• Add a minimum of 10 and no more than 50 keywords. 
• Keywords are comma-separated terms, like "alligator," "party," and "scream."
• You can use single words and common phrases. If it has a Wikipedia page, it’s probably acceptable to add (e.g. “Mother’s Day” is a common phrase you can use, but “happy mother” should be added as two separate keywords). 
• Add only the root version of each noun or verb (e.g. “laugh” and “cat,” not “laugh,” “laughing,” “cat,” and “cats.” Our dictionary takes care of this.
• Only copy and paste keywords to relevant assets, not across the entire series. 
• The most common mistake is stretching the possible concepts that a photo or video expresses. If the context is not visible, leave it out.

Don't Include

• Terms describing the model’s ethnicity, gender, or age. These terms are added automatically from the model release.
• Synonyms (e.g., “job,” “occupation,” and “career”). We make sure the terms clients use are connected behind the scenes.

Who 

The first rule of describing people is to ask. Be honest with your models about signing a release for stock and how the content might be used, and ask how they want to be represented.

Second, Instead of focusing your keywords on describing models’ diversity, differences, or physical characteristics, focus your metadata on what people are doing, the mood, or the concept. 

Finally, be considerate and accurate. In particular, metadata that describes aspects of gender expression or identity, sexual orientation, disabilities, or demographic status should only be added when confirmed with the model and relevant to the content.

Please do not make assumptions or embellish.

• Avoid subjective language, e.g. normal, weird, beautiful, ugly, healthy, unhealthy, cute, gross, etc. 
• Age/ethnicity/gender: Don't add these as keywords when a model release is attached. We take care of it for you.
  
  • Gender identity: You can add more specific descriptors of gender identity if accurate and confirmed with the model; for these terms, Stocksy follows the GLAAD Media Reference Guide for language
• Bodies: Use respectful and accurate descriptive language (discuss with your model beforehand)
  
  • Plus size models: use plus size or curvy
    • Do not use: any other terms to describe someone’s size or weight
  • Disabilities: Use the word “disability,” the medical name of the disability (if your model is comfortable sharing this information), and the correct name of any visible accessibility devices. All terms must be accurate and confirmed with the model.
• Body parts: mention specific body parts only when they’re an important part of the content (e.g. face, hand, skin)
• Hair: use hair when it’s an important element and hairstyle if a particular style is featured
  • blonde, brunette, redhead
  • specific names of hairstyles
• Body modification: tattoo, piercing, other body mod names 
• Clothing: include specific items if noteworthy (e.g. hat, sunglasses, wedding gown). No need to describe every piece of clothing on every person. 
  • Do not use: fashion unless it’s a fashion series
• Relationships: include relationships when context is visible (e.g. a portrait of a senior woman alone should not be keyworded “grandmother”)
  • Reminder: we take care of synonyms for you, so just pick one term for each role (e.g. mother, not mother, mom, mum, mommy, and mama)
  • Couples: use couple and the most accurate terms for their role (not all possible terms)
    • Husband, wife, boyfriend, girlfriend, partner 
    • LGBTQ and for any other terms, refer to the GLAAD Media Reference Guide
      • Don’t use: homosexual
    • Bride, groom (if there is visible wedding context)
  • Family: use family and the most accurate term for the relationship (not all)
    • • Parent
        • Mother, father
        • Child
      • Grandparent
        • Grandmother
        • Grandfather
        • Grandchild
      • Sister, brother
      • Aunt, uncle, niece, nephew, cousin
      • Pet
  • Friend
  • Coworker 
• Roles: include roles when context is visible (e.g. farmer, businesswoman, babysitter)
• Facial expressions: 
  • Looking at camera, looking away, looking up
  • Smile, laugh, frown, grimace 
• Emotions

What 

• Broad subjects (food, business, family)
• Specific subjects (sandwich, cat, toothbrush)
• Actions (eating, playing, typing)
• Concepts (togetherness, lifestyle, friendship)
• Moods (happy, depressed, bored)

When

• Time of day: sunrise or sunset (not both), day, night, evening
• Season: autumn, winter, spring, summer

Where

• Country, province/territory, city, neighbourhood, if any location context is visible
• Indoors or outdoors
• Specific locations
  • e.g. home, park, restaurant, school, office, beach
• Specific rooms
  • e.g. kitchen, closet, break room, office, garage
• For studio work, use studio

indoors, restaurant, day	outdoors, ocean, day, sunny
indoors, home, kitchen, day, morning	outdoors, street, night, direct flash
studio, hard light	indoors, store, warehouse

How

• Lens: wide angle, ultra wide angle, close-up, macro, fisheye, tilt-shift
• Angle: pov, birds-eye view, overhead, flat lay
• Any notable techniques (double exposure, timelapse, long exposure, motion blur)
• Lighting (if notable): dark, sunlight, direct flash, natural light, hard or soft light (for studio work), lens flare

When should my keywording be done?

You can submit content without completed metadata if you do a preselect. However, you should finish keywording before your content goes live on the site.

We don't know when Google will index your new portfolio additions, and you should complete your metadata before we send to Adobe when it is published.

How do I select keyword definitions?

We use a Controlled Vocabulary (CV), which is a dictionary and thesaurus bundled together. Select a definition when available to unlock the benefit of synonyms:

• Green terms are good to go, no action is needed
• Grey terms need a definition selected
• Blue terms have no available definitions (contact Support if you think we should add these to our dictionary)

You may also see the process of defining keywords referred to as "disambiguation."

Synonyms

Synonyms allow clients to find your assets when searching for keywords or phrases you didn't enter directly. For example, if your content has the green keyword "business," the CV will automatically match it to words like "occupation," "job," "career," and "employment," so you don't have to add all these manually. 

It also deals with regional spelling differences like "colour" vs. "color." 

If you don't select a definition, your asset will appear in search results for the term you entered but won't appear in searches for the associated synonyms.

Stemming

Stemming is how our CV deals with verb conjugations and pluralization, using the root of a word and modifying the letters on the end. The CV will automatically recognize the verb "walk" as meaning "walks," "walked," and "walking," and a noun like "bird" also to mean "birds." Stemming means you don't need to add multiple versions of each term. 

However, you still need to add separate verbs and nouns. For example, you may expect "dance" to also mean "dancer," but because "dancer" is a noun and not a verb, they both need to be entered separately.

Using Robokeywords

Underneath your keywords in Image Manager, you'll see a section called "Optional Keywords." You can add these one at a time or in bulk by pressing "add multiple."

Be cautious: this is a machine learning tool and is not always accurate. Only add the suggestions that are relevant to your content. We are looking to improve this tool in the future.