What is predictive text & how to use it

Predictive text is a keyboard feature that suggests your next word, or a full phrase, as you type. It shows up across smartphones, computers, and apps like Gmail and Google Docs, and its job is to help you type faster with fewer errors. Tap a suggestion, keep moving.

It’s become so common that most people use it without thinking about it. But knowing how to manage predictive text and when to reach for something more precise makes a real difference in how you communicate at work.

What is predictive text?

Predictive text is a feature that analyzes what you’ve typed and suggests likely next words or phrases based on language patterns, context, and your writing habits. On a phone, it appears in the predictive text bar, a row of suggestions above the on-screen keyboard. Tap a suggested word to insert it. On a computer, it often appears as gray suggested text that completes your sentence in Gmail.

The technology started with T9, the number-pad system on flip phones that guessed which word you meant from the letters on each key. Text messaging on T9 was clunky, but the idea was the same as today. Modern predictive text systems learn from your messages, factor in the app you’re using, and suggest whole phrases at once, from a single word on a phone to full-sentence Smart Compose in Google Docs. Today, predictive text is built into all major operating systems.

How does predictive text work?

Simple version: predictive text systems look at the words you’ve typed and calculate which word is most likely to come next, based on patterns learned from a large amount of text. Commonly used words or phrases rise to the top of the suggestion bar.

The longer version involves three generations of technology.

N-gram models were the earliest approach. An n-gram model looks at the last N words you typed and offers the most common follow-up as the suggested word. Type “Happy” and a bigram model suggests “birthday” because that pair appears often in text.

Machine learning moved the needle. Instead of fixed pattern tables, machine learning models learned from user behavior over time, picking up individual habits, slang, and domain-specific language.

On-device neural language models are where machine learning stands today. Apple says the iPhone keyboard in iOS 17 and later uses a transformer language model, the architecture behind large language models, scaled down to run on the phone. Google Research says Gboard’s next-word prediction uses neural network language models trained with federated learning, which keeps training data on the device. This kind of artificial intelligence reads context across a longer stretch of text, which is why suggestions sometimes feel eerily accurate.

A personalization layer sits on top. Your keyboard tracks the words you use, the contacts you message most, and the app you’re in. You write differently in Slack than in a birthday text, and the AI learns that difference.

How to use predictive text on your device

Each of the major operating systems has its own setting to turn predictive text on or off. Here are the current paths for each device.

Android

To enable predictive text in Gboard (the default keyboard on many Android devices):

1. Open Settings
2. Tap System > Keyboard > On-screen keyboard (on older versions, System > Languages & input > On-screen keyboard)
3. Select Gboard
4. Tap Text correction
5. Toggle Show suggestion strip or Next-word suggestions on

To enable predictive text on Samsung Keyboard: Tap the settings icon on the keyboard, then turn predictive text on. Or go to Settings > General management > Samsung Keyboard settings > Predictive text. Menu names vary by One UI version. Samsung Keyboard also offers Auto spell check, which underlines misspelled words in red.

To reset or retrain suggestions: In Gboard, open keyboard settings and go to Dictionary > Delete learned words. On Samsung Keyboard, go to Samsung Keyboard settings > Reset to default settings > Erase personalized predictions. Useful if suggestions picked up terms from a sensitive conversation, or you use the same device for professional and personal messages.

Practical tip: To remove words you don’t want suggested, touch and hold the suggested word. In Gboard, drag it to the trash. On Samsung Keyboard, tap OK.

iPhone

To enable predictive text (iOS 17 and later):

1. Open Settings
2. Tap General > Keyboard
3. Toggle Predictive Text on
4. Also enable Show Predictions Inline for full-sentence completions that appear directly in text fields

As a shortcut, touch and hold the emoji or globe key on the on-screen keyboard, tap Keyboard Settings, and turn predictive text on or off. Press the space bar to accept an inline prediction, or keep typing to ignore it.

To reset or retrain suggestions: Go to Settings > General > Transfer or Reset iPhone > Reset > Reset Keyboard Dictionary. This clears all learned vocabulary on the device.

Practical tip: Touch and hold the space bar to turn the keyboard into a trackpad, then drag to move the cursor mid-sentence.

macOS

To enable inline predictive text (macOS Sonoma and later):

1. Open System Settings
2. Click Keyboard
3. Under Text Input, click Edit
4. Toggle Show inline predictive text on, then click Done

Predictions appear as light gray text after your cursor. Press the space bar to accept, or keep typing to dismiss. Inline predictions aren’t available in every language, so check your input language if they don’t appear.

Windows

To enable text suggestions in Windows 11 (Microsoft’s name for predictive text on a physical keyboard):

1. Open Start > Settings
2. Go to Time & language > Typing
3. Toggle Show text suggestions when typing on the physical keyboard on

Practical tip: Text suggestions appear in most places where you enter text. If they don’t appear in a specific app, that app may be overriding the operating system’s text input.

New Outlook and Microsoft 365

Outlook and Word both offer predictive text, which Microsoft calls text predictions: inline completions you accept with Tab or the right arrow key. In the new Outlook, go to Settings > Mail > Smart suggestions, then under Text predictions, select or clear Suggest words or phrases as I type and select Save.

How to use predictive text in popular apps

Some programs go beyond the keyboards built into operating systems and add text predictions to their own interface:

Google Docs: Smart Compose, the predictive text feature in Docs, shows gray suggested text after your cursor as you start a sentence. Press Tab or the right arrow key to accept. Turn it on or off under Tools > Preferences > Show Smart Compose suggestions.

Microsoft Word: Editor text predictions behave similarly: suggested text appears inline as you type, accepted with the Tab key. Manage the feature under File > Options > Advanced, then select or clear Show text predictions while typing under Editing options.

Messaging apps: WhatsApp, iMessage, and most messengers use your phone’s predictive text, so switching keyboards is the fastest way to get stronger predictions. See our roundup of the best predictive text apps to compare options.

Predictive text vs. text expansion: What’s the difference?

Short version: they’re not competing for the same job.

Predictive text is reactive. The system watches what you type and offers suggestions based on patterns and context. You’re still doing most of the typing, and the output is the keyboard’s best estimate of what you meant.

Text expansion is proactive. You define an abbreviation, assign it to a block of text, and the expansion happens the moment you type that shortcut phrase. The output is exact. Every time.

Feature	Predictive text	Text expansion (TextExpander)
Triggered by	Your typing patterns	A specific abbreviation you define
Output	Suggested word or phrase	Exact text you specified
Learns over time	Yes, adapts to your habits	Only if you update your Snippets
Works across apps	Keyboard-level (most apps)	Mac, Windows, Chrome, iPhone, and iPad
Team sharing	No	Yes, shared Snippet libraries
Best for	Casual messages, short replies	Professional responses, templates, forms
Accuracy	Varies with context and your typing history	Exact: it’s what you saved

Predictive text is great for everyday typing: quick texts, common phrases, low-stakes communication. Text expansion is for when you need the exact right words, every time. Support responses, medical documentation, legal disclaimers, client emails. Contexts where “close enough” costs you.

No matter which software we use, which platform we’re on, we’re always going to have a form of consistency.

Alex Williams, Customer Onboarding Manager, CompanyCam

The two also operate at different scales. Predictive text works a word or phrase ahead. Text expansion handles anything from a two-word phrase to a multi-paragraph template with fill-in fields.

There’s also an autocomplete angle worth understanding. Autocomplete and predictive text are often used interchangeably, but they work differently at the technical level.

Go further with TextExpander

TextExpander starts from the answers you already know. You save the text once, assign it a short abbreviation, and it expands wherever you type. Predictive text makes an educated guess about your next word. TextExpander inserts the wording you’ve already decided on. For example:

Multi-line email reply

Abbreviation: ;reply

Hi [name],

Thanks for reaching out. I’ve reviewed your message and will put together a full response with everything you need by [date].

In the meantime, feel free to reply here or reach me directly at [phone].

Best,

[Your name]

The bracketed fields are fill-ins. TextExpander prompts you to complete each one before the expansion finishes, so you reuse what stays the same and apply judgment where the situation changes.

Introduction with fill-in fields

Abbreviation: ;intro

Hi, I’m [fill: Your name] from the [fill: team] team. I wanted to reach out about [fill: topic]. Let me know if you have a few minutes this week.

Short, high-frequency Snippet

Abbreviation: ;ty

Thank you for reaching out. I’ll review your message and get back to you within one business day.

The real advantage over predictive text is consistency. A Snippet expands to the same text every time, for every team member who uses it. With TextExpander for teams, you share Snippets with the people who need them, and one update reaches everyone. You can also use TextExpander for autocorrection, and your Snippets also expand on your phone with TextExpander for iPhone and iPad.

It was all very straightforward after we got TextExpander because the answers were right in front of us. We didn’t have to go look for them. It took away the second-guessing.
Kammikia Barnes, Population Health Strategist, CloseKnit Health

If you’re new to Snippets, we have a practical guide on creating and organizing your first Snippets.

Tips for getting the most out of predictive text

• Accept suggestions regularly. The more you accept text predictions, the more the system learns your patterns and commonly used words. On iPhone and Mac, the space bar accepts an inline prediction.
• Delete unwanted predictions. To remove words you don’t want suggested, such as typos, touch and hold the suggested word in the predictive text bar.
• Keep spell check on. Predictive text suggests words but doesn’t correct spelling after the fact, so leave autocorrect and spell check running.
• Match your register. Predictive text doesn’t know you’ve switched context from Slack messages to a client email. Reset suggestions when they drift, or turn predictive text off for formal writing. For more ways to cut errors, see these tips to speed up typing.
• Use both together. Predictive text handles casual filler. For anything professional, pair it with TextExpander Snippets that work like a spell-checker and more so the content that has to be right always is.

Conclusion

Predictive text has come a long way from T9’s number-pad guesses to today’s on-device language models, and the feature is worth configuring rather than leaving on defaults.

But it has a ceiling: it can only guess at what comes next. For professional communication where the words need to be exactly right, TextExpander keeps the answers you use most ready wherever you type. Use predictive text for quick messages and text expansion for anything that matters.

Want stronger predictions? Our guide to the best predictive text apps covers the top options. Ready to go beyond guessing? See our roundup of the best text expander tools.