There are three ways to generate ad variations, and they cost roughly €27, €15 and €5 per published ad. A person cutting each one on a timeline is the most flexible and the slowest. Modular template rendering fills hook, body and CTA slots inside a fixed timeline, which is fast and cheap per render but cannot change what the ad argues. Library-driven generation writes a new script and selects new shots from footage you already own, which moves more axes than a template but is bounded by what your camera actually captured.
Picking between them is not a question of which is best. It is a question of which axis of variation you need to move, and how many ads you need this month.
What the three approaches actually are
Manual editing means a person, a timeline and an export, whether that timeline is in a browser editor or a full NLE. Every decision is available and every decision costs minutes.
Modular recombination means building one timeline with named slots, then rendering combinations into it. The modular video ads glossary entry covers the mechanics. Swap the hook clip, swap the end card, render 200 files overnight. The pacing, the structure and the underlying argument are baked into the template, so they are the things you cannot vary without rebuilding it.
Library-driven generation means uploading footage once, having it transcribed and tagged, then generating each variation as a fresh script, shot selection, voiceover, caption set and export drawn from that library. It is not a re-cut of one timeline, which is the whole difference from the template route. It is also useless if you have no footage.
Cost per published ad, worked for 50 ads
The inputs are a blended €40 per hour for a person doing production work, €60 per hour for the motion designer who builds a template, and 50 published ads as the batch. Substitute your own rates in the video ad cost calculator.
| Approach | One-off setup | Tool cost for the batch | Human minutes per ad | Labour for 50 | Total for 50 | Cost per published ad |
|---|---|---|---|---|---|---|
| Manual editing on a timeline | €0 | €0 on a free tier | 40 | €1,333 | €1,333 | €26.67 |
| Modular template rendering, month one | €480 | €90 | 5 | €168 | €738 | €14.76 |
| Modular template rendering, month two onward | €0 | €90 | 5 | €168 | €258 | €5.16 |
| Library-driven generation | €40 | €76 | 4 | €133 | €249 | €4.98 |
The arithmetic behind each row, so you can check it:
- Manual: 50 ads at 40 minutes each is 33.3 hours, and 33.3 × €40 = €1,333. Divide by 50 for €26.67. A subscription editor adds a rounding error at this volume, not a cost.
- Modular, month one: 8 hours of template build at €60 is €480, plus about €90 of tool for the month, plus 5 minutes of asset prep and QA per ad, which is 4.2 hours at €40, or €168. Total €738, so €14.76 each.
- Modular, month two: the template is already built, so the €480 disappears and the same batch runs at €5.16.
- Library-driven: an hour to gather and upload footage is €40, one credit per variation at Growth rates is 50 × €1.52, and 4 minutes of review per ad is 3.3 hours at €40. Total €249, so €4.98 each.
The crossover points, which are the useful numbers
Manual against library-driven has no interesting crossover. Forty minutes per ad against four minutes plus an hour of setup means 40N against 60 + 4N, which flips at N of about 1.7. From the second ad onward, generation is cheaper. That is not a close call and nobody should pretend it is.
Modular against library-driven is the genuinely interesting one, because a template amortises. Modular runs at 570 + 3.33N in euros for N ads in the first month, library-driven at 40 + 4.19N. Those cross at N of about 616, both landing at €4.25 per ad. Under 600 renders off a single template, library-driven generation is cheaper. Past 600, the template wins on pure cost per render.
Six hundred renders off one timeline is a catalogue job or a localisation job, not a creative testing job. If you are rendering 600 variants of one 15-second spot for 30 markets and 20 SKUs, template rendering is the correct tool and we would tell you so. Plainly was listed at $69 per month in August 2026 and Sovran at $99 per month in August 2026. Prices move.
Which axis of variation each one moves
Cost per ad is only half the decision. The other half is whether the approach can change the thing you need changed.
| Approach | Axes it moves | Axes it holds fixed | Realistic throughput |
|---|---|---|---|
| Manual editing | Any of them, one at a time, including pacing and craft | Nothing, except by your patience | 1 to 2 ads an hour |
| Modular rendering | Hook clip, end card, on-screen text, sometimes length | The argument, the structure, the pacing, the voiceover performance | Hundreds a night |
| Library-driven generation | Script, argument, shot selection, voiceover, captions, length, structure | The footage itself, and anything not on the camera card | 10 to 50 a day |
That middle row is why template rendering plateaus for performance teams. Swapping the hook clip inside a fixed timeline moves the highest-effect variable, which is real, but the claim the ad is making stays wherever it was when the template was signed off. If your cost per acquisition problem is the proposition rather than the thumbnail, no amount of rendering touches it.
The bottom row has the opposite limit and it is worth being blunt about. Library-driven generation cannot invent a shot you did not film. If your library is 40 minutes of one wide angle, you will get four decent variations and then repetition, whatever the tool claims.
When each one is the right call
Manual editing is right below about five ads a month, for a flagship asset, and any time the craft is the point. A launch film, a founder piece, anything where a specific cut carries the meaning. It is also right when the footage needs rescuing rather than assembling.
Modular rendering is right for a locked brand timeline you are not allowed to deviate from, for catalogue-scale localisation of on-screen text, and for any volume past a few hundred renders from one template. It is the wrong tool for finding out what your ad should say.
Library-driven generation is right when you own footage, you need somewhere between 15 and 50 variants a month, and the argument itself needs to vary. Our 2026 fatigue benchmark, a synthesis of published platform and agency figures rather than our own measurement, finds brands shipping 15 to 50 creative variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers, and that most creative is effectively dead within three weeks. Fifty a month at manual rates is €1,333 of production time. The same batch at generation rates is €249. That gap is why the throughput conclusion in the benchmark, that throughput and not talent is the bottleneck, tends to be a budget statement rather than a talent statement.
Where Genyad sits, and what it will not do
Genyad is our product, so treat this as disclosure. It is the third row: upload footage once, get variations generated from the library. One standard variation costs 1 credit, AI-generated video costs 2 credits per second, and editing, re-exporting and uploading footage cost nothing. The free plan gives 5 variations with no card, Starter is €29 for 15 credits, Growth is €99 for 65 credits, and credits never expire. Batch creation of video ad variations is the route for a monthly refresh set. Scripts are written natively in English, German, French, Spanish, Italian or Hindi rather than translated, and exports come at 9:16, 4:5, 1:1 and 16:9, at 1080p on self-serve plans and 4K on Enterprise, with no watermark on any plan.
What it does not do, plainly. No AI avatars or synthetic presenters, so if you have no footage and need a person on screen, Arcads was listed at €100 per month for 10 AI actor videos in August 2026 and Creatify at $39 per month in August 2026. No product-URL import and no product-feed or CSV-driven template rendering, so catalogue work is not our lane. No static banner formats. No predicted performance scores, so it will not rank your variations before they run; AdCreative.ai was listed at $39 per month in August 2026 if a model score is what you want. And no direct publishing to Meta or TikTok, so you export the files and upload them in your ads manager.
Frequently asked questions
What is the cheapest way to generate ad variations?
Library-driven generation, at roughly €5 per published ad on a 50-ad batch, against about €27 for manual editing and €15 for template rendering in the month you build the template. Manual editing loses on cost from the second ad onward, because 40 minutes of a person's time is more expensive than four minutes of review. The exception is very high volume from a single timeline, where a template becomes cheapest past about 600 renders.
How many ad variations can I generate from footage I already have?
Multiply your count of usable distinct openings by the number of arguments your product can honestly make, and that product is your ceiling. Ten hook clips and five angles gives 50. A library of one wide shot and one voiceover gives you four or five before the variations stop being distinguishable, whatever the production method.
Is modular template rendering better than generating each variation?
For a locked timeline at catalogue volume, yes, and it is cheaper per render past roughly 600 renders from one template. For creative testing, no, because the template fixes the argument, the structure and the pacing at sign-off, and those are the variables that move cost per acquisition. The two tools answer different questions and plenty of teams run both.
Do I need a shoot before I can generate ad variations?
Not necessarily, but you need footage from somewhere: old campaign rushes, product videos, testimonial recordings, event capture. Library-driven generation is bounded by what is on the camera card, so an empty drive is the one situation where an avatar or AI-actor tool is the better starting point. Genyad has no synthetic presenters and cannot generate a person who was never filmed.