Shutterstock publishes more keywording guidance than most agencies, but only a fraction of it is enforced by software. Three things are checked on every upload: keyword count between 7 and 50, English-only metadata, and a commercial title that is a complete sentence of at least five words. The rest is judged by a reviewer, or caught much later in a collection audit.
That split matters, because contributors tend to optimise for the machine and get caught by the reviewer. The hard limits are trivial to satisfy. The rules that actually cost people files, and occasionally accounts, are the ones written in prose: what counts as repetition, when a trademark is allowed, and what happens if you re-add a keyword that staff removed.
The hard limits Shutterstock states in writing
Shutterstock's contextual metadata standard sets a minimum of 7 keywords and a maximum of 50. Both numbers are absolute. There is no tier, no exception for video, and no benefit documented anywhere for reaching the ceiling. Keywords are defined as individual words or compound words that describe the subject matter, which means a phrase counts as one keyword rather than several.
Descriptions and keywords must be in English. The documented exceptions are narrow: scientific Latin names for plants and animals, place names, and foreign terms already common in English. A description can run to 2,048 characters, which no sensible title ever approaches, but a commercial title must be at least five words and must read as a sentence rather than a comma-separated list. Special characters are excluded.
One category is required and a second is optional. Location is optional for commercial work, though Shutterstock asks you to be specific if you supply it, preferring a neighbourhood or city over a country. The full field-by-field breakdown, including video-specific differences, sits in our Shutterstock metadata requirements reference.
Where the spam line actually sits
Keyword spam on Shutterstock has two definitions, and only one of them is the obvious one. Irrelevant terms are spam, which everybody expects. The second definition catches far more honest contributors: repeating the same word or word stem across multiple keywords is also spam, even when every one of those keywords is accurate.
Shutterstock's own not-allowed example is a car photo tagged car ride, car road, car drive, car move, car adventure, car trip, car journey, car vector, car vector illustration, car illustration. Every phrase there is arguably true of the image. The pattern is the violation. A second published example does the same thing with seamless pattern argyle, seamless pattern green, seamless pattern blue and so on.
Not every repeat is banned, though, and the published allowed list is the useful half. Kansas City, city street, street scene passes. So does ski lodge, ski resort, mountain, ski lift, snow, Austria. The difference is that each repeated stem is doing different semantic work rather than padding a list with grammatical variants of one idea. If you can remove a keyword and lose no meaning, a reviewer will read it as padding.
Titles carry the same rule in a stricter form. Run, run, run is spam. So is run running runner, and so is the sentence Fishing scene with a fisherman fishing for fish with a fishing pole in a fishing boat with a fish on a hook, which Shutterstock cites verbatim as a rejection example. Runner running on city street is fine, because the repetition is grammatical rather than decorative.
The practical test is whether a stem appears because the sentence needs it or because you wanted the word counted twice. Adobe draws a similar line with different wording, which is worth reading alongside this one if you upload to both; we covered it in Adobe Stock keywording rules that actually get enforced.
Trademarks, and the one documented exception
Commercial content cannot carry trademarks in the title or keywords. That covers brand names and product names, and it is enforced at review rather than by a filter. You can photograph a generic smartphone that resembles a real one and sell it commercially; you cannot write iPhone or Apple in the metadata. Editorial and illustrative editorial content is exempt, because the whole point of an editorial caption is to name what is in the frame.
There is a published exception that surprises people. Shutterstock says Instagram may appear in a commercial title or keyword when it describes a look, such as an image processed to resemble an Instagram filter. The reasoning is that the word is functioning as a description of a visual style rather than a claim about the company. It is the only exception of its kind stated in the documentation, so it is not a template for using other brand names the same way.
What makes this rule difficult in practice is that Shutterstock publishes no list of restricted keywords. Its guidance says to infer the restrictions from the objects and subjects article and the known image restrictions list. So the rule is enforced strictly but expressed loosely, which is exactly the combination that produces surprise rejections on files a contributor thought were clean.
The rules that bite after publication
Approval is not the end of review. Shutterstock states that it regularly audits the collection for keyword spamming and for inappropriate or restricted terms. A file accepted two years ago can have keywords stripped today. Editing published metadata to introduce repetition is explicitly not allowed either, which closes the obvious workaround of tagging conservatively at upload and padding the file once it is live.
The most serious consequence in the whole ruleset attaches to what you do next. If a reviewer or an administrator removes keywords from an accepted file, do not add them back. Shutterstock says re-adding removed keywords may result in a warning, account suspension, or even closure, and directs contributors to contact support first to find out why the terms were removed. This is the one keywording rule with account-level consequences attached in writing.
For files that never got through in the first place, the rejection labels tell you which half of the rulebook you hit: Metadata - Commercial Requirements or Metadata - Editorial Requirements. Our breakdown of Shutterstock rejection reasons covers what each label means and what a resubmission needs to change.
Two rules high-volume contributors keep breaking
The first is reusing one title and keyword set across a series. It feels like an obvious time saver on a shoot that produced forty near-identical frames, and Shutterstock names it directly as a mistake: identical metadata makes the images compete against each other and against thousands of others, instead of each one reaching the buyer who wanted that specific frame. Bouquet of fresh red roses and Bouquet of hydrangeas outperform forty files called Flowers.
The second is copying another contributor's metadata word for word off a search result. Shutterstock acknowledges this is not plagiarism in a legal sense and prohibits it anyway. It is worth knowing because the practice is often recommended casually in contributor forums as competitive research, and the agency has drawn a line under it in its own visibility guidance.
Both problems come from the same place: treating a keyword set as a reusable asset rather than a per-file description. If you work at volume, the fix is structural rather than manual, and the principles carry across agencies; we set them out in our guide to structuring tagged keywords properly.
Editorial metadata is a separate rulebook
Documentary editorial captions carry two requirements commercial titles do not: the day, month and year the content was shot, and the geographic location where it was shot. Content taken before 2000 needs only the year or decade. Get either wrong and the file is rejected under Metadata - Editorial Requirements, which is also the label used when the date or place is simply inaccurate rather than missing.
The compensation is that editorial work escapes the trademark restriction. Naming a brand, a venue or a public figure in an editorial caption is expected rather than penalised. Contributors who shoot both types on the same day are the ones who get caught, because the two rulebooks pull in opposite directions and the metadata gets written in one pass.
What is folklore rather than rule
The most repeated claim about Shutterstock keywording is that position matters, that the first seven or the first fifteen keywords are weighted more heavily than the rest. Search that phrase and you will find it stated confidently on a dozen tool blogs, often with different numbers. It does not appear in Shutterstock's contributor documentation in any form. Adobe does publish a statement about keyword order; Shutterstock does not.
That does not prove ordering is irrelevant, only that nobody outside the company can source the claim. It matters because contributors spend real time reordering keyword lists on the strength of it, when the same time spent removing three padded terms would address a rule the agency actually publishes and enforces.
Two things Shutterstock does document are easier to act on. It asks for conceptual keywords alongside literal ones, so a photo of a laptop earns technology and communication as well as computer and desk. And it offers a keyword suggestion tool on the submit page, built to propose terms based on subject matter, which is a first-party signal about what the search index recognises.
A checklist that matches the documented rules
Count first. Between 7 and 50 keywords, and closer to the accurate number than the maximum. Read the list back and delete anything that repeats a stem already present without adding meaning. Read the title aloud; if it sounds like a list, rewrite it as a sentence of at least five words that answers who, what, where and why.
Then check the exclusions in one pass: no brand or product names on commercial files, no links, no camera settings, no emojis, no personally identifiable information, no case numbers meant for the reviewer. Confirm the metadata is in English apart from the documented exceptions. Set one category, and a second only if it genuinely applies.
Finally, look across the batch rather than at the single file. If ten images in the folder share a title or a keyword set, you have broken the rule Shutterstock says costs the most sales, and no individual file will show you that. This is the check that manual keywording almost never survives at volume, because the reviewer sees the batch and the contributor sees one row at a time.
Where this breaks in a multi-agency workflow
None of this is difficult for one file. It becomes difficult when the same image goes to several agencies, because the rules disagree. Shutterstock caps keywords at 50 and treats stem repetition as spam. Adobe caps them lower and publishes a position rule. Editorial requirements differ again. A single generic keyword set sent everywhere will be legal in some destinations and rejected in others, and the contributor only finds out days later, one rejection notice at a time.
The useful shape of a fix is per-agency rules applied before delivery rather than after rejection: a keyword count enforced per destination, banned and mandatory term lists you control, stem repetition flagged while the batch is still editable, and a way to roll back a batch that went out with the wrong profile. Speed is not the constraint here. Control is.
That is the part Rastock AI was built around. Metadata is generated against each agency's rules rather than one generic output, with banned and mandatory keyword lists you define, repetition checks before delivery, and a rollback if a batch goes out wrong. Delivery to 10+ agencies over FTP and per-file status tracking sit in the same place, so an image and its metadata never drift apart. There is no revenue share and no ownership claim on your work, and IPTC and CSV export stay open. The feature overview walks through the workflow, and pricing includes a 14-day free trial with no card required.
The short version worth remembering
Shutterstock's keywording rules are unusually well documented, and the enforcement is layered rather than uniform. A filter catches the counts. A reviewer catches the repetition, the trademarks and the list-shaped titles. An audit catches everything else, months or years later, and the only rule that can cost you the account is the one about putting removed keywords back. Write metadata that would survive all three passes and the ordering debates stop mattering.
Frequently asked questions
How many keywords does Shutterstock allow?
Shutterstock requires a minimum of 7 keywords and allows a maximum of 50. Both are stated in its contextual metadata standard and apply to photos, illustrations, vectors and video alike. Nothing in Shutterstock's documentation suggests that reaching 50 improves ranking, and pushing toward the ceiling raises the odds that one weak term reads as irrelevant to a reviewer. Accuracy is the constraint that matters, not volume.
What counts as keyword spam on Shutterstock?
Two things. Irrelevant keywords that do not describe the content literally or figuratively, and repeated words or word stems used to pad the list. Shutterstock's published example of the second is car ride, car road, car drive, car move and so on. Repetition is allowed when each term carries distinct meaning, as in Kansas City, city street, street scene. Spam is rejected at review and can also be removed after publication.
Can I use brand names in Shutterstock keywords?
Not for commercial content. Descriptions and keywords on commercial files cannot contain trademarks such as brand or product names. Editorial and illustrative editorial content may use them. Shutterstock documents one narrow commercial exception: Instagram may appear in a title or keyword when it describes a visual look rather than the company. There is no published list of banned terms, so the restriction has to be inferred from the known image restrictions material.
What happens if Shutterstock removes my keywords?
Contact contributor support before touching the file. Shutterstock states plainly that adding back keywords a reviewer or administrator removed may result in a warning, account suspension, or closure. Removal usually means the term was judged irrelevant, spammy or restricted, and re-adding it reads as deliberate evasion rather than an oversight. Ask why it was removed first, then correct the metadata accordingly.
Does the order of keywords affect Shutterstock ranking?
Shutterstock does not say so anywhere in its contributor documentation. The claim that the first seven or first fifteen keywords carry extra weight circulates widely on tool blogs and forums, but it is not an agency statement. What Shutterstock does document is that unique, specific, non-repetitive metadata improves discoverability, and that reusing one keyword set across a series hurts it. Treat ordering as untested folklore, not policy.
Why was my Shutterstock file rejected for title or keywords?
The most common documented causes are a title shorter than five words, a title written as a comma-separated keyword list, repeated words or stems in either field, keywords that are irrelevant or incomplete, and trademarks on commercial content. Editorial submissions are additionally rejected when the caption is missing the date or the geographic location. Shutterstock labels these rejections Metadata - Commercial Requirements or Metadata - Editorial Requirements.