Artwork is the most common reason a release gets held up, and it is almost never because the art was ugly. It is because it was the wrong size, or it had a web address on it, or it used a photograph the artist did not have the right to use. The aesthetic part is yours. The part below is just rules, and the rules are short.
The specifications, first
Make it 3000 × 3000 pixels, square, JPG or PNG, sRGB colour, under 4 MB. That single set of numbers satisfies every distributor and every store, and using it means never thinking about this again.
Spotify itself will accept anything from 640 × 640 up to 10,000 pixels square, as TIFF, PNG or JPG in sRGB. The reason to use 3000 anyway is that other stores are fussier and your distributor delivers one file to all of them.
Two details that cause real rejections: no embedded colour profile beyond sRGB, and no upscaled images. Taking a 500-pixel image and enlarging it to 3000 is detectable and gets refused. Start large or start again.
What gets artwork rejected
The list is short and almost entirely about text and ownership.
No web addresses, no social media handles, no email addresses. The stores will not carry artwork that advertises somewhere else.
No pricing, no 'out now', no 'available on iTunes'. Anything that reads as promotional copy rather than as a cover.
No blurry or pixelated images, which in practice means no upscaling and no screenshots.
No logos, trademarks or photographs you do not have the right to use — including record label logos, brand marks, and famous people's faces.
Text on the cover should generally be the artist name and the release title, and even those are optional. Anything beyond that invites a question.
Tools that cost nothing
Canva has a no-cost tier that will do this comfortably. Make a custom 3000 × 3000 canvas rather than using a template sized for something else, and check what licence any element you drag in actually carries.
Photopea runs in a browser, opens Photoshop files, and costs nothing beyond tolerating adverts. If you have ever used Photoshop, you already know how to use it.
GIMP is open-source, installs on everything, and is genuinely powerful once you get past the first hour of it being unfamiliar.
Any of the three will do the job. The choice between them is about what you are willing to learn this week, not about capability.
The licensing trap
This is the one that catches people, and it catches them after release rather than before.
'Free to use' is not one thing. An image can be free for personal use and not commercial use. It can require attribution you cannot put on a cover. It can be public domain in one country and not another. And a streaming release is commercial use — it is commercial even if you never sell a copy and never make a penny, because the work is being distributed through commercial channels.
So for any image you did not make: find the licence, read whether commercial use is permitted, and read whether attribution is required. Public domain sources and permissively licensed libraries exist and are genuinely usable — but check each image rather than trusting the site's front page.
The safest artwork is a photograph you took or a drawing you made. It is also usually the most interesting.
The only design advice that matters
It has to work at the size of a fingernail. Nearly everyone will see your cover as a tiny square in a queue on a phone. Shrink your artwork to about 50 pixels and look at it. If it becomes a grey smudge, it does not work, no matter how good it looks at full size.
That one test kills most of the usual mistakes: thin fonts, six colours, small text, a photograph with no clear subject. High contrast and one strong idea survive being shrunk. Detail does not.
On generated images
Stores currently accept AI-generated artwork and it is a real option when there is no money at all. Two things worth knowing before you use it. Work generated purely by a machine may not be protectable by copyright in the US, which means you may not be able to stop somebody else using your cover. And generated art carries the visual signature of the year it was made, which dates a record faster than almost anything else.
Neither is a reason not to. Both are reasons to treat it as the temporary answer it probably is.
Shrink it to fifty pixels and look at it. That one test catches nearly every cover-art mistake before a store does.
Three pages on this site that carry it further: the rest of what the stores need, choosing who delivers the file, and where artwork sits in the release sequence.
Where This Sits in the Release
Artwork has a deadline further back than most people realise, because the stores need it before the release date rather than on it. The start-here page lays out the whole road in order.
