You built the ad on a monitor. It was probably a good monitor, twenty-four inches or better, sitting on a desk in decent light. The image looked sharp. The headline fit. You read the whole thing back, decided it said what you wanted it to say, and hit publish.
Then LinkedIn served it to someone standing in a parking lot, holding a phone at arm's length, thumb already moving.
Those are not the same ad. Not because anything went wrong, and not because LinkedIn did anything to you. The feed simply renders differently on a narrow screen than it does on a wide one, and almost every decision that makes an ad readable on the phone is a decision you have to make before you upload it. LinkedIn publishes exactly where the text gets cut and exactly which shapes are allowed to appear where. The numbers are public. Almost nobody building the ad has read them.
This is the part of LinkedIn advertising that has nothing to do with strategy. Your targeting can be perfect and your offer can be right and the ad can still land as a half-sentence over a picture too small to read. We have written before about the ad that says nothing even after the targeting is fixed. This is the layer underneath that one. Not what the ad says. What survives of it by the time it reaches a thumb.
The First 150 Characters Are the Whole Ad
The block of text above your image is called introductory text. LinkedIn will accept three thousand characters in that field. It will show about a hundred and fifty of them before it collapses the rest behind a "see more" link.
That is LinkedIn's own number, published in their specifications, and it is the single most useful fact in this article. A hundred and fifty characters including spaces and punctuation. Two sentences if they are short ones. If the reason someone should care about your ad arrives in sentence four, it arrives behind a link that most people will never tap, on a screen where tapping means committing to something.
So the job is not to write shorter copy. Write as much as the ad deserves. The job is to make sure the first hundred and fifty characters can stand alone as a complete thought, and that everything after them is a bonus for the small number of people who expand it. Lead with the specific problem, or the specific claim, or the number. Not with a warm-up. Not with "In today's competitive B2B landscape." That phrase alone is forty-eight characters of nothing, which is roughly a third of everything you are guaranteed to be shown.
There is one exception, and it catches people out because it runs the opposite way. If you are running a carousel ad, the introductory text does not cap at three thousand characters. It caps at two hundred and fifty-five, total. Not two hundred and fifty-five before truncation. That is the whole field. If you have written a long intro for a single image ad and you are moving that copy across to a carousel, it will not fit and you will find out at upload.
Seventy Characters, and LinkedIn Will Not Promise You Even That
Under the image sits the headline. LinkedIn accepts two hundred characters and recommends seventy to avoid truncation, and the recommendation is the real limit.
What makes the headline trickier than the intro text is that seventy is not a hard edge. In their own documentation for document ads, LinkedIn says the shortening of the headline "can vary slightly depending on the device your ad is viewed on, and the type of characters you use." That second half is the part worth sitting with. Character type matters, because the cut is really about pixel width, not letter count. A headline in wide capital letters eats the available space faster than the same number of narrow lowercase ones. A headline full of capitals and wide letters like W and M can vanish sooner than a longer headline made of thin ones.
The practical version: treat seventy as your ceiling, aim under it, and do not put anything load-bearing in the last ten characters. If your headline only makes sense once the reader reaches the final word, you have built a headline that depends on a promise LinkedIn explicitly declines to make.
There is a third text field on single image ads called the description, and it has a hundred-character recommendation of its own. Before you spend any time on it, know when it actually appears. LinkedIn says the description shows "when your image is less than 200 pixels wide or if your ad is displayed beyond the feed through the LinkedIn Audience Network." In the ordinary case, running a properly sized image in the LinkedIn feed, nobody sees it. Fill it in, keep it honest, and do not put anything there that the ad needs in order to make sense.
LinkedIn Recommends a Shape That Cannot Run on a Desktop
Now the part that surprised me, and the reason this article is more than a list of numbers.
LinkedIn currently supports three shapes for a single image ad. Landscape at 1.91 to 1, which is 1200 by 628 pixels. Square at 1 to 1, which is 1200 by 1200. And vertical at 4 to 5, which they give as 720 by 900.
Vertical is the one they recommend. Here is the recommendation in full, from their specifications page, and read the second half carefully:
"We recommend using vertical aspect ratios for single image ads; however, paid vertical assets will only serve on mobile devices."
Vertical wins the mobile feed. It takes more vertical space in a scrolling column, which is exactly why it performs, and it is the shape most consistent with how people actually hold a phone. And it does not run on desktop. Not "runs smaller on desktop." Does not serve there.
For a large advertiser with a creative team, that is a non-issue: produce the vertical asset for mobile, produce something else for desktop, let the platform sort it out. For a small business that made one image because making one image is what the budget and the afternoon allowed, it is a real decision with a real cost, and LinkedIn's recommendation does not mention the cost in the same breath.
Square Is the Only Shape That Refuses to Pick a Side
So who is right?
AJ Wilcox has tested more LinkedIn creative than almost anyone alive, and his position is blunt: "We've tested a lot of different sizes and my recommendation is always to go with square whenever possible. It generally gets you the best performance across devices and ad formats."
Mitchell Gould, looking at the same problem from the legibility side, lands in the same place. Watching a wide image render on an iPhone, he notes that a wide aspect ratio "is going to limit the size of the ad on mobile as well," which shrinks any text baked into it, and suggests turning it into a one-to-one square, which "shows up good on mobile and desktop."
Here is our read, and it is a genuine call rather than a hedge.
If you are running one creative, make it square. 1200 by 1200. Square is the only one of the three shapes that renders well in both places without being locked out of either. Landscape gives up too much size on the phone, where most of your audience is. Vertical gives up the desktop entirely. Square is a compromise in the honest sense: slightly less commanding on mobile than a 4:5, slightly less wide on desktop than a 1.91:1, present and legible in both.
If you can produce two, then take LinkedIn's advice properly. Run the 4:5 vertical for its mobile advantage and a square alongside it so desktop is still covered. That is the version of their recommendation that actually works, and it is only available to you if you can make two images instead of one.
One more number belongs here, because it decides things quietly. LinkedIn says images less than 401 pixels wide display as a thumbnail. Upload something small and your ad does not simply look lower quality, it changes format and renders as a little thumbnail beside text. Their recommended dimensions are all comfortably above that line, so this is a rule you break by accident, usually by pulling an image off a website where it had already been compressed for page speed. Start from the recommended size and scale down. Never up.
A Carousel Card Is 312 Pixels Wide. Your Text Will Not Survive It.
Carousels get recommended a lot and they are genuinely useful, but there is a number in the specification that changes how you should design them, and it is easy to miss.
LinkedIn asks for carousel card images at 1080 by 1080. Then, in the same document: "Images will be scaled to 312 x 312px."
Three hundred and twelve pixels. That is roughly the size of a large app icon. Whatever you designed at 1080 is going to be displayed at less than a third of that, and every piece of text you put on the card shrinks by the same factor. A headline baked into the image at a comfortable size on your screen becomes unreadable. A logo with a tagline under it becomes a logo with a smudge under it. A chart becomes texture.
Card headlines are capped at forty-five characters, which is less than most people expect and less than half the single-image headline allowance. Between the 45-character headline and the 312-pixel card, the carousel is the format with the least room for words anywhere in LinkedIn's lineup.
So design carousels the way you would design road signs. One idea per card. Very few words, very large. If you want to check a card before you build nine more, shrink it to about a third of its size on your own screen and look at it from where you normally sit. If you cannot read it, neither can they. The format menu is worth understanding on its own terms, and we have gone through which of LinkedIn's ad formats actually earn their place separately.
Document Ads Break the One Thing You Built Them For
Document ads are one of the cheapest ways to put something genuinely useful in front of a professional audience, and they carry three constraints that quietly undo a lot of the work people put into them.
The first is the one that matters most. From LinkedIn's specification: "Hyperlinks in the document will only be clickable if the member downloads the document."
Read that against how these usually get built. Someone makes a nice ten-page PDF, puts a "Book a call" link on the last page, runs it as a document ad, and watches people flip through all ten pages in the feed without a single click. The link was never clickable. In-feed, the document is a viewer, not a browser. If you want action from a document ad, it comes from the ad's own call-to-action button or from a lead form attached to the campaign, never from a link inside the file.
The second: "Videos and other animations in documents aren't supported and will display as static images instead." An embedded demo clip becomes a frozen frame, usually a badly chosen one, sitting in the middle of your document looking like a mistake.
The third is small and slightly funny. By default, the headline of a document ad is pre-filled with the file name. Which means that unless someone changes it, a real advertiser somewhere is running an ad headlined `Whitepaper_v4_FINAL_reviewed.pdf`. Change it.
Worth knowing too: the lead generation objective supports PDFs only, so a PowerPoint that works fine for a traffic campaign will be rejected the moment you attach a lead form. And if your PDF has layers or mixed page sizes, flatten it and normalize the pages first. Where a document ad should send people once they are interested is a separate question, and we have argued that most LinkedIn clicks should not be going to a landing page at all.
Look at It on the Thing It Will Be Seen On
Everything above is checkable before you spend anything, and the check takes about two minutes.
Campaign Manager has a preview. Mitchell Gould's walkthrough puts it behind the three-dot menu on the ad, where you choose View and then the mobile feed option, and the point of it is not to admire your work. It is to read your own ad the way a stranger will, with the truncation in place and the image at its real size. If the preview shows your intro text ending mid-thought, the fix is to rewrite the first hundred and fifty characters, not to hope.
The cruder test catches things the preview will not. Put the ad on your actual phone, hold it at the distance you normally hold it, and look at it in daylight rather than at a desk. Small text that is merely tight indoors becomes invisible outdoors, and a good share of LinkedIn browsing does not happen at a desk. Gould's own conclusion after going through a set of live ads is the right instinct: if you cannot read the font, you are getting a worse click-through rate than you could, and the fix is usually fewer words at a larger size rather than a better image.
A short list to run before anything goes live:
- Does the first sentence still make sense at a hundred and fifty characters? Count it, do not estimate it.
- Is the headline under seventy, with nothing critical in the last few words?
- Is the image square unless you have deliberately chosen to trade away desktop or mobile?
- Is it at least 401 pixels wide, and ideally at LinkedIn's recommended dimensions?
- On a carousel, does each card survive being shrunk to a third of its size?
- On a document, is the only path to action the ad's own button rather than a link inside the file?
None of this makes a bad ad good. A weak offer rendered perfectly is still a weak offer. But this is the cheapest category of improvement available on LinkedIn, because it costs nothing but attention and it applies to every impression you have already paid for. The same failure happens one click later on the page itself, which we covered in what your site looks like to the customers who are actually on their phones.
We spend our days inside accounts like this, and creative that was built for the wrong screen is one of the most common things we find. If you would rather hand the whole thing over, we run LinkedIn ads for small businesses and this is part of what that involves. If you would rather do it yourself, open your ad on your phone this afternoon. The list above is the whole method, and it is yours either way.
