What is an emoji?

An emoji is a small picture character encoded in the Unicode Standard with its own name and codepoint, drawn slightly differently by each operating system, used inside text to add tone, emphasis, or a quick visual shorthand that the words on their own would have to spell out.

What is an emoji?

Emojis are the small coloured pictures sitting inside ordinary text on a phone, a laptop, or a social post. The word is Japanese, a combination of e meaning picture and moji meaning character, and the resemblance to the English emoticon is coincidence. The Unicode FAQ on emoji and pictographs defines them as picture characters originally associated with mobile phone use in Japan that are now popular worldwide, with each one carrying a fixed name like FACE WITH TEARS OF JOY and a permanent codepoint that travels with the character no matter which platform draws it.

The everyday meaning is looser. People use the word for the face emojis, the object emojis, the flags, the symbols, the skin-tone variants, and the platform-specific reactions like Slack and Discord custom emojis that are not part of the Unicode Standard at all. The strict Unicode set is what governs the characters that travel cleanly across iPhone, Android, Windows, Mac, and the web; the rest are platform features dressed in the same name.

Buffer's glossary entry on emojis lands in a similar place, defining them as small digital images used to express emotions, ideas, or concepts in online communication. The split between the technical definition and the everyday one matters for anyone trying to ship the same emoji across more than one platform, because the technical version is the one that survives the trip and the everyday one is the one people argue about.

Where emojis came from

The origin story most people know names Shigetaka Kurita and the year 1999. Kurita designed 176 twelve-by-twelve pixel icons for NTT DoCoMo's i-mode mobile internet service, a Japanese pager and early phone product, and the set covered weather, moods, food, transport, and the basic shorthand a teenager would need to send a short message without a full keyboard. The full set lives in the permanent collection of the Museum of Modern Art in New York.

The fuller version of the history is messier. Britannica's entry on emoji notes that the Japanese mobile carrier SoftBank had already shipped a 90-icon black-and-white set in 1997, two years before Kurita, and that the designer of the SoftBank set is unknown. Kurita stayed the famous one because the i-mode set spread further inside Japan and because his design language is the one Apple drew from when it brought the first Apple emoji set to the iPhone in 2008.

The point at which emojis stopped being a Japanese phone feature and became a global writing system is October 2010, when the Unicode Standard incorporated 722 emoji characters in version 6.0. Once the characters had Unicode codepoints, Apple, Google, Microsoft, and the platforms could draw their own versions and the same character would render on every device. Everything after that is a question of how each new release of the standard adds more, which platform ships them first, and which ones argue about what counts as a person.

How emojis actually work (Unicode and codepoints)

Every standard emoji is a Unicode character with three things attached: a unique codepoint that nothing else in the standard uses, a fixed English name, and a category. The grinning face is U+1F600. The heart emoji is U+2764 followed by U+FE0F, the variation selector that tells the system to draw the heart in colour rather than as the older monochrome heart symbol. The family-of-four emoji is six codepoints joined by zero-width joiners, which is why some older devices break it apart into the four individual people instead of rendering the group.

The standard is owned by the Unicode Consortium, a nonprofit run by Apple, Google, Meta, Microsoft, Netflix, Salesforce, and a small number of academic and government members, and the consortium reviews proposals for new emojis on a roughly annual cycle. Anyone can submit a proposal; almost all of them get rejected. The ones that survive run through an Emoji Subcommittee review and end up in the next major release of the standard.

The most recent release is Unicode 17.0, finalised on 9 September 2025, which added seven new emoji codepoints (Distorted Face, Fight Cloud, Orca, Hairy Creature, Trombone, Landslide, and Treasure Chest), a gender-neutral Ballet Dancer sequence, and skin-tone variants for existing Bunny Ears and Wrestling sequences. Emojipedia tracks the rolling total at 3,953 recommended emojis across all devices once Unicode 17.0 is fully supported, and the Unicode announcement blog covers the same release.

Why the same emoji looks different on different devices

Unicode defines the name and the codepoint; it does not define the picture. Each platform draws its own version of every emoji in its own house style. Apple's set leans round and glossy, Google's Noto set is flatter and more illustrative, Samsung's is closer to Apple than Google, Microsoft's Fluent set is bright and almost three-dimensional, and the Twemoji set that WhatsApp and Meta draw from is the flat cartoon style most third-party emoji pickers default to.

Sometimes the difference is cosmetic and the meaning carries. Sometimes the meaning shifts a little, and a small set of emojis have managed to change meaning entirely because the rendering split was wide enough to confuse the readers. Emojipedia has a long-running thread on the gun emoji, which Apple changed from a revolver to a green water pistol in 2016 and every other vendor eventually followed, and the older Emojipedia archive is the canonical place to check which platform draws what.

For a brand that ships posts to followers across iPhone, Android, and desktop, the practical effect is that the same emoji is going to read slightly differently in three places at once. The fix is not to stop using emojis; it is to pick the ones whose rendering is reliably similar across the major vendors, and to check the platforms the audience actually uses before committing to a less common one. Sparkles, fire, the red heart, the pointing-finger arrows, and the check mark all survive the trip cleanly. The yellow shrugging-shoulders person, the loudly crying face, and the gun do not.

Emojis in social media marketing

The numbers people quote for emoji engagement in marketing all trace back to a small set of studies and A/B tests, mostly run between 2015 and 2020. Sprout Social's guide on how to use emoji in marketing to drive engagement collects the headline figures: a 25 per cent engagement lift on what was Twitter, a 57 per cent lift on Facebook with shares up 33 per cent, an 85 per cent lift on push notification opens, and a 241 per cent click-through lift from an AdEspresso ad headline test. Hootsuite's emoji guide cites the same lifts and goes on to make the point that the bigger risk for a brand is not whether to use emojis but whether the brand actually knows what each one currently means.

The lift is real and it has narrowed. When five per cent of posts on a network used emojis, an emoji was a useful signal that the post was casual and worth reading. When ninety per cent of posts use them, the signal flattens out. Buffer's own analysis of its 2025 posting data, in the most popular emojis used in social posts in 2025, puts sparkles at the top of every major network and shows that the most-used emojis on professional accounts are the functional ones (sparkles, the pointing finger, fire, the check mark, the light bulb, the rocket) rather than the emotional faces. Professional users use emojis like punctuation; consumers use them like words.

The conclusion most brand teams have arrived at, after a decade of A/B testing, is that the emoji has to do a job the words could not. A sparkle at the start of a sentence that signals a launch line. A pointing finger that steers attention to a link in the next line. A check mark in a feature list where a hyphen would have looked thinner. The places it stops working are decorative rows at the top of the post, faces stacked on the end of a sentence that already had the emotion built into the words, and the same one or two emojis used on every single post for a quarter until the audience starts skipping the opening line on reflex.

Per-platform emoji conventions

Every platform has settled into its own emoji habits and a post that ignores those habits looks like it was written somewhere else and dropped in.

Practising one set of emoji habits and copying them straight across is the most common cause of cross-posted captions underperforming on the second platform. The same idea written with a different emoji budget on each network is the version that lands.

Emoji and accessibility

Screen readers do read emojis out loud, character by character, using the official Unicode name. FACE WITH TEARS OF JOY. RED HEART. PERSON SHRUGGING. A row of five emojis at the start of a sentence becomes a five-word audio interruption before the sentence even begins. The accessibility guidance from the Bureau of Internet Accessibility's piece on emoji best practice lands on three rules that most other accessibility guides repeat in slightly different words.

1. Put emojis at the end of the sentence. A reader using a screen reader can skip the trailing description without missing the actual message. Front-loaded emojis make the opening unreadable for anyone using assistive tech.
2. Do not use emojis to replace words. The sentence has to still make sense if the emoji is removed. A warning that depends on the construction worker emoji to communicate that there is a warning is a warning that does not work for screen reader users.
3. Cap the count at three per post. Most accessibility guides settle on three as the upper limit before the screen reader experience becomes hostile, and the same cap doubles as a useful editorial limit for sighted readers.

Common emoji mistakes in brand use

Most of the recurring mistakes in brand emoji use come from the same root: nobody read the post out loud before it shipped, or nobody asked whether the emoji adds anything the words do not already say.

1. Using emojis whose meaning has shifted under you. The peach is the obvious one, the eggplant is the next, and the lipstick mark, the snowflake, and the bone all carry secondary meanings in 2026 that they did not in 2018. A quick check on Emojipedia before a brand uses an emoji it has not used before is a five-second habit that saves a quote tweet later.
2. Decorating instead of communicating. Five emojis at the start of a LinkedIn post is the brand-team version of clearing the throat. The post reads better with the words alone and one emoji at the end.
3. Repeating the same emoji every post for a quarter. A sparkle on every launch post for four months stops being a launch signal and starts being wallpaper. Rotate, or use the emoji only on the posts that genuinely need a launch marker.
4. Skin tone and gender variants that do not match the speaker. Brand accounts that use a single default skin tone or gender variant on every emoji read as inattentive when the team running the account is diverse. Most schedulers now support inline emoji editing, and the small change is usually worth the thirty seconds.
5. Cross-posting emojis that only one platform supports. A TikTok square-bracket shortcode pasted to Instagram stays as the literal text. A custom Slack emoji name pasted to Threads stays as a colon-wrapped string. The cross-poster has to convert or strip platform-specific emojis before they ship.
6. Treating the emoji as the message. A post that depends entirely on the emoji for the meaning is a post that fails for every reader who is on the version of the platform that draws it differently. The words have to carry the meaning on their own.

For the wider context that decides whether an emoji even helps, the caption entry covers the writing the emoji is sitting inside, the algorithm entry covers what each platform actually rewards, and the content pillars entry covers the planning layer the emoji is decorating.