Short answer: on LinkedIn, three datasets put carousels at or near the top. On Instagram, two of three put carousels ahead of single images, and all three put them ahead of Reels on engagement rate, while Reels reach more people. On TikTok the data points the other way: video beats photo carousels on reach and interactions. On X there is no credible carousel data at all.
That's where carousel engagement benchmarks sit as of October 2026. The catch is that every study uses a different formula, so the headline percentages you have seen (21.77%, 7.00%, 0.55%) are not three opinions about the same thing. They are three different measurements.
Oakgen makes a carousel tool, so we have a stake in the answer. That is why every number below names its publisher, date, sample and formula.
Plan a week of carousels to test
Oakgen Carousels drafts a week of posts from your brand, so you can run a fair format test without building every slide by hand.
Carousel engagement benchmarks: the one-screen summary
Read the metric column first. Rates from different rows cannot be compared with each other.
| Platform | What the data says | Who says it | Metric definition | Sample | Date |
|---|---|---|---|---|---|
| Carousels (documents) top: 21.77% median vs video 7.35%, image 6.52% | Buffer, State of Social Media Engagement | Engagements / reach | 52M+ Buffer posts, 10 platforms; LinkedIn count not stated | 5 Mar 2026 | |
| Documents top at 7.00%; multi-image 6.45%; average 5.20% | Socialinsider | (Reactions + comments + shares) / impressions | 2.45M posts, brands with 1K to 1M followers | 16 Mar 2026 | |
| Carousels get 17x the interactions of images; images posted 6x more | Metricool | "Interactions", undefined | 673,658 posts, 63,108 accounts | 14 Apr 2026 | |
| Multi-image (photo grid, not PDF) 2.43% vs single image 2.13% | Ordinal | Non-click engagement / impressions | 270,000+ posts since Jan 2024 | Undated, read Oct 2026 | |
| Carousels top: 0.55% vs Reels 0.52%, images 0.37% | Socialinsider | (Likes + comments) / followers | Stated as 35M and 67.5M posts | 20 Feb 2026 | |
| Carousels 6.9% vs image 4.4%, Reels 3.3%; Reels get 1.36x the reach | Buffer, Best Content Format | Engagements / reach | Reach: 4M+ posts, Jan 2022 to Oct 2024 | 19 Mar 2026 | |
| Images ahead: 3.85% vs carousels 3.19%, Reels 2.73% | Ordinal | Non-click engagement / impressions | 12,459 carousels | Undated, read Oct 2026 | |
| TikTok | Video gets 5x the reach, 6x the interactions of images or carousels | Metricool | Not given | 2.3M+ posts, more than 92,000 accounts | 12 May 2026 |
| TikTok | Video 3.39% vs carousels and photo posts 1.92% | Buffer, Best Content Format | Not stated | Not stated | 19 Mar 2026 |
| X | No credible carousel data found | None | n/a | n/a | As of Oct 2026 |
Copy it; keep the verified date with it.
Researched October 22, 2026. We re-opened each publisher's page that day and quoted only numbers we could read there. We dropped widely repeated figures with no primary source, and one LinkedIn figure with no stated definition. We ran no benchmark of our own and use no Oakgen customer data. Disclosure: Oakgen is our product, and Oakgen Carousels makes carousels, so we benefit if you post more of them. Every publisher quoted here also sells social media software.
Why carousel statistics disagree
"Carousels get 21% engagement" and "carousels get 0.55%" can both be true. The difference sits in three places.
1. Different denominators
- Followers (Socialinsider's Instagram benchmark). Includes people who never saw the post, so rates come out tiny.
- Reach (Buffer). Only unique accounts that saw the post, so rates come out much higher.
- Impressions (Socialinsider's LinkedIn benchmark, Ordinal). Every view counts, including repeat views, so rates run lower than reach-based ones.
A made-up example, not data: an account with 10,000 followers posts a carousel that reaches 2,000 accounts, gets 3,000 impressions, and collects 100 likes and comments plus 40 saves.
| Formula | Calculation | Rate |
|---|---|---|
| Likes and comments per follower | 100 / 10,000 | 1.0% |
| Likes and comments per impression | 100 / 3,000 | 3.3% |
| All engagements, including saves, per reach | 140 / 2,000 | 7.0% |
Same post, three correct numbers, a 7x spread.
2. Different actions in the numerator
Socialinsider's Instagram formula counts likes and comments only, so saves and shares, which carousels are often credited with, are left out. Ordinal counts non-click engagement. Metricool reports "interactions" without defining them on its study page.
That matters. Metricool's press release says LinkedIn engagement rose nearly 14% year over year on "invisible interactions" such as clicks, carousel swipes and video views, while likes fell 13%, comments 17% and shares 10%. If swipes count, a 17x gap may partly measure the format's mechanics, not only how much people liked the post.
3. Different samples, and everyone is selling something
Buffer measures posts published through Buffer and says itself this is not a full-platform view. Socialinsider measures brands with 1,000 to 1 million followers. Ordinal measures its own users' posts. None is a random sample of all accounts.
Each publisher also sells social media tools, as do we, so prefer studies that publish their formula and sample. Even one publisher can frame a result two ways: Buffer's 5 March report says in its key findings that Instagram carousels earn 12% more engagement than Reels, and in its detailed section that they earn 2.09x the engagement rate (+109%). Quote the exact sentence and page you took a number from.
LinkedIn carousel engagement rate: strong across three studies
On LinkedIn a carousel is posted as a PDF document, and three separate datasets put that format at or near the top.
Buffer (5 March 2026) found carousels earned a median engagement rate of 21.77%, roughly three times video (7.35%) and images (6.52%). The more useful number is the spread: stronger carousels were above 41%, weaker ones around 5.4%. Buffer calls its interpretation "our theory": LinkedIn may be heading the way of Instagram, "where videos mean reach, but carousels mean engagement."
Socialinsider (16 March 2026) measured by impressions across 2.45 million organic posts and found native documents at 7.00% in 2025, up from 6.10% in 2024, ahead of the 5.20% platform average. It calls documents the top format.
Metricool's study page (14 April 2026) says carousels get 17x more interactions than images while images are posted 6x more often. Its press release, which compares January to February 2025 with the same months of 2026, says carousels, multi-image posts and polls consistently outperform image and video posts on Company Pages, and that on Personal Profiles video is now the most-used format but "underperforms relative to carousels, image posts and multi-image content." Remember that "interactions" is undefined and may include swipes.
Multi-image is a different format
A multi-photo post on LinkedIn is a photo grid, not a swipeable PDF. Socialinsider has it at 6.45%, close behind documents; Ordinal found a 2.43% median vs 2.13% for single images, with 3-image posts best at 3.11% (though most of its multi-image posts have no image count recorded). If someone quotes a "LinkedIn carousel" stat, check which format they mean.
Instagram carousel engagement rate in 2026: carousel vs Reels
Instagram is messier: carousels engage, Reels reach, and one dataset puts plain images ahead of both.
Socialinsider (20 February 2026) puts carousels at the highest average engagement rate in 2025: 0.55%, flat year over year, ahead of Reels at 0.52% and images at 0.37%. The carousel lead over Reels is three hundredths of a percentage point, on a formula that ignores saves and shares. Treat its year labels with care: the page notes that 2025 values are "presented as 2026", and it states its sample as 35M posts in the intro and 67.5M in the methodology.
Buffer (19 March 2026) measures by reach and finds a bigger gap: carousels at a 6.9% median, single images 4.4%, Reels 3.3%. In a separate analysis of 4M+ posts published through Buffer from January 2022 to October 2024, Reels got 1.36x the reach of carousels. That reach data is the oldest in this article.
Ordinal, measuring non-click engagement by impressions, found feed images at 3.85%, carousels at 3.19% and Reels at 2.73%. Images ahead of carousels; carousels still ahead of Reels.
What that means for carousel vs video on Instagram
Carousels tend to hold people who already see them; Reels tend to reach more people. If your audience scrolls past, test carousels. If nobody new finds you, the data does not support swapping Reels for carousels.
TikTok photo carousel reach: video still wins
Metricool (12 May 2026), across more than 2.3 million posts from more than 92,000 accounts, found video generates 5x more reach and 6x more interactions than images or carousels. Image and carousel posts also grew nearly 140% year over year, against 72.10% for video, so photo posts are getting more crowded, not less.
Buffer (19 March 2026) agrees: video at a 3.39% median engagement rate vs 1.92% for carousels and photo posts, with no sample or denominator stated for that cut.
Two datasets, same direction: do not move TikTok effort from video to photo posts expecting a lift. Photo carousels can still earn a place as extra posts when they are cheap to make, such as an Instagram carousel you already have, exported at 9:16. For TikTok's photo limits, see our guide to carousel sizes and limits on Instagram, LinkedIn, X and TikTok.
X: no credible carousel data
We found no credible 2025 or 2026 study of multi-image posts on X. X has no carousel object: a post can include up to 4 photos, 1 animated GIF or 1 video, according to X's developer documentation on media.
The nearest data point is Ordinal's: image posts on X averaged 1.47% engagement vs 1.22% for non-image, but 13,871 impressions vs 33,103. That is images in general, not multi-image sets. If someone quotes a confident "X carousel" stat, ask for the source.
What the data can and cannot justify
All of these studies are observational: they describe posts people chose to make.
What the data cannot justify:
- That carousels cause higher engagement. Brands that post carousels may simply put more budget and effort into each post.
- That a carousel will work for you. Buffer's LinkedIn spread (above 41% for strong carousels, around 5.4% for weak ones) shows the average hides huge variance. Skill and topic matter as much as format.
- That the mix effect is neutral. On LinkedIn, Metricool finds images posted 6x more than carousels. A rarer format can look good partly because only committed people use it.
What it can justify: testing carousels on LinkedIn, where three datasets point the same way; testing them on Instagram for engagement while keeping Reels for reach; keeping TikTok video-first; and measuring your own account before shifting your mix.
How to test carousels on your own account
A benchmark won't tell you what your audience does. Four weeks of paired posts will get you closer.
Pair your posts. Publish each topic twice, once as a carousel and once in your usual format (single image on LinkedIn, Reel on Instagram, video on TikTok). Otherwise you are comparing topics, not formats.
What to hold constant
- Topic pool. Write 8 topics before you start and assign each to a pair.
- Posting days and time. Same slots every week. Alternate which format goes first.
- Caption length and call to action. Same structure in both posts of a pair.
- Hook. The carousel's first slide and the other post's opening line should make the same promise.
- Posting frequency. The plan below is 4 posts a week. If that is more than you usually post, stretch it to 8 weeks rather than cramming.
Which metric to use per platform
Pick one primary metric before you start and write it down. Switching metrics after seeing results is how people talk themselves into what they wanted.
| Platform | Primary metric | Secondary metric | Why |
|---|---|---|---|
| Reactions + comments + shares, divided by impressions (calculate it yourself) | Impressions | Same ingredients as Socialinsider's formula, so you can sanity-check against it. Do not assume LinkedIn's built-in rate uses the same formula | |
| Saves and shares per account reached | Accounts reached | Reach-based, and counts the actions carousels are meant to earn. Use the same reach figure every time | |
| TikTok | Views | Engagements per view | Reach is where the 2026 data says photo posts fall short; test that directly |
The 4-week test plan
| Week | What you post | Hold constant | Record | Stop or continue |
|---|---|---|---|---|
| 0 (prep) | Nothing new. Pull your last 8 to 10 posts | n/a | Your current median on the primary metric: your baseline | Write the 8 topics and the metric down before week 1 |
| 1 | 2 pairs (topics 1 and 2), carousel and usual format | Days, times, caption structure, CTA | Primary metric 7 days after posting | Continue unless a post breaks a platform rule or gets removed |
| 2 | 2 pairs (topics 3 and 4), format order reversed | Same as week 1 | Same, at 7 days | Continue |
| 3 | 2 pairs (topics 5 and 6) | Same | Same | Continue. Flag any post far above everything else as an outlier |
| 4 | 2 pairs (topics 7 and 8), format order reversed | Same | Same | After the last post's 7-day mark, compare |
That gives you 8 carousels and 8 comparison posts, still a small sample. At the end:
- Count pair wins on your primary metric. If carousels win 7 or 8 of 8 pairs, act on it: that is unlikely to be chance (about 3.5% odds if the formats were equal). 6 of 8 is a lean, so run another round before changing your mix. 5 or fewer is noise: the topic mattered more than the format.
- Compare medians, not averages. One post that takes off will drag an average around. Report results with and without any outlier you flagged.
- Check the secondary metric. If Instagram carousels won on saves per reach but reached half as many accounts, you have learned the trade-off, not a winner.
When to stop early: if you change something you meant to hold constant (a launch, a paid boost, a collab), restart the affected pair. If one format loses every pair in the first two weeks by a wide margin on all metrics, you do not need two more weeks.
Our piece on A/B testing psychology for ad creative covers the biases that creep into tests like this. And if writing 16 posts in advance sounds like the hard part, the batch workflow for 30 days of posts applies here too.
Draft the carousel half of your test
Give Oakgen Carousels your brand once and plan a week of carousels, then edit each one before you post.
Reading benchmarks for your follower tier
Absolute numbers by tier show what "normal" looks like at your size. Socialinsider's Instagram benchmark reports average carousel saves and views per post:
| Followers | Avg. saves per carousel | Avg. views per carousel |
|---|---|---|
| 1K to 5K | 1 | 993 |
| 5K to 10K | 2 | 2,117 |
| 10K to 50K | 8 | 4,275 |
| 50K to 100K | 35 | 11,597 |
| 100K to 1M | 98 | 35,370 |
With 3,000 followers and 1 or 2 saves per carousel, you are around the average for your tier in that sample, not failing.
On LinkedIn, Socialinsider's impressions per post for pages with 10K to 50K followers were 1,850 for multi-image, 1,450 for single images, 1,420 for video and 1,150 for native documents. Documents top its overall engagement-rate ranking, yet at this size they averaged fewer impressions than three other formats.
Both samples are brand accounts with 1K to 1M followers. Personal profiles and smaller pages sit outside them, one more reason the week-0 baseline matters.
Where Oakgen fits, and when you don't need it
Oakgen is our product. The hard part of a fair test is producing enough decent carousels on schedule. Oakgen Carousels plans a week of posts from your brand book, writes and lays out the slides, and exports a LinkedIn PDF, Instagram 4:5 or TikTok 9:16 slides. Writing the copy is free; generated artwork costs credits, and generating needs a paid plan (see pricing). It can schedule to Instagram and Facebook Pages; LinkedIn and TikTok are downloads you post yourself.
For Instagram posts published through Oakgen, the week planner is built to read back saves and reach and, once at least 3 posts have them, use them to steer the purpose mix (teaching, story, offer and so on). It does not choose between formats, and we have not verified that read-back on live accounts, so do not count on it for this test.
You do not need Oakgen if you already have a designer or a template you like. To start from structures that already worked, see how to recreate a carousel that already works using the structure, not the words, or our older guide to creating an Instagram carousel with agentic chat.
Frequently asked questions
Do carousels get more engagement than video?
On LinkedIn, Buffer and Metricool both put carousels ahead of video. On Instagram, all three studies covered here put carousels ahead of Reels on engagement rate, though Reels reached more people in Buffer's data. On TikTok, video wins in both Metricool's and Buffer's data.
Do Reels get more reach than carousels?
In Buffer's analysis of 4M+ Instagram posts from January 2022 to October 2024, Reels reached 1.36 times as many accounts as carousels. Carousels earned a higher engagement rate among the people they did reach.
What is a good LinkedIn carousel engagement rate?
It depends on the formula. Buffer reports a 21.77% median using reach, Socialinsider reports 7.00% for documents using impressions. Compare your posts to your own baseline with one consistent formula rather than to either number.
Are TikTok photo carousels worth posting?
As a supplement, maybe. Metricool's 2026 study found video gets about 5x the reach and 6x the interactions of image and carousel posts, so do not expect photo posts to outperform your videos.
Why do carousel statistics contradict each other?
Each study divides engagement by something different (followers, reach or impressions), counts different actions, and samples a different set of accounts. Two numbers from two tools are not comparable.
How many posts do I need to test carousels on my own account?
Aim for at least 8 posts per format, paired by topic, over four weeks. That is still a small sample, so act only when carousels win nearly every pair.
Is there data on carousels for X?
We found no credible 2025 or 2026 study of multi-image posts on X. X allows up to 4 photos per post, so a longer deck needs a thread, a link or another platform.
Sources
All pages read on October 22, 2026.
- Buffer, The State of Social Media Engagement in 2026, 5 March 2026
- Buffer, Best Content Format on Social Platforms in 2026, 19 March 2026
- Socialinsider, LinkedIn benchmarks, 16 March 2026
- Socialinsider, Instagram organic benchmarks, 20 February 2026
- Metricool, LinkedIn trends study, 14 April 2026 (17x, 6x, sample)
- Metricool press release via Social Media Today, the rise of invisible interactions, 14 April 2026 (period, engagement change, Company Page and Personal Profile findings)
- Metricool, TikTok study 2026 press release, 12 May 2026
- Ordinal, social media images data, undated
- X developer documentation, media introduction
Start your carousel test week
Plan a week of carousels in Oakgen, post them alongside your usual format, and compare the results yourself.
