Keywording a science or lab photo works best when you describe what is physically in the frame before you describe what it means. A pipette, a gloved hand and a microplate are facts a reviewer can verify; cancer research is an interpretation the picture cannot prove, and agencies reject metadata that does not relate to the depicted subject.
That distinction is most of the answer to how to keyword science stock photos well. It is a discipline problem rather than a vocabulary problem: the literal layer has to be complete, the conceptual layer has to stay inside what a buyer could reasonably read from the image, and anything a reviewer cannot see has to come out before submission.
What makes a lab frame different from any other set
Three things, and all three land in the metadata. Real equipment carries legible brand names. Real clinical and research settings carry readable identifiers on labels, forms and screens. And the subject matter invites claims — a researcher at a bench looks like it is about something, and the temptation is to say what.
None of that is a keywording skill in itself, which is why the permissions and release side of this category is worth settling separately — the tooling and restrictions for science and lab content cover what reaches a reviewer before metadata is even read. What follows assumes the frame itself is clear and the releases are in hand.
The literal layer: name the instrument, not the science
Start with the objects. Micropipette, microplate, petri dish, cryovial, microcentrifuge, burette, fume hood, nitrile glove, white coat, safety goggles, bench. Then the action: pipetting, loading, weighing, examining, labelling. Then the setting: laboratory, cleanroom, research facility, workbench. Shutterstock's contextual metadata standards state that trademarks and brand or product names are not allowed in commercial content metadata, so the generic instrument name is the only option — and it happens to be the term buyers type anyway.
This layer is where most science submissions are thin. A contributor who shot ten careful frames of a PCR setup will often tag all ten with the same eight generic terms, when each frame has different visible objects. The literal layer is the only part of the metadata that is cheap to get right and impossible to argue with, so it should be the longest part.
The same lab frame, keyworded two ways
A pipetting shot tagged with brand names and unprovable claims against the same frame described by what is visibly in it. Brand and product names are banned in commercial metadata, and metadata that does not relate to the depicted subject matter is a documented rejection ground on Shutterstock.
Latin names are the one exception agencies write down
Shutterstock requires metadata to be in English, and then names its exceptions explicitly: scientific Latin names of plants and animals, place names, and foreign terms commonly used in English. For specimen, botany and microbiology work that exception is the whole point of the category. Escherichia coli, Arabidopsis thaliana, Drosophila melanogaster belong in the keyword set.
They belong alongside the English terms, not instead of them. A buyer searching for a fruit fly image will type fruit fly far more often than the binomial, and a buyer who needs the binomial needs it exactly. Carrying both costs two keyword slots out of an allowance that runs to fifty, which is the cheapest trade in this whole process.
The conceptual layer buyers actually search
Buyers of science imagery are rarely scientists. They are designers at pharmaceutical agencies, editors at health publishers, product marketers at biotech firms and presentation designers at universities, and they search in the vocabulary of their brief: innovation, research and development, biotechnology, healthcare industry, clinical trial, laboratory testing, scientific discovery, teamwork, expertise.
Those terms are legitimate because they describe a category the image plausibly belongs to, not a claim about what is in the tube. The line between the two is sharper than it looks. Biotechnology research describes a field and survives review; gene therapy trial asserts a specific procedure, and unless the frame visibly shows it, that is metadata which does not relate to the depicted subject.
The same test applies to anything that reads as a diagnosis or an outcome. Clinical frames carry a stricter version of this, and the healthcare keywording rules are worth reading alongside this one if your lab work crosses into patient-facing settings.
Two commercial terms earn their place in almost every lab frame and are routinely left out: copy space and close up. Science imagery is bought to sit behind headlines and beside body copy, so a buyer filtering for negative space is a buyer with budget. If the frame has a clean area, say so.
How many keywords, and which ten matter
Shutterstock sets a floor of 7 keywords and a ceiling of 50. Adobe takes a different approach: its contributor portal suggests a title and up to 25 keywords, and Adobe's own keywords and metadata guidance says the first 10 keywords listed are prioritised in search results, and asks contributors to reorder them so they reflect importance. For a lab frame that means the top ten slots are a budget, and they should be spent on what is unmistakably visible.
A practical split for a detailed laboratory image: roughly the first ten keywords on the dominant subject and action, the next ten to fifteen on secondary objects, setting and composition, and the remainder on the conceptual layer. Twenty-five to thirty-five total is comfortable on a frame with this much visible detail. Padding to fifty with synonyms of the same idea adds nothing a search engine reads as new.
Which ten keywords would you put first?
Adobe Stock's contributor portal prioritises the first 10 keywords in search results and asks contributors to reorder suggestions so they reflect importance. These terms come from the review-safe set above; drag the ten that describe what is unmistakably in the frame into the priority slots.
Writing the description as a factual sentence
Shutterstock asks for a factual sentence that describes the subject matter, and allows up to 2048 characters for it. That is far more room than anyone needs, and the common mistake is to use almost none of it — pasting the keyword list into the description field instead of writing a sentence. A reviewer reading a comma-separated list sees a metadata shortcut, and the field was never indexed as a list anyway.
Write it in the order a person would describe the picture: who, doing what, with what instrument, where. Researcher in nitrile gloves pipetting a clear sample into a microplate at a laboratory bench. No brand names, no diagnosis, no superlatives. Everything in that sentence is checkable against the pixels, which is the only standard the description has to meet.
What has to come out before you submit
Shutterstock prohibits sensitive, personally identifiable information in metadata and states plainly that case numbers do not belong in titles or keywords. Most categories never run into that rule. Research and clinical settings run into it constantly, because sample labels, requisition forms, screens and whiteboards carry readable identifiers, and a contributor transcribing what is in frame can walk straight into it.
The rest of the removal list is shorter: brand and product names, offensive or inaccurate terminology around identity, emojis, spelling errors and anything that reads as spam. Shutterstock also requires the metadata to be relevant and applicable to the depicted content, which is the catch-all that the unprovable claims fall under.
A repeatable order for a lab shoot
List the visible objects frame by frame, not set by set. Add the action and the setting. Add the Latin binomial where there is a species, with its English name beside it. Add the conceptual layer, and test each term against what the picture could prove. Order the top ten for Adobe. Write the description as one factual sentence. Strip identifiers, brands and superlatives last, because that is when they are easiest to see. The Shutterstock keywording rules in full are worth a read before a first science batch.
Done by hand, that order takes a few minutes a frame and most of it is mechanical. The literal layer in particular is pure transcription of what is in the picture, and the per-agency differences — Shutterstock's floor of 7 and ceiling of 50 against Adobe's 25 suggestions with the first 10 weighted — are the kind of thing worth handling once rather than per upload.
Rastock AI sits on that part of the job: it reads each frame, writes titles, descriptions and keywords against each agency's own rules — including banned and mandatory terms — and delivers to 10+ agencies over FTP and SFTP, with status tracking and rollback if a batch goes wrong. It takes no revenue share and claims no ownership of the files, and IPTC and CSV export stay open. There is a 14-day free trial with no card required; the feature list and the pricing are both public.
Frequently asked questions
How many keywords should a science or lab photo have?
Shutterstock requires a minimum of 7 keywords and allows up to 50, so a detailed lab frame has room for 25 to 35 without strain. Adobe Stock's contributor portal suggests up to 25 and states that the first 10 keywords are prioritised in search results, which makes the order of those ten matter more than the total count.
Can I use Latin species names in stock keywords?
Yes. Shutterstock requires metadata to be in English but carves out explicit exceptions for scientific Latin names of plants and animals, place names, and foreign terms commonly used in English. For specimen and biology frames that exception is the point: Escherichia coli or Arabidopsis thaliana can sit alongside the English terms rather than instead of them.
Why do lab photos get rejected on metadata alone?
Usually because the metadata names something the frame cannot show. Shutterstock rejects content under Metadata - Commercial or Editorial Requirements when the metadata does not relate to the depicted subject matter, and a tag like cancer research on a generic pipetting shot is exactly that. Brand names on visible equipment are a second common cause.
Can I name the equipment manufacturer in the keywords?
Not in commercial metadata. Shutterstock states that trademarks and brand or product names are not allowed there, which rules out the centrifuge, pipette and tube brands that are often legible in a real lab. Describe the instrument generically instead: microcentrifuge, micropipette, microplate, cryovial. The generic term is also what buyers search.
What should the description say for a laboratory image?
Shutterstock asks for a factual sentence describing the subject matter, with a maximum length of 2048 characters, so there is no reason to compress it into a keyword list. Write what is visibly happening, in order: who, doing what, with what instrument, where. Save the interpretation for the keyword layer, where it can be read as a concept rather than a claim.
Do I need personal details out of the frame and the metadata?
Both. Shutterstock prohibits sensitive, personally identifiable information in metadata and says not to include case numbers in titles or keywords. In a clinical or research setting that reaches further than most categories, because sample labels, requisition forms and monitors frequently carry readable identifiers that have to be cleared before the file is submitted.