Talking head video: setup, scripting and editing for 9:16 ads

What a talking head video is, and an example

The shot itself is simple: a person's head and shoulders, talking to camera. On TikTok, Reels and Shorts it is the default creator format, because a face looking into the lens holds attention cheaply and the creator's voice carries the argument. In ads, the same shot is the backbone of most UGC: a creator telling you why they switched, what surprised them or what they wish they had known.

Illustrative example (not a real ad): a 25-second vertical video. 0-2 s, the creator says "I stopped buying protein bars, here's why" over a close frame. 2-15 s, three short reasons, each a separate jump-cut line, with a cut-in of the product on the second reason. 15-22 s, the alternative they use now, held up in frame. 22-25 s, a call to action spoken and shown as text. No set, no music bed for the first seconds, captions throughout.

TikTok's creative best-practices page (June 2025) lists the ingredients this format already has: feature people such as creators or customers, use a less polished style that fits user-generated content, and put the hook in the first 6 seconds with the proposition in the first 3.

Talking head video setup: light

Light the face, not the room. The standard starting point is three-point lighting. Riverside's lighting guide (updated November 2025) describes it on a clock face with the camera at 6 o'clock: a key light, the brightest source, at about 4 o'clock; a fill light at about 8 o'clock at roughly half the key's intensity to soften shadows; and a back light behind the subject, about as bright as the fill or a little less, to separate them from the background. Keep the fill well below the key, or the face goes flat.

For a phone setup, a window can be the key light. Face it at an angle, not straight on, and use a white wall, card or a small LED panel as fill. Avoid overhead room lights as the main source; they put shadows under the eyes. Whatever you use, set exposure and white balance once and lock them, so the picture does not shift between takes.

Talking head video setup: audio

Treat audio as seriously as picture: both TikTok's best-practices page and Meta's Reels ad guide push for sound on. Use a clip-on lavalier or a small shotgun mic close to the speaker. StreamYard's lavalier guide recommends mid-chest placement, pointing up toward the mouth with a clear path to it, about eight inches from the chin to cut breathing and popping noise. Clip it to the outer layer of clothing and keep it from rubbing on fabric; too high can change the tone, too low drops the level.

Record in a small room with soft furnishings, turn off fans and fridges you can hear, and record ten seconds of silence at the start so the editor has room tone to fill gaps between jump cuts.

Framing for 9:16 safe zones

Shoot vertical at 1080x1920. Put the eyes about a third of the way down the frame, with the face in the upper-middle. The bottom of a vertical feed is covered by the caption, the account name and the call-to-action button, and the right edge by like, comment and share icons.

The platforms describe those areas differently. Meta's Instagram Reels ad guide suggests leaving about 14% of the top, 35% of the bottom and 6% of each side free of text, logos and other key elements. TikTok does not publish one margin; its In-Feed spec page (June 2026) says the safe zone depends on ad dimension, caption length and add-ons, and provides downloadable safe zone files to overlay in your editor. Use the platform's file before export. Our TikTok ad specs and YouTube Shorts ad specs posts cover the full specs.

Frame from mid-chest up, leave a little headroom, and keep the background simple and a few feet behind the speaker. If the product appears, it should come up to chest height, not sit on a table at the bottom of the frame where the caption will cover it.

Scripting a talking head ad

Write for the ear. Short sentences, one idea each, in the words a person would actually say. A structure that fits most talking head ads:

1. 01

   Hook (0-3 s): a claim, a question or a confession that names the viewer's problem. Pull a starting point from the UGC hooks library.

2. 02

   Context (3-8 s): why the speaker is worth listening to on this, in one line.

3. 03

   Proof (8-20 s): two or three specific reasons, each its own line so it can be its own cut.

4. 04

   Offer and call to action (last 3-5 s): what to do now, said and shown on screen.

Talking head video editing patterns

Jump cuts are the house style. Record each line two or three times, keep the best, and cut out every breath and pause between lines. Cut on the first frame of speech, not before it. A slight punch-in (scaling the frame up) on alternate lines hides the jump and adds energy without new footage.

Captions are not optional for silent viewers. Meta's Reels guide marks captions as recommended, and TikTok suggests on-screen text at about 5-10 words per second, which is a good ceiling for caption speed. Show two to five words at a time, centered in the safe area, and highlight the key word.

Use inserts sparingly. One cut-in to the product, a screen recording or a proof point per 5-10 seconds keeps the face as the anchor. TikTok's best-practices page also recommends transitions, stickers or graphics to hold attention; use them to mark a new point, not to decorate.

End on the call to action, then stop. Do not fade out on the speaker smiling for two seconds.

When talking head works in ads, and when it does not

It works when the argument is the product: a new point of view, a clear offer, a myth to correct, a before-and-after story, or a service or app with nothing physical to show. It is weaker when the product has to be seen working, such as texture, fit, size or a physical result. There, cut in product footage, or switch formats to product in hand or an unboxing (see our unboxing video ideas).

Test the opening, not just the whole ad. Run two hooks on the same body and compare 3-second view rates with the hook rate calculator. Put the angle, the claim and the shot list in a UGC brief template, and draft the lines with a UGC script template.

If you do not want to film at all, FireplaceUGC's Talking Video mode turns a script into a 9:16 talking take with one of 500+ AI actors, which you own and upload yourself. Script iteration is free, and the AI actor must never be presented as a real customer; see our AI ad disclosure rules. The SaaS and supplements pages show where message-led ads fit.