formats. Once you have a winner, test new hooks against it.

Which metric decides the winner?

Cost per result, such as cost per purchase. Meta picks its winner on cost per result, and TikTok uses the key metric you select when you set up the test.

Ready to make your test videos? Check what UGC videos cost, plan with our script template, or browse the creator shortlist.