How to Make AI Video Ads: Tools, Prompt Templates and Costs (2026)

A few years ago, a 15-second ad meant a film crew, a studio, editing and several weeks of work. Today you can produce AI video ads on the same day with a few well-written sentences and the right tools. For SMEs with limited ad budgets this is a real opportunity: you can test different messages at once, refresh videos for each season and create a separate creative for every product.

On the other hand, there are a lot of tools, the names are confusing and it is not clear what each one costs. In this article we look at how to make a video with AI from a business perspective: what each tool is good for, how far the free versions go, how to write prompts for ads, what concrete examples look like for furniture and hotels, and which risks to watch out for.

In the final section we cover something most businesses skip: even the best video will not bring in sales if the campaign is poorly targeted or the landing page loads slowly.

Text-to-video and image-to-video: the two core methods

Almost every AI video tool works in one of two ways:

• Text-to-video: You describe the scene, lighting, camera movement and mood in words, and the AI generates a clip from scratch. This suits brand stories, mood videos or abstract campaign ideas.
• Image-to-video: You upload an existing product photo and the tool adds motion, a camera pan or ambient effects. When the product must look exactly right (furniture, packaged goods, a hotel room), this is usually the safer route.

The most common mistake in advertising is trying to generate a product video entirely from text. The AI cannot know your sofa's fabric, its stitching or your hotel's real pool; it will produce something similar but different. That creates a risk of misleading customers. The rule is simple: if the product itself will be visible, start from your own photo, and use text-to-video for mood and transition scenes.

Which tools can you use for AI video ads?

Google Veo 3 and Gemini

Veo is Google's video generation model. It is available in the Gemini app with the relevant subscription plans and produces short clips (typically around 8 seconds with Veo 3). Its standout feature is that it can generate sound effects, ambient audio and dialogue together with the visuals.

How do you use Google Veo 3? In Gemini, you select the video creation option, describe the scene in a detailed prompt and, if you like, add a starting image. The clip is ready within a few minutes; if you do not like it, you refine the prompt and generate again. The number of generations is capped by daily or monthly limits depending on your plan.

Veo and Asset Studio inside Google Ads

Google is also bringing AI video generation directly into the ad platform. Asset Studio in Google Ads offers features such as generating images, replacing backgrounds and creating short videos from product images, and Veo-based video generation is gradually being rolled out to accounts. The advantage is that the videos you create can be connected straight to your Performance Max or YouTube campaigns. Which features are enabled in your account can vary by region and account type. If you want to understand the campaign setup on Google's side, see our Google Ads guide.

Meta Advantage+ creative tools

For Facebook and Instagram ads, Meta's Advantage+ creative features can add motion to your images, build short videos from several images, generate text variations and adapt formats automatically. There is no separate fee; it comes built into Ads Manager. When preparing Instagram video ads, this is often the most practical starting point, because the output is already shaped to the platform's formats. For the campaign side, our article on how to run Instagram ads may help.

CapCut

Think of CapCut as an AI-assisted editing tool. It offers automatic subtitles, text-to-speech voiceovers, background removal, templates and video creation from a product link. CapCut is usually the final stop where you combine clips made in Veo or Firefly and add subtitles, your logo and music.

Adobe Firefly

Firefly is Adobe's family of generative AI tools and offers both text-to-video and image-to-video. Adobe stresses that Firefly models are trained on licensed content that should not raise copyright issues and are designed for commercial use. For businesses that care about brand safety, this is a meaningful advantage. It works on credits: each generation uses a certain number of credits.

Free vs paid: what the tools cost

The table below is only meant to give a general idea. Prices change often with exchange rates, promotions, region and plan updates. The ranges are not exact prices; they are observations from the Turkish market in TL, so check the current price on each tool's own website for your country.

Step by step: how to make a video with AI

1. Pick one message. For example, “New season sofa set, free assembly” or “Direct discount for early booking”. Do not squeeze three messages into one video.
2. Decide on the format. Vertical 9:16 for Reels, Stories and Shorts; horizontal 16:9 for YouTube and the web. Producing both formats separately gives better results than cropping.
3. Prepare your visual material. Clear, well-lit product photos, your logo and your brand colours.
4. Generate the clips. Image-to-video for product scenes, text-to-video for mood. Create 3-4 variations of each scene and pick the best.
5. Edit. Combine the clips in a tool such as CapCut and add subtitles, your logo and a short call to action (“Visit our showroom”, “Book now”).
6. Check everything. Does the product look true to life? Are there odd hand or face distortions? Is the text readable? Does it make sense without sound?
7. Launch with at least two versions. Run an A/B test with a different opening scene or a different message.

Prompt templates for AI video ads

In AI video ad production, quality depends on how specific your prompt is. A vague request (“a nice hotel video”) gives an average result. Use this structure: subject + action + setting + lighting + camera + style + length/format.

Product template: “[Product] stands in [setting]. The camera [slowly moves closer / orbits around the product]. [Daylight / warm evening light]. Realistic, commercial-quality, clean background. Vertical 9:16, no text and no logo.”

Mood template: “[Location], [time of day]. [Moving element: curtains rippling in the breeze, gently rolling sea]. Camera [static / slow pan]. Ambient sound: [birdsong, waves]. Cinematic, natural colours.”

Template with people: “A [age range] [description of person] [action: sits down on the sofa and smiles]. Natural behaviour, understated expressions. Home setting, soft light. Close-up, slow camera movement.”

Saying “no text” in the prompt matters: AI can still struggle with rendering text and may produce meaningless letters. Add any text yourself during editing.

Example scenarios: a furniture product video and a hotel promo video

Furniture: a 15-second video for a sofa set

Imagine a sofa manufacturer in İnegöl, Türkiye's best-known furniture hub. You have studio product photos to work with.

• Scene 1 (image-to-video): Upload the sofa photo with the prompt “the camera slowly pans from right to left, morning light comes through the window”.
• Scene 2 (image-to-video): A close-up photo of the fabric, with “the camera slowly moves in on the fabric texture”.
• Scene 3 (text-to-video): A mood scene: “a family having evening tea in a modern living room, warm light”. The sofa is not clearly visible here; the scene conveys a feeling.
• Closing: In the edit, add the logo, “From our factory to your home” and the website address.

In furniture advertising, the real work starts after the video: users must be able to find the product on your website in the same colour and dimensions. We describe our experience in this sector on our furniture industry page.

Hotel: a season-opening promo video

In a hotel video, going beyond the real property carries serious risk; guests expect the pool they saw.

• Scene 1 (image-to-video): A real pool photo, with “gentle ripples on the water, palm trees swaying in the breeze, sunset”.
• Scene 2 (image-to-video): A room photo, with “curtains open gently in the breeze, the sea is visible from the balcony”.
• Scene 3 (text-to-video): A breakfast table mood shot, close-up, a generic scene that does not represent the actual property.
• Closing: A “Better rates when you book direct” message and the website.

The video brings guests to the hotel website; if they cannot find quick answers to their questions there, they often go back to an online travel agency. You can find our hotel-specific work on our hotels sector page.

Risks: copyright, labelling and misleading advertising

AI-generated content is fast, but the responsibility stays with you. The main points to watch:

• Copyright and licensing: Each tool has different terms for commercial use, and free plans may restrict it. Do not use prompts to imitate a famous brand's logo, a well-known character or a recognisable person's face.
• Music: Any music you add must be licensed for advertising. Not every track in an editing app is cleared for commercial use.
• “Made with AI” labels: Meta may display a label on content that is AI-generated or significantly altered, and Google adds an invisible watermark called SynthID to content created with Veo. Platform labelling rules are updated frequently.
• Misleading advertising: Showing a feature the product does not have, or a view the hotel does not offer, can lead to consumer complaints and penalties.

A video alone will not drive sales

Because AI makes video production easier, anyone can now make a good video. That means video is no longer the only place where you stand out from competitors. The difference comes from the system around the video:

• Targeting: Are you showing the sofa video to people who have just bought a home in your city, or to the whole country? A good video shown to the wrong audience quietly burns through your budget.
• Budget allocation: Shifting budget towards the videos that bring more clicks and conversions can make a bigger difference than producing new ones.
• Measurement: If Meta Pixel, the Conversions API, the Google Ads conversion tag and Google Analytics are not set up correctly, you cannot tell which video drives sales. Views are not sales.
• Landing page: If users click the video and land on a page that loads slowly, breaks on mobile or is a generic homepage unrelated to the product, the click you paid for is wasted. Our article on how to speed up your website is a good starting point.

At Fietto Creative, we treat video in ad campaigns together with these four elements; when creative, targeting, measurement and the landing page are discussed at the same table, results become much easier to read. If you would rather not load this work onto your own team, take a look at our SEO and ad management service.

Before spending money on ads, if you want to see whether your site is ready to welcome visitors coming from your videos, you can scan it with our free Site X-Ray tool. Your site goes through more than 40 checks, including speed, mobile-friendliness and technical SEO; the top 10 issues appear on screen instantly and the full report is sent to your inbox.

Frequently asked questions