LinkedIn multi-image posts: the format most guides skip

A LinkedIn multi-image post is several photos attached to one ordinary update: up to 20 of them, 5MB each, shown as a grid preview that opens into a full viewer. It is not a carousel, it needs no design tool, and it takes minutes rather than an hour to make. Socialinsider's benchmark of 1.3 million posts ranks it second among all formats at a 6.45% engagement rate, behind only native documents at 7.00%. Our own dataset cannot confirm or contradict that, because it does not distinguish multi-image posts from single-image posts, and we would rather say so than borrow a number.

What is a LinkedIn multi-image post?

You write an update, click the image icon in the composer, and attach more than one photo. That is the whole feature. LinkedIn arranges the images into a preview in the feed, with a counter showing how many are hidden, and tapping opens a viewer where readers move through them one at a time.

There is no separate button, no special post type, and no file to prepare. It is the same composer flow as a single image post with more files selected. That simplicity is the format's main argument: everything else on LinkedIn that produces a swipe requires design work, and this does not.

What you give up in exchange is structure. There is no page counter, no document title field, and no strong sense that the images form a sequence. LinkedIn presents them as a set. If your content is a set, that is exactly right. If your content is an argument that proceeds in order, it is the wrong container, and we go through that boundary in our piece on document posts versus carousels.

How many images can you put in a LinkedIn post?

Twenty. From LinkedIn's help page on sharing photos: “You can upload a maximum of 20 photos in a multi-photo post.” The same page gives the rest of the constraints.

Two of those deserve attention because they cause most of the problems people run into.

The aspect ratio range is narrower than it looks. 4:5 is the tallest allowed, which is portrait but not very portrait. A screenshot of a full phone screen is roughly 9:19.5 and will be cropped or rejected. If you are posting screenshots, crop them yourself so you control what gets cut rather than letting the upload decide.

Mixed aspect ratios look bad. LinkedIn will accept a landscape photo next to a portrait one, and the resulting grid is untidy in a way that reads as careless. Pick one shape for the whole set. This is the single cheapest thing you can do to make a multi-image post look deliberate.

Do multi-image posts actually perform well?

The best public evidence says yes, and it is not our evidence.

Socialinsider's LinkedIn benchmarks, built from 1.3 million posts across 16,645 business pages between January 2024 and December 2025, rank formats by engagement rate per impression like this:

The interesting row is not the second one. It is the gap between multi-image at 6.45% and single image at 5.30%, because those are the same content type with a different number of files attached. Whatever is happening there is a property of the format rather than of the subject matter.

Two caveats before you take that as settled. Their population is company Pages, not personal profiles, and Page audiences behave differently from personal followings. And their engagement rate is computed per impression, which is a different denominator from ours. Neither invalidates the ordering, and the ordering is the useful part.

What does our own data say about multi-image posts?

Nothing, and we are going to be plain about it rather than reaching.

Our cohort of 12,988 English LinkedIn posts from 65 creators records a media type for each post. One of those types is “image.” It contains 4,728 posts from 57 accounts, and it was 30.7% of our top-performing decile against 40.3% of the bottom half, with a median engagement rate of 0.33 per 1,000 followers against a cohort median of 0.40.

That bucket mixes single-image posts and multi-image posts together. The source data does not tell us how many files were attached, so we cannot split it. Any number we quoted for multi-image specifically would be invented, and inventing it would be worse than having nothing to say.

What we can say is that the combined image bucket is the one format that showed up more often among our worst posts than our best. The likeliest reading is not that images hurt. It is that attaching a stock photo is the default dressing on a low-effort post, so the image bucket collects a lot of filler that a text post would never have bothered to produce. That confound is the same one that shows up in every format comparison, ours included, and we work through it in our comparison of text posts and image posts.

If Socialinsider's split is right, then the multi-image half of our image bucket is performing better than the single-image half, and our combined number understates it. That is a plausible story and we have no way to test it. Take it as a reason to run your own test rather than as a finding.

When should you use a multi-image post instead of the alternatives?

The question that resolves almost every case: does the order carry meaning? If page three only makes sense after page two, you want a document post. If any order works, you want a multi-image post.

The row people get wrong most often is the seventh. Exporting slides as individual PNGs and attaching them as a multi-image post is a common workaround, and it produces a worse version of a document post: no page counter, no document title, and image compression on text that was meant to be crisp. If you designed slides, upload the file.

For the format with the strongest evidence behind it in our own data, native video was 26.3% of our top decile against 10.1% of our bottom half, across 1,613 posts from 46 creators. That is in our piece on LinkedIn video posts, and it is worth reading before you decide that photos are the answer.

What are multi-image posts actually good for?

Ten uses that hold up, roughly in order of how often they are worth doing.

1. Before and after. Two images, no caption gymnastics required. The single most efficient use of the format.
2. Event coverage. Four to six photos from a conference, with a caption about what you actually learned rather than who you met.
3. A process shown in stages. Photographs of real stages, not designed slides. A build, a repair, a rollout.
4. Screenshots of a thing working. Product changes, dashboards, a bug you fixed. Crop them to a consistent shape first.
5. The same data three ways. One chart per image, each answering a different question about the same dataset.
6. A team or a place. The format where personal photography beats stock by the widest margin.
7. Annotated examples. Four real examples of something done well, each with the annotation baked into the image.
8. A physical product from angles. Obvious and under-used by anyone selling something you can hold.
9. Whiteboards and notebook pages. Photographs of actual thinking. They look unpolished, which is the appeal.
10. A short receipts post. Three screenshots that prove a claim you made in the caption.

What they are not good for: anything that is really text. If your images are quotes in a nice font, you have made a document post badly. If you want slide-shaped content, our list of carousel concepts by goal covers what belongs there instead.

How does a multi-image post render in the feed?

Worth understanding, because it changes what you should put where, and no spec sheet covers it.

In the feed, LinkedIn shows a preview grid rather than the full set. The first image gets the largest cell, the next few get smaller ones, and if there are more than the grid holds, the last visible cell carries a counter. Tapping any of them opens a viewer where readers move through the images one at a time at full size.

Three consequences follow. First, the first image is a thumbnail and gets judged as one: reduced, next to text, in a moving feed. Second, images two and three are visible but small, so anything with fine detail in those positions is wasted until somebody taps. Third, anything past the grid is invisible until a reader commits, which means the back half of a twenty-image post is effectively unpublished for most of your audience.

The grid also differs between mobile and desktop, and the mobile version is the one that matters. Most of your audience is on a phone, the cells are smaller there, and text baked into an image at desktop size will be unreadable. The habit that fixes this is the same one we recommend for document posts: look at it on your phone before you publish, every time.

One practical note about scheduling tools. Support for multi-image posting varies between third-party schedulers, and some of them will silently post only the first image or convert the set into something else. If you schedule rather than post live, test the format once with a low-stakes post before you rely on it. Our rundown of LinkedIn scheduling tools covers what to check.

Should the photos be original or stock?

Original, and this is the format where it matters most, because a multi-image post is implicitly a claim that you have several things worth showing.

A stock photo says nothing. A stock photo repeated six times says nothing six times, and it actively signals that the post is filler with decoration attached. Our combined image bucket showing up more often among the worst posts than the best (30.7% of the top decile against 40.3% of the bottom half, across 4,728 posts from 57 accounts) is most plausibly explained by exactly this: the reflex of attaching a generic picture to a post that had nothing visual to say.

Where original photos come from, if you think you have none:

• Your camera roll from the last month. Most people have four usable photos of work they actually did and have never thought of them as content.
• Screenshots. A dashboard, a message thread with permission, a document with the sensitive parts removed. Crop consistently.
• Whiteboards, notebooks, sketches. Deliberately unpolished, and readers respond to them precisely because nobody could have bought them.
• Charts you made. Even a plain one from a spreadsheet beats an illustration of a person pointing at a graph.

The bar is not photographic quality. It is that the image could only have come from you. A slightly badly lit photo of the actual thing outperforms a perfectly composed image of a generic version of the thing, and it does so for the same reason first-person writing outperforms institutional writing in our hook data.

How should you order and choose the images?

The first image is doing a job the others are not. It appears in the feed preview at reduced size, next to a counter telling people how many more there are. Most readers decide whether to tap based on that one frame.

• Lead with the most legible image, not the best one. A beautiful wide shot that turns to mush in a grid cell loses to a simple one that reads instantly.
• Put the payoff second, not last. Unlike a document post, there is no page counter promising an ending, so people do not reliably reach image nine.
• Keep the set to five to ten. Twenty is the ceiling, not a target. Past about ten, the tail is unseen.
• Use one aspect ratio throughout. Covered above and worth repeating, because it is the difference between a set and a pile.
• Cut any image that is there for volume. The same discipline as slides. If deleting it loses nothing, delete it.

On accessibility: add alt text to every image. LinkedIn provides the field, almost nobody fills it in, and a multi-image post without it is a set of blank rectangles to anyone using a screen reader. If the substance is in the pictures, summarise it in the caption too.

How do you write the caption for a multi-image post?

The caption carries more weight here than on a document post, because there is no cover page stating the claim and no document title underneath. The images are evidence. The caption is the argument.

Keep the claim inside the fold. The visible portion runs to roughly 210 characters on desktop and about 140 on mobile, per AuthoredUp's character limit breakdown. Our hook data, which covers all 12,988 posts and so is on much firmer ground than any format-specific claim, points in a few consistent directions: saying “you” somewhere in the visible hook ran 47.5% among top-decile posts against 32.7% in the bottom half, and first-person openers ran 19.6% against 10.1%.

Image posts in our cohort (4,728 posts, 57 accounts) carried four or more hashtags 42.3% of the time, which is the highest rate of any format we measured and runs against the cohort-level pattern where heavy hashtag use appears in 41.9% of bottom-half posts against 25.9% of top-decile posts. One to three relevant tags at the end is plenty, and the full breakdown is in the hashtag analysis.

A caption structure that works: one line stating what the images show and why it matters, one line of context or credibility, one line inviting a specific response. Then stop. Everything else you were going to write belongs in a reply to the first person who asks. Our guide to how to end a LinkedIn post covers why a single specific ask beats a menu.

What are the common mistakes with multi-image posts?

1. Twenty images because twenty is allowed. The tail is unseen and the post reads as an album dump.
2. Mixed aspect ratios. The most visible unforced error in the format.
3. Slides exported as PNGs. Use a document post. You are losing the page counter and adding compression artefacts to text.
4. A weak first image. It is the thumbnail. Treat it like one.
5. No alt text. Free to fix, and it is the difference between a post being available to somebody and not.
6. Posting the photos instead of the point.A conference album with “great few days” on top is not a post, it is a caption looking for one.
7. Assuming you can fix it later.LinkedIn's own documentation notes that images cannot be resized or edited after posting. Check the set on a phone first.

How do you test whether multi-image works for your account?

Since we cannot tell you from our data, here is the test that will. It takes about six weeks.

1. Baseline first. Pull your last thirty posts and record impressions, reactions, comments, and saves. Use the median, not the average, because one outlier will distort the average badly.
2. Post six multi-image posts over six weeks. Same topics, same posting time, same caption style as your usual content. Six is not statistically impressive, and it is six times what most people try before forming an opinion.
3. Hold everything else constant. If you change the format, the cadence, and the hook style at once, you learn nothing.
4. Compare medians, then read the comments. If multi-image comments are more substantive than your usual ones, the format is reaching a more engaged slice of your audience, which matters more than the impression count.

For a scale to measure against, our whole-cohort benchmarks are a median of 0.40 engagement per 1,000 followers, 1.27 at the 75th percentile, and 5.95 at the top-decile threshold. The percentile tables are in our benchmark piece on what good looks like at your follower count, and the engagement rate calculator will put your own numbers on the same scale. Everything in our dataset is correlational rather than causal, and the cohort skews toward established creators with 1,000 or more followers, so treat it as a reference rather than a target.

So are LinkedIn multi-image posts worth using?

Yes, particularly if you have real photographs and no appetite for design work. LinkedIn multi-image posts take up to 20 images, need no file preparation, and rank second by engagement rate in the largest public benchmark available, at 6.45% behind documents at 7.00%. The format is genuinely under-used relative to how cheap it is.

What we will not tell you is that our own data confirms it, because it does not. Our image bucket mixes single and multiple images across 4,728 posts from 57 accounts, and no honest split exists. Use the external benchmark as a reason to try the format, use the order test to decide between multi-image and a document post, keep the set to five to ten photos in one aspect ratio, and let your own six-week comparison settle the question for your account.