Text Expanders and Autocomplete: Type Less, Not Faster

The fastest keystroke is the one you never type, and two kinds of tool try to remove keystrokes for you. A text expander replaces an abbreviation you type with a phrase you have saved. A word suggestion offers the word you are probably about to type, and you accept it with a tap or a key. At the average online typing speed, a text expander saves about 13.0 seconds on each use of a 60-character phrase ¹, which adds up to about 9.1 hours a year at ten uses a day over 250 working days ². Word suggestions are less clear. In controlled studies, more assertive suggestions cut keystrokes ³, but checking and accepting them takes attention, and that cost can offset the saving ⁴.

The fastest keystroke is the one you never type

Typing speed is measured in words per minute, and the 2018 online typing study counts a word as any five characters ⁵. The average in one large online typing test was 51.56 words per minute ⁶. On phones, the average in a study of 37,370 volunteers copying sentences shown on screen was 36.2 words per minute ⁷.

At those speeds, every character you do not type returns a little time. Two things decide whether that happens. The first is how often you type the same text. The second is whether the shortcut is quicker to trigger than the phrase is to type. Both are covered below. Our overview of typing speed in general explains the baseline, and the guide to how to type faster covers the other route: more practice at the keys.

Savings are also tracked as keystrokes per character, a count that falls when a tool fills in text for you. A lower count is not the same as a faster finished text, because the time you spend checking the tool is part of the total.

Text expanders replace short abbreviations with whole phrases

A text expander keeps a list of abbreviations and the text each one stands for. When you type the abbreviation, the expansion replaces it. People use expanders for email sign-offs, addresses, product codes, standard replies and form text. If you want the definition in plain terms first, the question on what a text expander is gives the short answer.

The simplest version is built into the operating system. On a Mac, text replacement can replace certain text with other text or symbols, for example replacing (c) with a copyright sign ⁸. On an iPhone, Apple's example is typing "GM" to produce a longer greeting, because text replacement lets a shortcut become a longer phrase or word ⁹.

Dedicated expander apps add more: larger libraries, sharing across a team, and expansions in more apps. One open-source option is espanso, which describes itself as a privacy-first, cross-platform text expander written in Rust ¹⁰. Commercial expanders such as TextExpander are another option. Their prices and plan limits change, so compare them on the vendor's own pricing page rather than in a third-party list.

The main risk of an expander is forgetting the abbreviation. If you cannot recall it, you type the full phrase anyway, and the expander adds nothing. Choose short codes that you would not type by accident in ordinary writing, and keep the list small enough to remember.

How much a text expander saves depends on how often you repeat yourself

Take a 60-character sign-off, about twelve words of five characters ⁵. At the average online typing speed it takes about 14.0 seconds to type ¹¹. On a phone at the average transcription speed in the phone study it takes about 19.9 seconds ¹². Speaking it on a phone at the study rate takes about 4.7 seconds ¹³. A four-character abbreviation, such as "sig" and a space, takes about 0.9 seconds to type at the keyboard speed ¹⁴.

The saving per use is the keyboard time minus the abbreviation time: about 13.0 seconds ¹. The total depends on how often you use it. The yearly figure assumes 250 working days.

At ten uses a day, the saving is about 9.1 hours a year ². The saving is simply proportional to use: half as many uses saves half as much time, and a phrase you send once a week saves a small fraction of that.

Those numbers ignore three costs. You have to build the list, which takes a few minutes for each phrase. You have to remember each abbreviation. And you have to find the expansion when you are not sure it is the right one. A phrase you type once a month is rarely worth an expander. A reply you send every day usually is.

Expanders also work best on short, fixed text. For long text that changes each time, speaking it is usually quicker. That is the subject of the last section. For the baseline speeds behind these figures, see our guide to average typing speed.

Word suggestions can cost more time than they save

A word suggestion saves keystrokes only if the suggestion is right, and only if you notice it, read it and accept it without slowing down. The research measures both sides of that trade. The question on whether word suggestions save typing time takes the same evidence in short form.

In a 2016 CHI study, the authors varied how assertively a mobile keyboard showed suggestions. Suggestions were always shown, never shown, or shown only above a probability threshold. The authors found that the costs of attending to and using the suggestions impaired average time performance ⁴. More assertive suggestions did cut keyboard actions, and participants preferred them ³. Fewer actions and a faster finish are different things, and the study shows the gap between them.

The same pattern shows up in other populations, with different results. The table gathers the main studies and their limits.

The blind-user study shows the cost most clearly, and it has an inconsistency to note. Its abstract says prediction slightly improves typing speed. Its results section reports a lower average speed with prediction ¹⁶, and it adds that the paired test found no significant difference between the two keyboards ¹⁸. The same paper found that prediction did not enhance typing accuracy, and that it increased both physical and temporal workload compared with the default keyboard ¹⁹. That group types slowly, so the speeds do not transfer directly to a fast typist. The workload finding is the one most likely to carry over, though this study did not test sighted typists. For more on this group, see our overview of accessibility.

The lesson is not that suggestions are useless. The gain depends on how good the suggestions are. Turn suggestions off on a keyboard you already type well, and test your own speed with and without them for a week before you decide. Track typing accuracy alongside speed, because a faster finish with more errors to fix is not a gain.

Autocorrect is linked to faster typing because it lets you type loosely

Autocorrect has the most consistent evidence behind it of the suggestion features, because it works without asking you to read anything. You type quickly and loosely, and the keyboard fixes the word when you finish it. In a phone study of 37,370 volunteers ⁷, people who used auto-correction typed faster, while people who tapped the words the keyboard suggested typed more slowly ²⁰. That is a correlation in observed behavior. Faster typists may simply use suggestions less, so the data cannot prove that suggestions slow people down.

The reason autocorrect works is that the keyboard combines two models: one for where your finger landed, and one for which words are likely. A 2015 study of a touchscreen keyboard simulated noisy touch input and found that a combined spatial and language model reduced word error rate from a pre-model baseline of 38.4 percent down to 5.7 percent ²¹. The result comes from a simulation, so it describes how well the model can decode touches, not the error rate you will see on your own phone.

Autocorrect still needs checking. A fix you did not want costs a keystroke to undo, and a fix to a word you meant to type exactly can slip past you. Our guide to how to type faster covers how speed and accuracy trade off against each other. For the question of whether autocorrect helps on phones specifically, see the question on whether autocorrect makes you type faster, and the guide to phone typing speed for the full phone picture.

AI autocomplete writes whole phrases, and nudges your wording toward its own

AI autocomplete goes beyond the next word. It can offer a whole sentence, and that changes what the saving is. In controlled tasks, AI assistance has cut completion time by large amounts. Developers who had GitHub Copilot finished one coding task 55.8 percent faster than the control group, with a 95 percent confidence interval of 21 to 89 percent ²². In a working paper, professionals who had ChatGPT took 37 percent less time on a writing task, a drop of 10 minutes ²³.

Those figures are not typing speeds. The coding study measured completion time on one task, the writing study measured time on a task with a limit, and the AAC study measured communication rate. They show that assistance can save a lot of time when the task is bounded and the output is checkable. They do not show that AI autocomplete makes everyday typing faster. The working paper also reports that evaluator grades rose by 0.45 standard deviations ²⁴, but its authors did not check the writing for factual accuracy, so that gain says nothing about whether the text is true. The published Science version may report different figures, so check it before you quote the working paper. Our question on whether AI can help you type faster gives the short answer.

The second effect is on what you write. Suggestions shape content, not only speed. In an online experiment with 1,506 participants ²⁵ who wrote a post on whether social media is good for society, some were shown suggested text generated by a large language model ²⁶. The model had been configured to argue one side. Using the opinionated model affected the opinions expressed in participants' writing, and it shifted their opinions in a later survey ²⁷.

That result is the reason to read an AI suggestion before you accept it. The model's wording is a choice, and it can pull your text toward its own view. For a quick reply, a suggestion you have checked is fine. For writing with a point of view, type the first draft yourself, then use the assistant to tighten it.

Mac and iPhone include free text replacement, with steps you can follow today

Mac and iPhone both have built-in text replacement, so you do not need a separate app. On a Mac running macOS Ventura 13 or later, the steps are in System Settings: open Keyboard, click Text Replacements, type the text to replace and its replacement, then click Add ²⁸. Apple's example replaces teh with the.

On an iPhone, the path is different. Open the Settings app, tap General, tap Keyboard, then tap Text Replacement ²⁹. Tap the Add button, enter the phrase and the shortcut, then tap Save. When you type the shortcut in a text field, the phrase replaces it automatically ⁹.

We could not open an official Google page that gives the steps for adding shortcuts on Android, so this guide does not give them. Check your keyboard's settings for a personal dictionary or text shortcuts, and use the steps from Google's help for your phone's version.

Windows is not in the table because we did not verify a built-in text replacement feature there. For Windows, espanso describes itself as cross-platform, so it is the option from the earlier table to check first ¹⁰.

Dictation is the other way to type less

Speaking is much faster than typing for raw text. In a controlled study on an iPhone, English speech entered text at 153 words per minute ³⁰, while the built-in keyboard reached 52 words per minute in the same study ³¹. Speaking a 60-character sign-off takes about 4.7 seconds ¹³. That is still longer than the 0.9 seconds an expander abbreviation takes ¹⁴, so for fixed text you repeat, the expander stays quicker.

Dictation is not a free replacement for every task. Correction changes the sum. In a 1999 CHI study, for users with good typing skills, correcting errors by typing was faster than multimodal correction, which in turn was faster than speech-only correction ³². The honest comparison is the finished text, after editing. Our guide to dictation vs typing works through when speaking wins, and our comparison of speaking and typing speed covers the research behind the speed gap. The overview of dictation covers setup.

Yap, which we make, is an on-device dictation app for Mac and Windows, so speech recognition runs on your own device. If you want to try dictation next to your expanders, the download page has the current versions. For free dictation without an install, use online dictation in the browser.

The practical plan is simple. Use an expander for short, fixed text you type every day. Use autocorrect, and keep word suggestions off unless you have measured a gain for yourself. Use AI assistance for bounded tasks you will check, and read suggestions before you accept them. Speak long, changing text, and correct it by hand. For the wider picture of typing speed, including how fast people speak and type, start with the overview.