Meta Ad Formats: Which One, When, and Why

Most advertisers pick a Meta ad format out of habit, not strategy. They run single images because they always have, or they chase Reels because someone said video is the future. Both reflexes leave money on the table. The format is not a style choice. It is a lever that changes cost, attention and conversion depending on where someone sits in your funnel and which placement Meta serves. This article walks through the six formats that matter in 2026, with real specs and real data, and tells you which one to reach for and when. It also kills two myths along the way: that video always beats image, and that the carousel is dead. Both are wrong, and believing them quietly drains budget.

Why vertical comes first now

Before formats, placements. The screen is vertical now, and the numbers are no longer debatable. Tinuiti benchmark data showed Reels ads reaching 26 percent of all Instagram ad impressions in Q3 2025, up from 19 percent a year earlier. On the same platform, Stories held a plurality at around 44 percent of impressions, while Feed dropped to roughly 31 percent. On Facebook, total Reels-related inventory rose from 17 percent to 31 percent of impressions year over year in Q2 2025. The vertical surfaces are where the volume lives. If your creative is built only for the square or the horizontal Feed, you are forcing Meta to crop it for the placements that now carry most of the delivery.

Vertical first does not mean vertical only. It means you design the 9:16 frame as the primary asset, then adapt down, instead of designing a square and praying the crop survives. Meta data cited across 2025 found 9:16 vertical video drove around 41 percent higher engagement than cropped formats, and a meta-analysis of fifteen Reels split tests reported sound-on vertical video respecting safe zones beating still-image assets with high statistical confidence in ecommerce verticals. Treat the agency split-test figure as directional, not gospel. But the structural point holds: a vertical-native asset reaches the high-volume placements at full frame, while a cropped one shows up boxed, shrunk and easy to scroll past.

It helps to name the six formats cleanly before pitting them against each other, because the marketing chatter blurs them. Single image is one static frame. Video is one moving asset. Carousel is a swipeable set of 2 to 10 cards. Collection is a cover plus a product grid that opens a storefront. Stories and Reels are vertical placements you fill with video. Confusing a placement with a format is the first mistake; treating all six as interchangeable is the second. Each was engineered for a different job, and Meta prices and delivers them differently. Once you see them as distinct tools rather than a menu of styles, the question stops being which is best and becomes which fits this audience, this offer and this moment in the funnel.

One last reframing on vertical before the formats. The shift is not aesthetic, it is economic. Meta sells impressions through an auction, and the auction now overwhelmingly serves vertical inventory because that is where attention sits. When you upload only horizontal or square creative, you do not just look dated, you literally compete in fewer auctions, because the model has nothing well-shaped to place in the surfaces carrying the most volume. Vertical-native creative is therefore not a creative-team preference, it is auction access. The advertisers who treated vertical as optional in 2024 quietly lost reach to the ones who made it the default, and the impression-share gap above shows exactly where that lost reach went.

The safe zone trap nobody mentions

Here is the detail that wrecks otherwise good vertical creative. Stories and Reels overlay your ad with interface elements: the profile name and call to action at the bottom, the progress bars and close button at the top. Meta guidance points to keeping roughly the top 14 percent, the bottom 35 percent and the sides 6 percent clear of key content. That bottom third is enormous, and most advertisers ignore it, then wonder why their headline is half-hidden behind a Shop Now button. A vertical asset that puts the logo at the very bottom and the price at the very top is technically 9:16 and practically broken. Design to the safe zone or the placement quietly punishes you with worse delivery.

Single image: underrated, not obsolete

The single image is the format everyone calls dead and everyone keeps using, because it works. The recommended spec is 4:5 (1080 x 1350) for Feed, which a single asset can cover across Facebook Feed, Instagram Feed, Threads and several other surfaces. Add a 9:16 version for Stories and Reels. The myth is that static lost to video. The data says otherwise depending on the goal. Hootsuite Social Trends 2025 put average video click-through at 1.14 percent versus 0.90 percent for static, so video wins on clicks. But aggregated Meta benchmarks showed image creative beating video on cost per thousand impressions by roughly 38 percent in cold prospecting. Cheaper reach, faster comprehension. The image is not a relic, it is the efficient default.

Reach for the single image when speed and cost matter more than narrative. A clean image with strong copy can land an offer in under a second, which is exactly what a tight CPA conversion campaign needs. It is also the fastest creative to produce: a static ships in under an hour, so you can run four variants for the cost and time of one video. That iteration speed is its real weapon. In cold prospecting tests where you are hunting for a winning angle, the image lets you test more hooks per week on smaller budgets, exit the learning phase sooner, and only then commit production effort to the angle that proved itself in video.

There is a quieter reason the image keeps winning where it matters: it survives bad conditions. Most people scroll with sound off, in a hurry, on a cracked phone in poor light. A static frame still communicates fully in that environment, while a video that depends on audio or a slow reveal simply fails. The image also ages well in your library. A strong static can run for months before it fatigues, whereas a video angle often burns out faster because the novelty that carried it wears off. Treat your best static as an account workhorse, not a placeholder you upgrade away from. Many top-performing conversion ads on Meta are still a single well-cropped image with one sharp line of copy.

Video: powerful, and overprescribed

Video is the format people reach for to look modern, and it earns its reputation on the upper funnel. It holds attention, it demonstrates a product in motion, it builds the brand association that a static cannot. For Reels and Stories, the spec is 9:16 (1080 x 1920), capped at 90 seconds, with a performance sweet spot of 15 to 30 seconds. For in-Feed video, 4:5 mirrors the single image. The mistake is treating video as the universal upgrade. It is not. Video costs more to produce, takes longer to iterate, and on tight-CPA conversion goals it often loses to a sharp static because it makes the viewer wait for the value instead of stating it instantly.

The honest answer is a mix, not a winner. Sprout Social Index 2025 reported brands running mixed-format strategies saw around 19 percent higher overall ROAS than single-format competitors, and aggregated ecommerce data through early 2026 pointed to roughly a 70/30 video-to-image split outperforming single-format accounts. So video earns the larger share, but starving your account of static is as much a mistake as never shooting video. Use video to build attention and demonstrate, use image to close efficiently, and let the model serve whichever wins the auction for each person. The format war is a false frame. The accounts that win run both and let performance, not fashion, decide the ratio.

One real change worth flagging: Meta removed post-loop Reels ads starting in November 2025, the format that played an ad after an organic Reel finished, while keeping the overlay format. If your playbook leaned on post-loop placements, that lever is gone and your vertical video now competes in-feed against organic Reels directly. That raises the bar for the first second. A video that opens on a logo animation or a slow establishing shot loses the viewer before the message lands. Front-load the payoff: show the product, the transformation or the hook immediately, then let the rest of the clip justify the watch. The 90-second ceiling tempts people to fill it; the data on the 15 to 30 second sweet spot says do not.

Carousel: the format wrongly declared dead

The carousel is the format the trend crowd buried years ago and the data keeps resurrecting. It lets you show 2 to 10 cards, each with its own image or video, headline, description and destination URL, in a single ad unit. The recommended card ratio is 1:1 (1080 x 1080) because it displays cleanly across the most placements. Meta has stated that carousel link ads can deliver 30 to 50 percent lower cost per conversion and 20 to 30 percent lower cost per click than single-image ads when used well. Far from dead, the carousel is one of the most cost-efficient direct-response formats Meta offers, especially for retargeting warm audiences.

The carousel earns its keep in three jobs. First, multi-product showcasing: each card is a different SKU, ideal for a category retargeting set. Second, sequential storytelling: card one hooks, the next cards build the argument, the last card closes, which suits a considered purchase. Third, the how-it-works breakdown: one feature or step per card. The critical discipline is the first card. Reporting across accounts suggests roughly 70 percent of carousel viewers never swipe past card one, so that opening frame must carry the hook and the core value on its own. Build the carousel as if most people will only ever see the first card, then reward the ones who swipe.

One caveat worth stating plainly: the cost-per-conversion figures come from Meta and from agencies promoting the format, not from independent audits. Treat the exact percentages as directional. The pattern, though, is consistent across enough sources to trust the direction: for warm retargeting and multi-product catalogs, the carousel routinely beats a single image on efficiency. If you dropped the carousel because a blog told you it was outdated, you most likely cut one of your cheapest sources of conversions. Test it again against your single-image control on a warm audience and read the cost per result, not the trend article.

Collection: the mobile storefront

Collection is the format built for catalog ecommerce on mobile. It pairs a cover image or video with a grid of product cards pulled from your catalog underneath, and tapping it opens a full-screen Instant Experience storefront without leaving the app. The cover spec is 1:1 (1080 x 1080), with product images drawn from your feed. It runs across Feed, Stories, Reels, Marketplace and Search. Note an important change: as of March 2026, Collection was moved out of the main Ad Setup flow and is now configured through Ad Creative under Format Display settings, so if you cannot find it where it used to be, that is why, not because Meta retired it.

Use collection when you have a real catalog and a mobile-heavy audience, because that is exactly the situation it was designed to monetise. It shines paired with Advantage+ Shopping, which Meta benchmarks credit with cutting CPA by up to 32 percent versus manually configured campaigns, though independent agency data such as MHI puts the realistic gain closer to 17 percent for brands meeting the bar of 30-plus SKUs, 15-plus creatives and clean conversion tracking. The collection plus Advantage+ Shopping pairing is the strongest combination for ecommerce, but only when the catalog and signal are healthy. With three products and a leaky Pixel, the format has nothing to dynamically populate and the advantage evaporates.

Collection also quietly solves a creative-volume problem. Because the grid is populated dynamically from your catalog feed, you do not hand-build a creative per product. You design one strong cover and the format assembles the rest, pulling fresh products, prices and availability automatically. That is leverage: a small team can run a deep catalog without producing hundreds of static ads. The trade-off is that your catalog feed becomes the creative. Bad product photography, missing titles or stale prices show up directly in the ad, because the format displays exactly what your feed contains. Before you scale collection, audit the feed the way you would audit a landing page, because for this format the feed is the landing page.

Stories and Reels are placements, not just formats

A common confusion: people talk about Reels and Stories as if they were ad formats like the image or the carousel. They are placements, the vertical full-screen surfaces, that you fill with a format, usually video. Stories ads sit between organic Stories on Facebook and Instagram at 9:16 (1080 x 1920), with each story card holding up to 15 seconds before Instagram auto-splits a longer video into 15-second segments. Reels ads play in the dedicated Reels feed, also 9:16, capped at 90 seconds. The practical consequence: you do not choose Reels instead of video, you choose to run your vertical video in the Reels placement, ideally as a Reels-native asset rather than a recycled Feed clip.

The native-versus-recycled distinction matters more than the placement choice itself. A Reels ad that looks like an ad, polished, branded, clearly produced, gets scrolled past in the same feed where organic creators rack up views with raw, sound-first, fast-cut clips. The Reels placement rewards content that matches the surrounding native content, which means real audio, a hook in the first second, and motion that earns the next second. Meta has repeatedly said Reels ads with key messages inside the safe zone, quality audio and true vertical video see better delivery to the placement. Recycling a horizontal brand film into Reels is the single most common reason a vertical campaign underdelivers.

There is also a budget angle people skip. Different formats reach the learning phase at different speeds, and that shapes what you can afford to test. A single image, cheap to make and cheap to serve, can exit learning on a small daily budget, so it is the rational choice when you are validating an audience or an offer you are unsure about. Video and collection demand more spend and more time before their numbers stabilise, so committing to them before you have a proven angle wastes budget on assets you may scrap. Sequence the formats to your certainty: cheap and fast while you are still learning what works, richer and pricier once a winner has earned the investment. The order matters as much as the choice.

A practical decision guide

Strip away the format folklore and the choice gets simple. Cold prospecting on a tight budget: lead with single images to test angles cheaply and fast, then promote the winners into vertical video. Building awareness or demonstrating a product: vertical video in Reels and Stories, native and sound-first. Warm retargeting and multi-product reminders: carousel, with the hook front-loaded onto card one. Catalog ecommerce with a mobile audience and a healthy feed: collection paired with Advantage+ Shopping. Notice none of these is a single answer. A healthy account runs several formats at once and lets the model serve the right one to each person, because the model reads the auction better than any rule of thumb you could memorise.

The deeper rule sits above all of them: produce in vertical-native ratios, design to the safe zone, and give Meta format diversity to choose from. Advantage+ placements work best when fed 4:5, 9:16 and 1:1 versions of your creative so the model can serve the right shape to the right surface without cropping. The format question is real, but it is downstream of a bigger one. Have you given the algorithm a clean, full vertical frame and more than one format to test? Do that, and most of the agonising over which single format is best simply dissolves, because you stop betting on one and let the auction pick the winner per impression.

If you remember one thing, make it this: stop asking which format is best and start asking which job needs doing. Best is a question with no fixed answer, because it changes with the audience temperature, the offer, the placement and the week. The format is a tool, and you would not ask which tool in a workshop is best either. You would ask what you are building. Prospect cheaply with images, demonstrate with vertical video, remind and cross-sell with carousels, monetise a catalog with collection, and feed Meta several shapes so the auction can finish the job. Do that consistently and the format anxiety that fills the forums simply stops being your problem.

Sources

Tinuiti Digital Ads Benchmark Reports Q2 and Q3 2025 (Reels 26 percent of Instagram impressions Q3, Stories 44 percent plurality, Feed 31 percent, Facebook Reels inventory 17 to 31 percent); Meta Business Help Centre best practices for aspect ratios (4:5 Feed, 9:16 Stories and Reels, 1:1 carousel, safe zones top 14 percent, bottom 35 percent, sides 6 percent); Meta for Business Reels ads guidance (90 second cap, 15 to 30 second sweet spot, native vertical sound-on delivery); Hootsuite Social Trends Report 2025 (video CTR 1.14 percent vs static 0.90 percent); aggregated Meta for Business benchmarks (image roughly 38 percent lower CPM in cold prospecting); Sprout Social Index 2025 (mixed-format 19 percent higher ROAS); Meta statement on carousel link ads (30 to 50 percent lower cost per conversion, 20 to 30 percent lower CPC); reported carousel first-card drop-off (around 70 percent never swipe); Meta benchmarks and MHI agency data on Advantage+ Shopping CPA (up to 32 percent Meta, around 17 percent independent); Meta product update moving Collection into Ad Creative Format Display (March 2026). Figures attributed to Meta and agencies are reported by their authors and not independently audited.