Veo 3 for ads: a practical guide

What Veo 3 does well for ad work

Veo is Google's video model, and for ad work its main strength is native audio, including dialogue written into the prompt. Google describes it as "a model for generating 8-second videos (720p, 1080p, or 4k) with natively generated audio" (Gemini API docs).

A note on names. Searches still say "Veo 3", but on the Gemini API Veo 3 and Veo 3 Fast are marked "(Deprecated)", and the current models are Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite. Google's video overview also recommends a newer model, Gemini Omni Flash, "as your default model for video generation", and Veo 3.1 when you need "scene extension, last-frame control, or integration with legacy pipelines" (Gemini API docs). The facts below are for Veo 3.1 on the Gemini API.

For ads, three of these shape everything: the 8-second clip, the two aspect ratios, and English as the only fully supported language. If the ad is not in English, test the dialogue carefully or record the voice separately.

Which ad formats Veo suits

Veo suits short dialogue-led formats in English and cinematic shots in 16:9. The table is guidance from the published capabilities, not test results.

Formats with long unbroken monologues fit Veo less well, because each line has to fit inside 8 seconds or be carried across an extension.

Workflow: from script to finished Veo ad

Plan the ad as a sequence of 8-second shots, generate each with its own line of dialogue, then join them in an editor. A 24-second ad is three shots; a 40-second ad is five.

1. Cut the script into 8-second lines. Each shot gets one short spoken sentence and a few seconds of reaction or product. If a line will not fit, split it across two shots with a cut between them.
2. Make reference images. One presenter image and one product image, used on every shot (up to 3 per generation). Reference images require 8-second duration, so set that as your default.
3. Choose the ratio once. Pick 9:16 for Reels and TikTok or 16:9 for YouTube and brand cuts. Veo has no 1:1 or 4:5, so crop from 9:16 in the editor if you need a feed version.
4. Generate shot by shot. Put the dialogue in the prompt with the reference images attached. Generate a few takes and choose on mouth sync and pronunciation of the product name.
5. Match the cuts. Where two shots must join smoothly, export the last frame of shot one and use it as the first frame of shot two with frame-specific generation.
6. Use extension for continuous shots. For a brand-film walk or a slow product reveal, extend one clip by 7 seconds at a time. Extension works at 720p only, so upscale or accept 720p for those shots.
7. Edit and finish. Add burned-in captions, music, the end card, and any price or offer as editor text.

For a hook test, regenerate only the first shot with a new line and keep shots two onward the same. That isolates the hook as the variable.

Veo 3 prompt structure for ads

A Veo ad prompt should read like a one-shot script: shot type, subject, setting, action, the spoken line, then sound. Keep to one speaker and one idea per 8-second clip.

The templates below are written for this guide. Square-bracket items are placeholders.

UGC selfie hook (9:16, 8 seconds)

Vertical handheld selfie shot, front phone camera, arm's length.
[presenter from reference image], [age range], in [location].
She stops walking, looks into the lens and says:
"[hook line, under 15 words]"
Then she raises [product from reference image] into frame.
Audio: her voice, light street ambience, no music.

Product hero shot (16:9, first and last frame)

Studio product shot of [product from reference image] on [surface].
First frame: product in shadow. Last frame: product fully lit, label readable.
A light sweeps across from left to right as the camera slowly pushes in.
[one physical detail, e.g. droplets on glass].
Audio: low hum rising to a soft chime. No dialogue.

Podcast single (one host per shot, cut between two)

Medium close-up of [host A from reference image] at a studio microphone,
over-ear headphones, [studio set with lamp and bookshelf] behind.
He leans toward the mic and says:
"[one sentence of the conversation]"
Static camera, shallow depth of field.
Audio: close mic voice, slight room tone.

Repeat the same prompt with host B's reference and the next line, then cut the two together as a two-camera edit.

Animated explainer scene

Bright 3D animated scene, [character from reference image] in [setting].
The character holds up [object showing the problem], frowns, then swaps it for [product].
Camera: slow dolly in to a medium shot.
Character says: "[one-line benefit]"
Audio: character voice, light cartoon sound effects.

Brand-film establishing shot (16:9, 4k)

Wide cinematic establishing shot of [location] at [time of day].
[weather and light], [one moving element crossing the frame].
Slow crane up revealing [landmark or product placement in scene].
Audio: [ambient sound bed], distant [single detail sound]. No dialogue.

Limits, terms and policy notes

Veo's main limits for ads are the 8-second clip, two aspect ratios, and English as the only fully evaluated language. Check terms and pricing for the access route you use.

Commercial use. The Gemini API terms say: "Google won't claim ownership over that content. You acknowledge that Google may generate the same or similar content for others and that we reserve all rights to do so." They also say the Gemini API is "for developers building with Google AI models for professional or business purposes, not for consumer use", and users must be 18 or older (Gemini API terms).

Watermark. Every Veo video carries a SynthID watermark (Gemini API docs). Treat every Veo shot as AI-generated when you set platform labels.

Pricing, as published (Gemini API, paid tier, per second, USD).

Free tier: "Not available". "You will only be charged if your video is successfully generated" (Gemini API pricing).

Google's consumer plans, as shown on the US page: Free $0 a month; Google AI Plus $4.99 a month; Google AI Pro $19.99 a month, which includes "a limited trial of Veo 3.1 Lite"; Google AI Ultra "Starting at: $99.99 / month", with Ultra listing "video generation with Veo 3.1" (Gemini subscriptions). Prices may differ outside the US.

Platform labels. Meta labels ad images "created or edited using third-party AI tools with AI info" (Meta). TikTok allows AI-generated ad content only with the AIGC label or your own clear disclaimer, and says undisclosed AI content means "your ad will be rejected or restricted" (TikTok ads policy). See AI-generated content labels on Meta and TikTok ads.

Synthetic presenters. A Veo presenter is not a real customer. Do not write lines that present them as one ("I've used this for a year"), and keep health or money claims out unless the brand has approved them.

Veo in a one-script, every-format workflow

Veo fits the one-script method because the script is already cut into 8-second lines. The same lines can be regenerated as a selfie, a podcast single, a street-interview answer or an animated character with only the scene description changed.

Keep the dialogue identical across formats and change the setting, framing and reference images. The full method is in one winning script, made in every ad format, and the UGC script template shows how to write beats that fit short clips.

Anglr pays AI creators, editors and agencies on performance to remake winning ad scripts in every format. Anglr provides the scripts and angles, and every video gets a QC and compliance review before it goes live. If you make full ads with Veo and know they perform, see how it works for creators.