TikTok Hooks for UGC Ads: Templates and a Priced Test

By Influverse Editorial Desk

Published 23 October 2026. Last updated 23 October 2026. How this was written and checked

A TikTok hook is the opening of a video: the first picture, the first spoken line and the first on-screen text a viewer meets before deciding whether to keep watching. In a UGC ad the hook has one more job, because the viewer did not ask to see an ad, so the opening has to make a promise about their problem before it shows the product. Below are hook types with templates for product ads, how to make the spoken line and the text carry the same promise, and three ways to test three openings on one ad, each priced before you render anything.

If you already have an ad script and need only the first line, start at the hook types. If you have three openings and want to find out which one holds viewers, go straight to the test.

Key takeaways

  • Write every hook in three layers that make one promise: the first frame, the spoken line and the on-screen text.
  • Pick the hook type from the product: a visible result opens on the result, a familiar frustration opens on the problem, a crowded category opens on a comparison.
  • Give an AI creator lines about the problem, the viewer or what the ad shows, never a story of using the product that nobody lived.
  • Test the on-screen text first, on a take you already have, before you render a new spoken line.

What a TikTok hook is made of

A hook is not one sentence. In the opening of a TikTok a viewer takes in three things at once, and each of them can hold the viewer or lose them:

  • The first frame. What is on screen before anyone speaks: the creator mid-gesture, the product in a hand, the result already visible. Do not spend it on an empty or static shot.
  • The spoken line. The first sentence the creator says. It states the promise in words and sets the voice of the whole ad.
  • The on-screen text. A short line laid over the video for viewers watching without sound. It restates the promise in fewer words instead of transcribing the speech.
Diagram of a UGC ad: the opening holds three stacked layers that make one promise (the first frame, the spoken line and the on-screen text), followed by the body, a product demonstration and a call to action that stay the same in every hook version.

The three layers fail together when they disagree. A spoken question, on-screen text about a discount and a first frame of the creator smiling make three different promises, and the viewer gets none of them. Write the promise first in plain words, then make each layer carry it. For the rest of the ad the hook sits in front of, from the creator to the product to where it runs, see how AI UGC ads are put together.

Hook types for UGC ads, with examples

Each type below has a template and an example for one product, an insulated lunch box, so you can see the same product opened several ways. Swap the bracketed words for your own product, problem and result, and keep only promises the rest of your ad shows on screen.

  • Problem first. Template: "If [problem] keeps happening, [common fix] is not the answer." Example: "If your lunch is cold by noon, a bigger bag is not the answer."
  • Result first. Template: "This is [result], and here is what made it happen." Example: "This is a hot lunch at my desk, and here is the box it came out of."
  • The callout. Template: "If you [specific situation], this one is for you." Example: "If you eat lunch at your desk every day, this one is for you."
  • The objection. Template: "Everyone says [objection]. Here is the part they skip." Example: "Everyone says insulated boxes are bulky. Here is the part they skip."
  • The comparison. Template: "[Option A] or [option B]: here is the difference that matters." Example: "Glass container or insulated box: here is the difference at lunchtime."
  • The mistake. Template: "Stop [common habit] if you want [result]." Example: "Stop packing soup in a plastic tub if you want it warm at lunch."
  • The demo in progress. Template: "Watch what happens when I [action]." Example: "Watch what happens when I open this at my desk."

Match the type to the product rather than to a list of what went viral. A product with a result you can film opens on the result or the demo; a product that solves a familiar annoyance opens on the problem or the mistake; a product in a crowded category opens on the comparison or the objection. Whichever you pick, name the product only after the promise has landed, so the opening gives the viewer a reason to stay before it reads as an ad.

Three openings for one ad, spoken and on screen

Here are three openings for the lunch box written as pairs, so you can see how the on-screen text carries the promise in fewer words than the spoken line:

  • Problem first. Spoken: "If your lunch is cold by noon, a bigger bag is not the answer." On screen: "Cold lunch by noon?"
  • Demo in progress. Spoken: "Watch what happens when I open this at my desk." On screen: "Opening it at my desk"
  • The comparison. Spoken: "Glass container or insulated box: here is the difference at lunchtime." On screen: "Glass or insulated?"

Spoken hooks versus on-screen text hooks

The spoken line and the on-screen text do different work, so plan them differently:

  • The spoken line sets the tone and the voice. Settle it before you render, and plan a new take for any change to it.
  • The on-screen text carries the promise for viewers watching without sound. Add it in your editor, on top of the finished take.

That difference sets the order of the test further down. For the spoken line itself, the model you render on matters.

In Influverse, a character speaks on camera in its own voice on Wan 3.0 Prime and Seedance. On any other model, check how a voiced line will play before you build a hook around the creator's mouth.

Keep the on-screen text shorter than the spoken line: a few words a viewer can read in the time the line takes to say. Then write the caption, the third piece of text, which sits below the video and should add something the opening did not say. If you want a first draft to work from, try the TikTok caption generator.

Writing hooks for an AI creator

A filmed creator who says they used a product for a month either did or did not; an AI creator never did, so a first-person story of use is a testimonial nobody gave. Write the hook about the problem, the viewer or the result the ad shows, and keep the creator's first person for opinions and demonstrations the viewer can watch happen.

  • One short sentence per opening. Write a line the creator can say in one breath. There is more on keeping a spoken line natural in a short clip.
  • Words this character would say. Match the voice you gave the character: a calm reviewer does not open with a shout.
  • No figure you cannot show. A hook that names a time, a price or a count has to prove it inside the ad.
  • Labels where they are asked for. Follow the rules of the platform you post on for labelling ads and AI-generated content.

If you are a freelance creator making sample ads for brands with an AI character, the same rules apply to your portfolio, and the guide to producing UGC samples with an AI character covers how to build one.

Testing three hooks on the same ad

Once you have three openings, the question is how to compare them without paying to make the whole ad three times. Split the ad in two: the opening, which changes between versions, and the body, the demonstration and the call to action, which stays the same. Then choose the method by what your three openings change, and start with method 3 when they differ only in their on-screen text.

Every figure below names its model, resolution and length, so you can check it against the credit table. Here are the three methods side by side:

MethodModel and resolutionTakes you renderCredits
Text onlyWan 3.0 Prime at 720pOne 10-second take54 credits
One body, three openingsWan 3.0 Prime at 720pOne 5-second body and three 5-second openings108 credits
One body, three openingsWan 3.0 Prime at 720pOne 10-second body and three 2-second openings87 credits
Three full takesWan 3.0 Prime at 720pThree 10-second takes162 credits

On Wan 3.0 Prime at 720p, a 10-second clip costs 54 credits, a 5-second clip costs 27 credits and a 2-second clip costs 11 credits.

Method 1: three full takes

Render the whole ad three times, each time with a different opening. Each version comes out of the model as one finished clip.

Three full 10-second takes on Wan 3.0 Prime at 720p cost 162 credits, three takes at 54 credits each, which is $8.10 at $0.05 a credit.

Method 2: one body, three openings

Render the body once and each opening as its own short take, then place each opening in front of the same body in your editor. Give each opening take a start image of the creator and the first frame of the body as its end frame.

Wan 3.0 Prime takes an end frame as well as a start frame.

One 5-second body take plus three 5-second opening takes on Wan 3.0 Prime at 720p cost 108 credits, the body and the three openings at 27 credits each, which is $5.40.

One 10-second body take plus three 2-second opening takes on Wan 3.0 Prime at 720p cost 87 credits, 54 credits for the body and 11 credits for each opening.

Add the two lengths to see how long each finished ad runs, and pick the split that keeps the ad as long as you planned. Save the shorter split for your shortest lines, since few spoken words fit in it.

Method 2 stops working when an opening runs into the body, such as a question whose answer is the first sentence of the body. Change the question and the answer has to change with it, so that version goes back to method 1.

Method 3: text only

For a text-only test, use a take you already have, or render one 10-second take on Wan 3.0 Prime at 720p, which costs 54 credits. Then add each of the three on-screen texts to a copy of that take in your editor.

Only the text changes, so this test tells you nothing about the first frame or the spoken line. Run it first anyway: if the text alone changes how many viewers stay past the opening, you have learned something before paying for a new line. For any other model or length, see every video model with its credits.

Posting the versions

Post the versions with the same caption and the same sound, either on one day a few hours apart or on separate days at the same hour, and compare how many viewers stay past the opening rather than total views: that count is this guide's measure of a hook, because total views also reflect everything after the opening. Posting on different days mixes the day into the result, so treat a small gap between versions as noise and run the winner against a fresh opening before you settle on it.

Influverse schedules posts to TikTok on the Pro, Max and Enterprise plans and during the 7-day trial. Scheduling a post spends no credits.

To render these takes, pick Wan 3.0 Prime and 720p in the AI video generator.

Common hook mistakes

  • Opening on a logo or a greeting. The first frame is the first layer of the hook, so spend it on the problem or the result.
  • Promising what the ad never shows. The call to action then follows a promise the ad did not keep.
  • Changing several things between versions. Change only the opening, or the test cannot tell you which change mattered.
  • Writing the on-screen text as a transcript of the speech. Restate the promise in fewer words.
  • Giving an AI creator a first-person story of using the product. It is a testimonial nobody gave.

Render your three openings in the AI video generator. Start your free 7-day trial with 120 credits. $0 today, cancel anytime. Not ready? Sign up free and get 20 credits, no card needed.

Frequently asked questions

What is a TikTok hook?

A TikTok hook is the opening of a video: the first picture, the first spoken line and the first on-screen text, which together give a viewer a reason to keep watching.

What makes a good hook for a UGC ad?

One promise about the viewer's problem or the result, carried by the first frame, the spoken line and the on-screen text together, and delivered by the rest of the ad.

Should a TikTok hook be spoken or written on screen?

Both, making the same promise. The spoken line sets the voice for viewers with the sound on, and the on-screen text carries the promise for viewers watching without sound.

How do I test TikTok hooks on the same ad?

Change only the opening. Start with the on-screen text on one take you already have, then render each spoken opening as its own short take in front of one body take, and render full versions only for an opening that runs into the body.