How to test UGC ad creatives: a practical method

What is creative testing?

Creative testing is running deliberately different versions of an ad against each other to learn which one performs better, and why. The aim is a repeatable lesson, such as "question hooks beat claim hooks for this product", not just a single winning ad.

UGC ads suit testing well because they are cheap to vary: the same script can open five ways, be filmed in two looks, or end with two offers.

What should you test first in a UGC ad?

After the hook, work down the ad in order of how much each part tends to move results. This order is a rule of thumb, not a law; skip a step when you already know the answer.

For hook ideas to test, see UGC ad hooks.

Why test one variable at a time?

If two versions differ in the hook, the actor and the offer, and one wins, you don't know which change did it. Changing one thing per test turns a result into a lesson you can reuse.

• Keep the audience, placement, budget and dates the same for every variant.
• Use the platform's split-testing tool where you can. TikTok's split test divides the audience into two equal groups so each person sees one version, and declares a winner at a 90% confidence level.
• Meta's A/B test works the same way for ad sets and campaigns.
• When you must test several things at once to save time, test whole concepts against each other, then go back and test parts of the winner.

Which metrics tell you a UGC ad is working?

Read the metrics in the order a viewer experiences the ad: did they stop, did they keep watching, did they click, did they buy. Platforms name and calculate these slightly differently, so use the same definition for every variant.

Attention metrics settle quickly and help you diagnose an ad; cost metrics decide which ad you keep. A hook with a high hook rate and a poor CPA usually means the opening promises something the rest of the ad doesn't deliver.

How much budget and time does a creative test need?

Give each variant enough impressions and conversions that the difference isn't luck. There is no single right number, but these rules of thumb are widely used:

• Run a test for three to seven days, so weekday and weekend behavior both show up.
• Judge hook rate after a few thousand impressions per variant; it settles faster than conversion metrics.
• Before judging CPA, let each variant spend at least one to two times your target CPA.
• On Meta, an ad set usually leaves the learning phase after about 50 optimization events in a week, so very small budgets give noisy results.
• Test two to five variants at once; more than that spreads the budget too thin to learn anything.

If the budget can't reach conversions, optimize the test for a cheaper event, such as clicks or add-to-cart, and confirm the winner on sales afterwards.

How do you run a UGC creative test?

1. Write a hypothesis

   Name the variable and the expected result: "An objection hook ('It's not another meal kit') will beat our feature hook on hook rate."

2. Make the variants

   Change only that variable. For a hook test, keep the script after the first line, the actor, the look and the captions identical.

3. Set up a fair test

   Same audience, placement, budget and schedule for each variant, ideally in the platform's split-testing tool.

4. Wait for enough data

   Don't turn variants off on day one. Wait for the impressions and spend you set out to collect.

5. Read and record

   Note the winner, the margin and the lesson in a testing log, then pick the next variable.

What mistakes ruin a creative test?

Most failed tests don't fail because the ads were bad; they fail because the test couldn't tell the ads apart.

• Calling a winner on day one, before either variant has enough impressions or spend.
• Changing the hook, actor and offer together, then not knowing which one mattered.
• Putting all variants in one ad set and letting delivery favor one early, so the others never get a fair share.
• Uneven budgets or schedules between variants.
• Testing differences too small to matter, such as one word in the caption, before the hook is settled.
• Not writing results down, so the same test gets run again in three months.

What is creative fatigue, and how often should you refresh?

Creative fatigue is when an audience has seen the same ad so often that it stops responding, and cost per result rises. Meta's Ads Manager can flag it on an ad set as creative fatigue.

• Signs of fatigue: frequency climbing, hook rate and CTR sliding, CPA rising week over week on a steady budget.
• First fix: give the winning ad new hooks. The body that converted usually still works.
• Next fix: a new look, actor or angle on the same message.
• Last resort: a new concept, tested from the hook up.

How do you make hook variants with Probably?

Probably is an AI video studio that makes UGC-style ads, and its variants are built for hook testing. Ask the crew for alternate hooks in plain words, and each one is made as its own version of the same video, so the script, actor, look and captions after the hook stay the same.

• Ask for a different kind of hook, such as a question, an objection or a result, or write the line yourself in the script editor.
• Keep the actor fixed across tests with a persona in your Space, so a hook test stays a hook test.
• Try a different look, such as selfie vlog against POV hands, when you move on to testing format; see UGC styles.
• Earlier versions are kept, and the studio shows a credit estimate before a paid change runs. Changing the captions on a delivered video re-renders it free.

Download the variants and upload them to your ad account; see review and revisions for how changes work.