Definition
Search autocomplete is a technology that predicts the words or phrases most likely to come next each time you type a character into a search box, and shows them as a list. You don't have to type everything out. It guides you to the information you want faster.
A tree where one letter splits into tens of thousands of words
When you look for a book in a library, shelves sorted alphabetically help you find it fast. A search box works on a similar idea. A computer doesn't scan hundreds of millions of words from start to finish. It uses a special word dictionary, such as a data structure called a trie, with branches that grow letter by letter. Real services often use compressed or modified versions of this structure.
Say you type 'c'. The computer jumps straight to the big branch for words starting with 'c'. It only needs to look at the words hanging below it, like 'cat', 'car', and 'cake'. Even with tens of millions of words stored, it just follows the letters in order, so it can narrow down candidates in well under a second.
A cache (a temporary memory spot) helps too. It holds searches people just made. Popular words are kept within easy reach, so the list can show up in the blink of an eye, often before you press the next key.
Scoring thousands of candidates to pick the winners
Tens of thousands of words may start with 'c', but the suggestion list under the search box usually shows only about 5 to 10. How does the computer pick the real winners from this crowd? The key is how many people actually searched for each one.
The search system looks at live search counts, current trends, and even seasonal patterns, then gives each word a score. In fall foliage season, 'fall foliage near me' scores high. On the day a popular smartphone launches, 'pre-order bonus' automatically takes the top spot.
Autocomplete also fixes common typos. If your finger slips and you type 'aple', or you misspell a word like 'recieve', the search engine works out what you meant and neatly puts the word you were actually looking for in the list.
A bit more precisely: AI and personalization
To be more precise, modern autocomplete goes far beyond the old approach of matching letters and ranking by search volume. Today it combines artificial intelligence (AI) with natural language processing, so it can use the context of a sentence and tailor suggestions to each user.
Typing the same letters can give very different suggestions depending on whether you're in New York or Hawaii, or whether you follow sports or play video games. This is why a programmer who types 'ja' might see the programming language 'Java' before the grocery item 'jalapeño'.
All this complex work is usually designed to finish in about 0.1 seconds. That's because many high-performance servers spread around the world pass data back and forth at lightning speed, delivering the best-fitting words in real time.
Common misconceptions
The words in autocomplete show up because your phone is listening in on your everyday conversations.
It isn't eavesdropping on your voice. It's a probabilistic prediction built from the patterns in the letters you type, huge search statistics from people all over the world, your location, and your earlier search history.
Where you meet it
You type just 'wea' in the search box, and 'weather' and 'weather forecast' pop up in the list right away.
In English, typing 'you' brings up 'youtube' at the very top as a suggestion.
In one sentence
This technology combines a trie data structure, search statistics, and AI to instantly predict and show the words most likely to come next as you type.