Digital Product Ad Format Benchmarks: How to Choose Between Image, Carousel, and Video Without Borrowed Averages

Searches for ad format benchmarks usually return tables of average click-through rates and costs by format. We are not reproducing one, because we do not have a source we can verify, and because the averages answer a question you are not really asking. A benchmark pooled across industries, regions, platforms, audiences, and budgets tells you roughly where the middle of an enormous distribution sits. It does not tell you what your image ad will do for your product.

What you want to know is narrower: for my product, with my audience and offer, which format earns attention at a cost I can sustain? That is a measurement you make yourself, on a small budget, in your own ad accounts. Infinall AI helps with the production side: from a product URL it drafts image ads, carousels, and (on Pro) video drafts, which land in an Approval Queue for you to edit before export. It does not buy media or report on performance, so the benchmarking itself happens in Meta Ads Manager, Google Ads, LinkedIn Campaign Manager, or wherever you run the ads.

This article therefore does two things. It explains what each format is structurally good at for different kinds of digital products, and it describes how to build benchmarks of your own that are better than borrowed ones. Where a figure would normally appear, we note [NEEDS VERIFICATION] only if the claim depends on one; otherwise we reason from how the formats work.

Formats differ in how much information they can carry and how quickly they must earn attention.

Static image: one idea, one glance. Good for a single clear promise, a price or offer, or a recognizable interface. It is cheap to produce and easy to vary, which makes it the natural format for testing angles. Its limit is that it cannot explain anything with more than one step.

Carousel: sequence and structure. Good for showing a before-and-after, listing several benefits, walking through a process in three or four steps, or presenting a product range. It rewards a strong first card, since later cards are only seen by people who swipe. Its limit is the same: if the first card does not earn the swipe, the rest is wasted.

Short video: motion and demonstration. Good for showing how a product actually behaves, such as an app flow, a dashboard, an unboxing, or a transformation. It carries more information and more emotion per second than a still, at higher production cost and a harder first second, since viewers decide very quickly whether to keep watching.

Text-led formats such as search or newsletter ads: pure language. Good when people are already looking for the thing and need the right words, less so for creating demand.

The practical implication is that the format should follow the job of the ad. Interrupting a cold audience and explaining a complicated product are different jobs.

Different products lean on different formats, mostly because of what is easiest to show.

SaaS: the product is interface and workflow, which is hard to convey in one still. Carousels walking through a problem and a solution, and short screen-recording videos, tend to be natural candidates, with image ads for single-claim angles. B2B channels such as LinkedIn change the context: the audience is working, so clarity beats flash.

Mobile apps: the product is an experience on a small screen. Video that shows the first thirty seconds of use is a common choice, with static images for a single feature or promise. The store page that follows has to continue what the ad showed.

E-commerce and Shopify stores: the product is visible and physical. Product imagery and lifestyle images are strong, carousels suit ranges or collections, and short video suits demonstration or unboxing.

Online courses: the product is a transformation and a person. Image ads with a clear outcome promise, and short video of the instructor or a lesson excerpt, can both work; proof from students matters more than polish.

Newsletters: the product is a sample of the writing. Text-forward images or carousels featuring an excerpt, and the sign-up promise, are common.

These are starting hypotheses, not findings. Plenty of products succeed with the format that seems wrong for them, and the reason to test is that your audience may differ from the typical one.

A useful internal benchmark takes weeks, not months, and does not need a large budget, provided you design it to answer a single question.

Hold everything else constant. If you are comparing formats, use the same angle, same offer, same audience, and same landing page across an image, a carousel, and a short video. Otherwise you cannot tell whether the format or the message produced the difference.

Decide in advance what you will measure and what would change your decision. For a lead-generation offer that might be cost per qualified signup; for an app, cost per completed first action rather than cost per install. Judge on the metric closest to business value that you can measure with enough volume. [NEEDS VERIFICATION: minimum sample guidance varies by platform; check each platform's current documentation.]

Run the test long enough for the platform to stabilize before reading it, and avoid declaring winners from a handful of clicks. Record the result in a simple log: product, audience, angle, format, spend, outcome. After a few rounds, that log is a benchmark more relevant to you than any industry table.

Production cost belongs in the comparison. A video that performs slightly better but costs several times more to make may not be the better choice. Infinall's Free plan ($0, 1,000 credits per month) offers Lite mode with up to 3 image ads and 1 carousel per campaign and no video generation; Pro ("$30/month") adds up to 6 image ads, 3 carousels, and video up to 4K, which matters if video is part of the test.

A benchmark is only useful if it feeds a decision, so close the loop deliberately.

Start by separating format findings from message findings. If a carousel beat an image on one angle, check whether it held on a second angle before generalizing. A format that wins once may simply have carried a stronger message.

Next, look for interactions with placement and audience. The same video can behave differently in a feed than in a story or search context, and differently with cold audiences than with people who already know you. Keep the log specific enough to see this.

Then decide how to allocate production effort. If images and carousels perform comparably and cost far less, spend your time on more angles in those formats rather than on video. If video clearly earns more attention for your product, invest in it, and budget the review time, because a video draft still needs someone to check the script, captions, and claims.

Finally, retest periodically. Audiences tire of formats and creative, and platform behavior changes. A benchmark from last year is a hypothesis this year.

Throughout, remember what Infinall does and does not do here: it drafts the creative variants and holds them in an Approval Queue for your edits, and it can produce several formats from one campaign, but it does not run the test, measure it, or guarantee any format will perform.