What Makes Social Media Content Perform, 2027 Report
What 1.2 million posts from 22k brand profiles show about images, video, captions and hashtags on Facebook, Instagram, LinkedIn, TikTok, YouTube, X and Threads
Posts from August 1 to October 31, 2026. Published .
- Data window
- Aug 1 to Oct 31, 2026
- Posts
- 1.2 million posts
- Brands
- 22k brand profiles
- Coverage
- 7 networks
- Published
- November 2026
On this page
Key findings
Vista Social data from 1.2 million brand posts, August 1 to October 31, 2026. Every comparison is against the same brand's other posts.
- 01
+36%
Brand videos of 1 to 3 minutes get 36% more engagement on YouTube than the same brand's other videos, and three in four brands see the gain. The lift is 14% on Instagram and 10% on TikTok.
- 02
-17%
Brand videos under 15 seconds get 17% less engagement on Instagram than the same brand's other videos, and about three in four brands see the drop. They do 25% worse on YouTube and 15% worse on TikTok.
- 03
+18%
Brand videos with someone speaking get 18% more engagement on YouTube than the same brand's videos without a voice. The lift is 14% on TikTok and 11% on Instagram. Videos with music only do 10 to 15% worse on all three.
- 04
7 of 7
Opening a video with a question is the most common hook brands use, and it gets less engagement than the same brand's other videos on all 7 networks we measured. Opening with a bold claim or direct statement does better on all seven.
- 05
+12%
Brand videos with on-screen captions get 12% more engagement on YouTube than the same brand's videos without them, and 9% more on Instagram and TikTok.
- 06
-20%
Facebook image posts with a link in the caption get 20% less engagement than the same brand's image posts without one, and two in three brands see the drop.
- 07
+11%
Brand videos with a call to action at the end get 11% more engagement on YouTube than the same brand's other videos. Six in ten brand videos have no call to action at all.
- 08
+20%
On Instagram, brand videos get 20% more engagement than the same brand's images, and two in three brands see it. On X it's the other way: videos get 36% less engagement than images.
- 09
+14%
Inspirational videos get more engagement than the same brand's other videos on every major network, up to 14% more on YouTube.
- 10
+1 to 5%
Brand images with a face get only 1 to 5% more engagement than the same brand's images without one, on Facebook, Instagram and LinkedIn.
About this report
Most advice on what content works comes from a handful of viral examples or from what marketers say they do. This report measures 1.2 million images and videos that 22k brand profiles, managed from 4.9k Vista Social accounts, published on seven networks, and how audiences responded to each one.
Vista Social analyzes every image and video when it is uploaded: what's in it, who's in it, how it opens, what it sounds like and how long it runs. We matched those labels with each post's engagement and impressions.
Every number compares a brand with itself. For each brand, we compared the engagement rate of its posts with a feature (for example, a voiceover) to its posts without it, then took the median across brands. So "+11%" means the typical brand gets 11% more engagement on posts with that feature than on its own posts without it. Big brands don't outweigh small ones, and an industry that posts a lot of one kind of content doesn't skew the result.
No figure describes a single customer. A number is published only when it rests on at least 50 brand profiles managed from at least 10 different Vista Social accounts.
Terms used in this report
- Engagement rate: a post's engagements (reactions, comments, shares, saves and similar) divided by its impressions.
- Lift: how much more or less engagement a brand gets on posts with a feature than on its own posts without it, as a percent. "+11%" means the typical brand gets 11% more.
- Same brand comparison: each brand is compared only with itself, then we take the median across brands. This keeps big brands, and industries that favor one kind of content, from skewing the result.
- Brand profile: one account on one network, such as a company's Instagram account or its LinkedIn page.
- Hook: how a video opens in its first seconds, for example with a question, a bold claim or a visual reveal.
Table 1.1 Images and video by network
| Network | Image posts | Video posts | Share video |
|---|---|---|---|
| 263k | 155k | 37% | |
| 115k | 187k | 62% | |
| 51k | 58k | 53% | |
| TikTok | 11k | 124k | 92% |
| YouTube | - | 125k | 100% |
| X | 35k | 38k | 52% |
| Threads | 11k | 20k | 64% |
Posts with exactly one image or one video, August 1 to October 31, 2026. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
Table 1.2 Engagement rate by format
| Network | Median engagement rate, image posts | Median engagement rate, video posts | Video against the same brand's images |
|---|---|---|---|
| 0.8% | 0.8% | +19% | |
| 2.6% | 2.7% | +20% | |
| 4.8% | 3.1% | +6% | |
| TikTok | 0.8% | 2.8% | +96% |
| YouTube | - | 1.0% | - |
| X | 1.8% | 1.2% | -36% |
Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
The median columns are the typical brand's median engagement rate per post. The last column compares video and images within brands that post both. On TikTok, videos get nearly twice the engagement of the same brand's photo posts. The two kinds of column can disagree: on LinkedIn, the brands that post video have lower engagement overall, but each of them still does slightly better with video than with images.
- About half of brand images show no people. 47 to 50% of images on Facebook, Instagram and LinkedIn have no face in them. One image in five shows exactly one face.
- Graphics are the most common image. Designed graphics and text cards are the largest single scene type, 14 to 21% of images depending on the network, ahead of photos of people (8 to 12%) and events (8 to 10%).
- Most brand video has a voice. A voiceover is the most common soundtrack (40 to 52% of videos), followed by music only (17 to 28%) and dialogue (15 to 22%).
- The typical brand video runs 30 to 59 seconds. That band covers 29 to 40% of videos. Under 15 seconds is 13 to 27%, 1 to 3 minutes is 23 to 30%, and only about 1 to 2% run longer than 3 minutes.
- Six in ten brand videos have no call to action. When there is one, it almost always comes at the end.
- Hashtags are fading on Facebook. 70% of Facebook image posts carry none, against 37% on Instagram.
Engagement of videos in each length band, against the same brand's other videos.
Figure 1
Video length: engagement against the same brand's other videos
- Under 15 seconds-17%
- 15 to 29 seconds-4%
- 30 to 59 seconds+4%
- 1 to 3 minutes+14%
- Under 15 seconds-15%
- 15 to 29 seconds-4%
- 30 to 59 seconds+5%
- 1 to 3 minutes+10%
- Under 15 seconds-25%
- 15 to 29 seconds-14%
- 30 to 59 seconds+1%
- 1 to 3 minutes+36%
- Under 15 seconds-8%
- 15 to 29 seconds-3%
- 30 to 59 seconds+1%
- 1 to 3 minutes+2%
- Under 15 seconds-3%
- 15 to 29 seconds-4%
- 30 to 59 seconds-1%
- 1 to 3 minutes0%
- Under 15 seconds+5%
- 15 to 29 seconds-4%
- 30 to 59 seconds-1%
- 1 to 3 minutes0%
Table 2.1 Video length
| Length | TikTok | YouTube | X | |||
|---|---|---|---|---|---|---|
| Under 15 seconds | -8% | -17% | -3% | -15% | -25% | +5% |
| 15 to 29 seconds | -3% | -4% | -4% | -4% | -14% | -4% |
| 30 to 59 seconds | +1% | +4% | -1% | +5% | +1% | -1% |
| 1 to 3 minutes | +2% | +14% | 0% | +10% | +36% | 0% |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
On Instagram, the few brand videos longer than 3 minutes did best of all, 27% above the same brand's other videos. On YouTube, three in four brands get more engagement on their 1 to 3 minute videos than on their shorter ones.
The advice to keep everything under 15 seconds doesn't hold for brands. Very short clips are often loops, teasers or reposted ads, and audiences treat them that way. A video long enough to say something is the better bet on Instagram, TikTok and YouTube. On X and Threads, short clips do slightly better.
Table 2.2 Voice and sound
| Soundtrack | TikTok | YouTube | ||
|---|---|---|---|---|
| Voiceover | +7% | +8% | +5% | +7% |
| Voiceover and music | -1% | +7% | +4% | +10% |
| Dialogue | -7% | -9% | -3% | -8% |
| Music only | -6% | -10% | -10% | -15% |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
Any voice, voiceover or dialogue, against videos with no voice: +18% on YouTube, +14% on TikTok, +11% on Instagram, +4% on Facebook, with about two in three brands seeing the gain on YouTube, TikTok and Instagram. LinkedIn shows no lift, and on X videos with a voice do 6% worse.
Note the split between voiceover and dialogue. A narrator explaining what's on screen does better than people talking to each other in a clip.
- A presenter (someone speaking to the viewer) lifts engagement by 9% on YouTube and TikTok and 7% on Instagram.
- A visible face lifts it by 9% on Instagram, 8% on TikTok and 7% on YouTube.
- On X, both go the other way: videos with a presenter do 6% worse.
Videos with captions or subtitles on screen: +12% on YouTube, +9% on Instagram and TikTok, +5% on Facebook. No effect on LinkedIn. Most people scroll with the sound off, and captions let them follow anyway.
The first seconds of each video are classified by how they open.
Table 2.3 Hooks
| Hook | Share of videos | Lift range across networks |
|---|---|---|
| Question | 23 to 29% | -2% to -9%, worse on all seven networks |
| Visual reveal | 16 to 29% | about 0, mixed |
| Problem or pain point | 8 to 27% | mixed, -5% to +4% |
| Bold claim or direct statement | 14 to 20% | +2% to +6%, better on all seven networks |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
"Have you ever wondered...?" is the most common way brands open a video, and the weakest. Saying the point outright works better everywhere.
Table 2.4 Calls to action
| Call to action | TikTok | YouTube | |
|---|---|---|---|
| At the end | +8% | +4% | +11% |
| None | -7% | -3% | -13% |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
A call to action throughout the video, or right at the start, does not help.
Inspirational videos beat the same brand's other videos on every major network: +14% on YouTube, +11% on LinkedIn, +9% on Facebook, +7% on TikTok and Instagram. Warm and inviting videos do even better where brands post them (up to +20% on TikTok), though they are only 1 to 2% of videos. Informative videos, the most common mood at about 40 to 68% of videos, do slightly worse on most networks.
Table 3.1 Faces
| Image | TikTok | X | Threads | |||
|---|---|---|---|---|---|---|
| Image has a face | +1% | +4% | +5% | +10% | 0% | -11% |
| Face looking at the camera (among images with faces) | +3% | +2% | +5% | -2% | +2% | -2% |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
Faces help, but far less than the usual advice suggests. On LinkedIn, group photos of 4 or more people do best (+13%), which fits team, event and award posts.
Table 3.2 What's in the picture
| Scene | |||
|---|---|---|---|
| Food and drink | +12% | +6% | - |
| Sports and fitness | +1% | +11% | - |
| People | +1% | +7% | +5% |
| Outdoor and nature | +3% | +6% | +1% |
| Event | -3% | -1% | +6% |
| Promotional | -8% | -3% | 0% |
| Graphic or text design | -7% | -7% | -3% |
Change in engagement rate against the same brand's other posts, median across brands. Source: Vista Social Research, posts from Aug 1 to Oct 31, 2026.
Graphics and promotional images are what brands post most, and they do worse than the same brand's photos. The gap is widest on TikTok, where promotional images do 17% worse.
Images the AI reads as informative do worse on Facebook (-11%), Instagram (-12%), LinkedIn (-6%) and TikTok (-9%). Inspirational images do better on Facebook (+10%) and X (+10%), and energetic and fun images do better on Facebook, Instagram and LinkedIn (+4 to +5%). Serious images do worse everywhere except X.
Part 5
Figure 2It depends on the network
The same choice can help on one network and hurt on another.
Figure 2
Video features: engagement against the same brand's other videos
- Someone speaking+4%+4% · 55% of brands did better · 1,022 brands
- On-screen captions+5%+5% · 55% of brands did better · 953 brands
- A presenter on camera+4%+4% · 55% of brands did better · 957 brands
- A visible face+5%+5% · 55% of brands did better · 617 brands
- A call to action0%0% · 49% of brands did better · 1,246 brands
- Someone speaking+11%+11% · 63% of brands did better · 1,261 brands
- On-screen captions+9%+9% · 60% of brands did better · 1,320 brands
- A presenter on camera+7%+7% · 57% of brands did better · 1,170 brands
- A visible face+9%+9% · 58% of brands did better · 898 brands
- A call to action+6%+6% · 59% of brands did better · 1,664 brands
- Someone speaking-1%-1% · 48% of brands did better · 224 brands
- On-screen captions0%0% · 50% of brands did better · 280 brands
- A presenter on camera+1%+1% · 51% of brands did better · 219 brands
- A visible face+4%+4% · 56% of brands did better · 182 brands
- A call to action+2%+2% · 51% of brands did better · 432 brands
- Someone speaking+14%+14% · 63% of brands did better · 787 brands
- On-screen captions+9%+9% · 63% of brands did better · 930 brands
- A presenter on camera+9%+9% · 61% of brands did better · 759 brands
- A visible face+8%+8% · 60% of brands did better · 584 brands
- A call to action+2%+2% · 53% of brands did better · 1,107 brands
- Someone speaking+18%+18% · 64% of brands did better · 580 brands
- On-screen captions+12%+12% · 63% of brands did better · 695 brands
- A presenter on camera+9%+9% · 58% of brands did better · 485 brands
- A visible face+7%+7% · 57% of brands did better · 387 brands
- A call to action+10%+10% · 60% of brands did better · 905 brands
- Someone speaking-6%-6% · 42% of brands did better · 154 brands
- On-screen captions-3%-3% · 44% of brands did better · 166 brands
- A presenter on camera-6%-6% · 44% of brands did better · 131 brands
- A visible face-7%-7% · 45% of brands did better · 92 brands
- A call to action+4%+4% · 55% of brands did better · 196 brands
- Short clips: much worse on YouTube, Instagram and TikTok, slightly better on X and Threads.
- A presenter or a voice: up to 18% better on YouTube, TikTok and Instagram, about 6% worse on X.
- Faces in images: 10% better on TikTok, 11% worse on Threads.
- LinkedIn video barely responds to any of it. Voice, captions, presenters and length all land within about 4% of zero. On LinkedIn, the topic and the poster matter more than the production.
Part 6
8 industriesBy industry
Industries with enough brands for at least a few comparisons, using the industry names from the Social Media Benchmarks by Industry, 2027 Report.
The strongest effects in the report. On Instagram, videos with a presenter do 24% better, with a voice 22% better and with captions 20% better. TikTok videos with captions do 31% better. Facebook image posts with a link do 37% worse, the largest drop we measured, and three in four health care brands see it. Videos under 15 seconds do 24% worse on Instagram, and 1 to 3 minute videos do 33% better on YouTube.
Food photos beat a restaurant's other images by 9% on Facebook and Instagram. A call to action in an Instagram video lifts engagement by 20%, and eight in ten restaurants see it. Voice helps on Instagram (+16%) and TikTok (+18%). Images with text on them do 18% worse on Instagram and 15% worse on Facebook.
On YouTube, 1 to 3 minute videos do 54% better than shorter ones. Voice lifts Instagram and TikTok video by about 15%. Facebook image posts with a link do 22% worse. Videos under 15 seconds do 22% worse on Instagram and 17% worse on TikTok.
Voiceover lifts video by 17% on Facebook and YouTube. TikTok videos with captions do 20% better, and YouTube videos with a call to action 18% better. Facebook image posts with a link do 20% worse, and with a question 16% worse. On LinkedIn, images with faces do 11% better.
On Instagram, 1 to 3 minute videos do 20% better, and eight in ten brands see it. Voice lifts video by 12 to 17% on Instagram, TikTok and YouTube. On Facebook, promotional images do 24% worse and images with text 19% worse.
On YouTube, 1 to 3 minute videos do 65% better than the same school's shorter videos. Voiceover lifts Instagram video by 19%. On Instagram, images with faces do 15% better.
Videos with a call to action do 17% better on Instagram. Videos of 1 to 3 minutes do 24% better on TikTok and 44% better on YouTube, and videos under 15 seconds do 31% worse on Instagram.
Questions this report answers
Yes. Brand videos with someone speaking get 18% more engagement on YouTube than the same brand's videos without a voice, 14% more on TikTok and 11% more on Instagram. Videos with music only do 10 to 15% worse on those three networks. LinkedIn shows no difference, and on X videos with a voice do 6% worse.
Yes. Brand videos with captions or subtitles on screen get 12% more engagement on YouTube than the same brand's videos without them, 9% more on Instagram and TikTok, and 5% more on Facebook. LinkedIn shows no difference.
With a bold claim or a direct statement. That hook gets more engagement than the same brand's other videos on all seven networks we measured. Opening with a question, the most common hook brands use (23 to 29% of videos), gets less engagement on all seven.
Yes. On Instagram, brand videos get 20% more engagement than the same brand's images, and two in three brands see it. Video also does better on Facebook (+19%) and much better on TikTok, where videos get nearly twice the engagement of photo posts. On X, videos get 36% less engagement than images.
Yes. Facebook image posts with a link in the caption get 20% less engagement than the same brand's image posts without one, and two in three brands see the drop. Facebook video posts with a link do 9% worse.
At the end. Brand videos with a call to action at the end get 11% more engagement on YouTube than the same brand's other videos, 8% more on Instagram and 4% more on TikTok. A call to action at the start or throughout the video doesn't help.
Data source
Anonymized, aggregated data from Vista Social. The report covers images and videos uploaded to Vista Social and published from August 1 to October 31, 2026, on brand profiles connected to Vista Social, with engagement and impressions as reported by each network.
Included: organic posts with exactly one image or one video, with impressions reported. Excluded: carousels and multi-image posts, boosted posts, Stories, competitor profiles, and networks that don't report impressions per post. After those rules, the data covers 1.2 million posts (488k images and 707k videos) from 22k brand profiles managed from 4.9k accounts.
How content is labeled
Vista Social runs every uploaded image and video through AI vision models, which describe the scene, mood, faces, text, people on camera, soundtrack, opening hook, call to action and other properties. The labels in this report are model labels, not human coding. We grouped free-text labels (for example, many ways of describing a question hook) into the categories shown, and left the rest as "other". Caption properties (length, hashtags, emoji, questions, links) are counted directly from the caption. No captions, images or videos were read by people for this report.
Metric and comparison
The metric is engagement rate by impressions for each post: engagements divided by impressions. For each brand profile, network and media type, we compared the median engagement rate of posts with a feature against posts without it, and report the median of those ratios across profiles. A profile counts toward a comparison only if it has at least 10 posts of that media type on that network, and at least 5 posts on each side of the comparison. Categories with more than two values (for example video length) are compared against the rest of the same brand's posts.
Privacy
No figure describes a single customer. A number is published only when it rests on at least 50 brand profiles from at least 10 Vista Social accounts. The analysis used only labels and counts. No captions, media or links left our systems.
Limitations
- Correlation, not causation. Comparing a brand with itself removes the effect of brand size, industry and audience, but not everything. A brand's longer videos may also be its better-planned ones.
- AI labels make mistakes. The models are good at clear-cut questions (is there a face, is someone speaking) and less reliable on judgment calls like mood. Results for clear-cut features are the most solid.
- Three months of data. Seasonal content (back to school, holidays) is not fully represented. The 2028 edition will cover a full year.
- Results collected in early November 2026. Posts from late October had about a week to collect engagement and impressions, so their totals may still grow a little.
- Single-item posts only. Carousels, which do well on Instagram and LinkedIn, are not covered.
How to cite this report
Vista Social. What Makes Social Media Content Perform, 2027 Report. November 2026. https://vistasocial.com/research/social-media-content-performance/
This report is published under a Creative Commons Attribution 4.0 license (CC BY 4.0). Charts, tables and figures may be reused with credit to Vista Social and a link to this page.
For press questions or custom cuts of the data, contact the Vista Social team at research@vistasocial.com.