WhatPulse 7.0 is the biggest release since the app was first written. The app has been restructured from one program into two, and that change touches almost everything: how your stats are counted, how the window gets its data, how updates are installed and everything else. Most of it is invisible when everything works, which is exactly the point.

Around that new foundation, 7.0 brings a new sidebar navigation, league standings on the Overview page, an in-app feedback window, better application recognition, a universal macOS build, a big round of Linux work, accessibility improvements, and a long list of fixes. It went through six public betas (a new record), and the feedback from beta testers shaped a large part of this release. Thank you to everyone who tested it.

Counting and the window, finally apart​

Until now, WhatPulse was a single program. The code that counts your keystrokes and the code that draws your charts lived in the same process and went down together. If the window hit a bug, counting stopped. If the counting side was busy, for example processing a flood of network packets, the whole app felt it.

7.0 splits that into two:

• The collector is a small background process that does all the counting: keys, clicks, mouse distance, uptime, application and window title tracking, network traffic, browser extension traffic, pulsing, backups and uploads. It starts when you log in and runs without a window or Dock icon.
• The app is the window you already know: the pages, charts, heatmaps, tray popup, Geek Window and settings. It reads your statistics directly and talks to the collector for everything else.

What this means day to day:

• A busy collector no longer slows down the window. Heavy work in the collector, like counting a lot of network packets, exporting data, can no longer make the interface sluggish, and the window itself is a lot snappier - you'll definitely notice this.
• Updates are atomic. Quicker and more reliable updates, with full rollback if anything fails. macOS and Linux ship both inside the same package.
• Tools built on the Client API keep working. The local Client API and the browser extension connection are now served by the collector, on the same ports and settings as before, nothing has changed there.
• Your database is safer. The database now uses a journaling mode that lets the app read while the collector writes, and online backups are taken as consistent snapshots of the live database.

If you're looking, you will see two entries in Activity Monitor or Task Manager: WhatPulse and WhatPulse Collector. Both are expected, and on Windows the collector shows up with a proper name and icon.

macOS: a one-time permission prompt​

macOS ties Input Monitoring and Accessibility permissions to a specific program. The collector is a new program, so the permission you granted WhatPulse before does not carry over. After updating, macOS prompts you once to allow the collector. WhatPulse walks you through it and keeps counting uptime and network traffic while you do. Once granted, the permission sticks across future updates. Screen readers now announce when a permission is granted.

Linux: network capture permission​

Network monitoring now runs in the collector. If you used setcap to give WhatPulse permission to capture network traffic, grant it to whatpulse-collector instead. This is only needed when you're not running the AppImage or Flatpak, and use an extracted WhatPulse binary.

A new sidebar navigation​

The old top navigation bar and the separate tab strips on each page are replaced by one sidebar on the left. Every page and its sub-pages sit in one tree, so getting from Input > Keyboard to Network > Interfaces is one click instead of two.

• Collapsible. Shrink the sidebar to a slim icon rail to give your charts more room. Hover an icon to reveal its sub-pages, and WhatPulse remembers whether you left it collapsed.
• One Applications page. Uptime > Applications and Uptime > Application Activity are now one page, with an Activity / Data toggle to switch between the 24-hour activity graph and the per-application list. WhatPulse remembers which view you picked.
• Prefer the old layout? The classic top navigation is still there under Settings > Appearance > Navigation.

League standings on the Overview page​

If you're taking part in the Leagues, your standing is now right on the Overview page: your division, your rank, which zone you're in, and how much time is left in the round. Near the end of a round, when scores are hidden, it says so. Click the block to open the full leaderboard on the website.

The Geek Window gets matching league variables, so you can put your division, rank or zone in your own layout.

Send feedback from inside the app​

For a long time, the only way to get diagnostics to us was finding log files by hand and attaching them to an email. 7.0 adds a feedback window, reachable from a new icon in the title bar while you're logged in, and from Settings > Advanced > Report a problem.

• Report a problem. Describe what happened and optionally attach your logs and statistics database. WhatPulse packages them and uploads them to a support session tied to your account, so we can look at the actual data instead of guessing.
• Suggest a feature. Tell us what you'd like to see and how much it matters to you. Opening the window from the title bar preselects the page you were on.
• Other feedback. For everything else.

Better application recognition​

Many applications, games in particular, have generic file names like game.exe or launcher.exe and no publisher information. The website could only match these by name, so unrelated apps were merged into one entry. WhatPulse now also sends what identifies an app more precisely:

• Windows: the publisher from the app's digital signature.
• macOS: the developer's Team ID from the app's signature, so Mac App Store apps are no longer all attributed to Apple.
• Games: the store and store ID for games installed through Steam, the Epic Games Store, GOG or the Microsoft Store.

Applications you already use are updated once in the background.

Cleaner window title lists​

Window title tracking, new in 6.3, had one weak spot: applications whose title changes constantly. A game showing its FPS, a media player counting seconds, a download showing a percentage, or a chat app with an unread count each produced a brand-new title every time the number changed, cluttering your lists and growing the database.

WhatPulse now recognizes these volatile titles on its own, by watching how a title changes over time instead of using a fixed list of patterns, and records them under one stable entry. Genuinely different titles, like switching between numbered tabs, are left alone. Learned patterns are remembered across restarts, and a slow background cleanup folds titles recorded before this update into the same entries over the coming weeks.

A fresh look​

• One icon set. WhatPulse had collected icons from several sources over the years. They are now all replaced by one consistent icon family across the Overview page, the sidebar, every settings page and the title bar. Sharper at any display scaling, and consistent in weight and size.
• A new installer look. The installer follows your system's light or dark mode, and it's a lot prettier. On macOS, the disk image has a clear background with instructions, and the installer is called Install WhatPulse.
• Small touches. The title bar is slightly darker than the window body, message banners use their full width, and the Advanced settings menu is sorted alphabetically.

macOS​

• One universal build. WhatPulse for macOS used to ship as two builds. There is now one universal build that runs natively on Intel and Apple Silicon. The installer and the maintenance tool run natively on Apple Silicon too, instead of under Rosetta. If you had the Apple Silicon build, you are moved onto the universal one automatically through the normal update.
• Reopening the app. Launching WhatPulse from the Dock, Finder, Spotlight, Launchpad or a launcher like Raycast while it is already running now opens the main window.

Windows​

• Launch on startup, always available. Launch when computer starts was missing from Settings > General for some users, mostly on all-users installs or after switching install types. A leftover entry from older installers hid the setting and could stop WhatPulse from starting at all. The setting is now always shown and always under your control. On all-users installs, every account keeps starting WhatPulse at sign-in after updating.
• Task Manager startup toggle. If you turn WhatPulse off under Startup apps in Task Manager, the checkbox in WhatPulse reflects that, and ticking it again turns startup back on in Task Manager too.
• Windows 10 windows. Square corners no longer poke out past the rounded window frame, the main window no longer opens with its right and bottom edges cut off, and the thin gray line along the bottom edge of windows is gone.
• High DPI override. If the "Override high DPI scaling behavior" compatibility setting was enabled on WhatPulse.exe, the tray popup opened in the wrong place on scaled displays. WhatPulse now detects the setting and tells you how to turn it off.
• Installer elevation. If you decline or cancel the administrator prompt, the installer offers to try again or stop, instead of showing a dialog that could be unresponsive.

Linux​

7.0 includes the largest batch of Linux work in a long time.

• AppImage updates now replace the installed version and restart WhatPulse, instead of leaving the old version running, and update checks no longer stop working until a restart. The beta channel works on Linux too.
• Flatpak no longer starts without a tray icon or window on KDE Plasma and GNOME.
• Desktop integration. The WhatPulse window is grouped with its launcher in the dock. The AppImage follows a light GNOME desktop instead of starting dark, and the tray icon is visible on a dark top bar. Removing WhatPulse from your startup applications also stops data collection at login.
• Network. IPv6 traffic is attributed to the right application.
• Wayland. Native apps no longer disappear from application uptime when an XWayland app such as Steam is open, and the sidebar flyout opens next to the sidebar instead of mid-window.
• Hardware. Processors and graphics cards are now detected.
• Sleep detection. Sleep and wake are detected through the system's power management, so counting pauses correctly when you suspend.
• Crash reporting is now available on Linux, with the same opt-out as Windows and macOS.
• Smaller packages. Unused database drivers are no longer included.

Accessibility​

• The tray popup is now fully operable with the keyboard: Tab and the arrow keys move between controls, Space and Enter activate, Esc closes. Every control has a proper name for screen readers like VoiceOver and NVDA, and the pause control announces what it will do ("Pause tracking" / "Resume tracking").
• On macOS, a right-click or Ctrl-click on the tray icon opens a native menu, which VoiceOver users can reach with VO+Shift+M. The native menu also has a Pause/Resume tracking action, on macOS and Linux.
• The interface has been fully tested with VoiceOver and NVDA.

Fixes​

• Mouse heatmap retention. The "Prune older than 3 months" checkbox on Input > Mouse did not stop the pruning; heatmap data older than three months was removed regardless. It is now honored. This bug dates back to when the setting was introduced.
• Tray popup double counting. The Today and Week views added everything since your last pulse on top of totals that already included it. They now match the Overview page.
• Deleted profiles counted twice. Deleting a profile copied its stats to General but left the originals behind, so the combined view counted them twice. Existing duplicates are repaired automatically.
• Total uptime. Total uptime on the computer uptime page was calculated from boot times reported by the system. A wrong system date at any point skewed it permanently. It now sums the uptime WhatPulse counted itself, matches the chart and respects the profile filter. The separate "longest uptime" value is removed.
• Geek Window labels. A label added from the settings could disappear on the next save, or several new labels could end up sharing one entry so that editing one changed another. Labels are now stored correctly, keep the position you dropped them at, and the layout is saved in one go.
• Network attribution. On machines with two or more active network adapters, traffic could be attributed to the wrong address, port or direction. On dual-stack connections, IPv4 and IPv6 traffic could be merged. Both are fixed.
• Website icons. A site's icon address pointing at an endless stream, such as an internet radio feed, could keep a connection open and slowly use memory. Icon downloads now check the response up front, cap their size and give up after a deadline.
• Crashes. A long list of crash fixes, most found through crash reports during the beta: when closing, quitting or switching theme or language while a heatmap was loading; when resizing a large mouse heatmap mid-generation; memory corruption in Windows input capture; the Geek Window refreshing while a pulse arrived; looking up running processes on Windows; and several rare crashes during shutdown. We also reduced how often a Windows time zone bug seen on some Windows 11 systems could crash WhatPulse.

Privacy and security​

• Crash reporting on every platform. Crash reports were Windows-only. macOS and Linux now send them too. This follows the Send bug and usage reports setting in Settings > General, and reports never include what you type, window titles or application names.
• Scrubbed diagnostics. Log lines sent with crash reports are scrubbed of web addresses, home folder paths, e-mail addresses and text values from database queries before they leave your machine.
• Stricter connections. Online database backups fail on any certificate error instead of ignoring it, and website icon redirects to local network addresses are refused.
• Tighter boundaries. The Windows update helper only touches files inside the WhatPulse installation folder. The collector only accepts read-only queries from the window and only uploads the specific files it expects.

Before you update​

• macOS: expect the one-time permission prompt for WhatPulse Collector.
• Linux: if you use setcap for network capture, apply it to whatpulse-collector.
• Everyone: a backup never hurts. Premium members can use Account > Start Backup; everyone else can copy the WhatPulse data folder.

If anything looks off, the feedback button in the title bar is the fastest way to reach us, and it can attach the logs we need.

Happy pulsing!

Two years ago, AI lived in a browser tab. You opened ChatGPT in Chrome, typed a question, copied the answer and closed the tab. To WhatPulse, that looked like "Chrome" (before we added Web Insights at the beginning of 2026).

Then the AI companies started shipping real desktop apps. ChatGPT came to the Mac. Anthropic launched Claude for desktop. AI code editors like Cursor and Antigravity replaced VS Code on a lot of machines. Suddenly these tools had their own window, their own process and their own row in the WhatPulse application stats.

That gives us something most AI usage reports can only estimate: how much people actually type into AI apps. Not downloads, not sign-ups, not survey answers. Keystrokes and time, counted on real desktops, every hour, for almost three years.

Here's what 33 months of data say.

The short version​

In January 2024, fewer than 1 in 100 active WhatPulse users had a desktop AI app they typed into. By September 2026, more than 1 in 5 did. The average WhatPulse user now types 143 times as much into AI apps as they did at the start of 2024, and AI's share of everything people type went from a rounding error to almost 2%.

Two percent doesn't sound like much until you remember what's in the other 98%: every email, every Slack message, every line of code, every game. Almost one in every fifty keystrokes now goes into an AI app.

How we measured this​

WhatPulse counts keystrokes, clicks and active time per application, per hour, on every computer running the client. For this article we:

• followed the same group every month: everyone who was active on WhatPulse that month, about 12,000 to 13,000 people, from January 2024 to September 2026;
• grouped all builds of each app into one product, so "Claude" covers every Mac and Windows build of the Claude desktop app;
• only counted native desktop apps. AI you use in a browser tab counts as your browser, so the real numbers are higher than these.

WhatPulse users are not a cross-section of everyone. They're people who spend a lot of time at a computer: developers, gamers, writers, power users. That makes them an early signal. When their habits change, everyone else's tend to follow.

We never see what anyone types, only how often a key is pressed and in which app.

2024: AI moves to the desktop​

At the start of 2024, about 1 in 250 active WhatPulse users typed into a desktop ChatGPT app, and only a handful had tried an early AI code editor called Cursor.

Two things happened that year.

ChatGPT became an app. In May 2024, OpenAI released ChatGPT for macOS. That month, ChatGPT desktop's reach among active WhatPulse users quadrupled, from 0.3% to 1.1%. It never went back down.

Developers found Cursor. In the summer of 2024, Cursor took off. Monthly keystrokes went from about 500,000 in June to 3.1 million in September. By August 2024, Cursor had passed ChatGPT to become the AI app people typed into most, and it held that spot in 17 of the next 20 months.

Then, at the end of October 2024, Anthropic released the Claude desktop app. In its first month, it reached a quarter of a percent of active WhatPulse users. Modest, but it was the start of the biggest story in this dataset.

2025: everyone gets an AI app​

In 2025, desktop AI went from early adopters to normal. The share of active WhatPulse users with an AI app nearly doubled, from 7% in January to 13% in December, and new apps kept arriving:

• February 2025: 1.4% of active WhatPulse users installed a local LLM app (LM Studio or Ollama) in a single month, the biggest wave of "run AI on my own machine" in our data.
• July 2025: Perplexity launched Comet, its AI browser. At its peak in October 2025, it reached 4% of active WhatPulse users.
• October 2025: OpenAI launched ChatGPT Atlas, its AI browser. Within ten days, 1.4% of active WhatPulse users had it running.
• November 2025: Google launched Antigravity, its AI code editor. 1.4% of active WhatPulse users installed it in its first month.

Meanwhile the "big three" settled into a pattern. Cursor and ChatGPT each saw around 10 million keystrokes a month. Claude was a solid but distant third, at around 1 to 2 million.

That didn't last.

2026: the year of Claude​

In early 2026, Claude's line goes almost vertical.

In nine months, Claude desktop usage grew 12 times. Since its launch month in November 2024, it has grown 215 times. Today, 1 in 9 active WhatPulse users types into Claude every month.

Along the way:

• In March 2026, Claude passed ChatGPT in keystrokes (8.4 million against 7.8 million), and four times as many people installed Claude as ChatGPT.
• In April 2026, Claude passed ChatGPT in reach (7.0% of active users against 5.2%) and Cursor in keystrokes, making it the most-typed-in AI app on WhatPulse.
• By September 2026, Claude accounted for 45% of all keystrokes typed into AI apps.

Claude users don't just open the app, either. In September 2026 they spent an average of 15 hours in it, about half an hour every day.

Developers are moving from AI editors to AI agents​

The other big shift in 2026 is happening on developers' desktops.

Cursor peaked in March 2026 at 13.4 million keystrokes, reaching 3.7% of active WhatPulse users. By September 2026 it was down to 7.7 million keystrokes (-42%) and 2.7% of active users. New installs more than halved over the same period.

Cursor didn't collapse. It slid, steadily. The developers who stayed still spend about 14 to 15 hours a month in Cursor, the same as before, but they type around 20% less in that time. We're not the only ones seeing this. In JetBrains' 2026 research on AI coding agent adoption, the share of developers using Cursor at work fell from 18% to 12% between January and mid-2026, even though more developers than ever had heard of it.

At the same time, two things grew fast:

• Claude Code, Anthropic's coding agent, went from almost nobody in October 2025 to 3.4% of active WhatPulse users in September 2026, more than Cursor. You won't see its keystrokes in our data, because people type to Claude Code inside their terminal. But the share of people running it grew faster than any AI editor.
• OpenAI's Codex app appeared in February 2026. By June it reached 5.4% of active WhatPulse users and 13.3 million keystrokes.

The one AI editor still growing is Google's Antigravity, which went from 0.7 million keystrokes in its launch month to 6.1 million in September 2026, close to Cursor's level.

One likely explanation: more and more coding work is being handed to agents that run in the background, while developers type the instructions. Less time spent editing code by hand means fewer keystrokes in the editor.

OpenAI's second wind​

ChatGPT's desktop numbers were remarkably flat for a long time. From early 2025 to spring 2026, about 5 to 6% of active WhatPulse users typed into it each month, at 6 to 10 million keystrokes.

In July 2026, that changed. A new ChatGPT desktop app arrived (the previous version shows up as "ChatGPT Classic" in our data), and ChatGPT had its best month for new installs yet, more than double its usual pace. Keystrokes went from 7.6 million in June to 19.0 million in September 2026, and its reach grew to 6.9% of active users.

Over the same months, Codex usage dropped sharply, and in September a ChatGPT installer that bundles Codex appeared in our data. It looks like a lot of Codex usage moved into the ChatGPT app itself.

ChatGPT is growing again, but Claude now reaches about 50% more WhatPulse users and gets about 50% more keystrokes.

Who types the most?​

Over the full history, these are the desktop AI apps WhatPulse users have typed into the most:

A few things stand out:

• Cursor users spend the most time in it, almost 95 hours on average, and type over 115,000 keystrokes each. An editor is somewhere you work all day, not something you check now and then.
• ChatGPT users type the fastest, at about 2,000 keystrokes per active hour. Chat apps are pure writing: long prompts, follow-up questions, context pasted and edited.
• Claude has reached more WhatPulse users than any other AI app, and at its current pace it will pass Cursor and ChatGPT in lifetime keystrokes soon.
• Comet users spend lots of time but type little, because a browser is mostly for reading.

Local AI: installed, not used much​

Local LLM apps like LM Studio and Ollama are a surprisingly common sight: about 4% of active WhatPulse users have one running every month. But they barely type into them, about 3 hours per user over the app's whole lifetime. For most people, running a model locally seems to be an experiment, not a daily tool.

What this means​

1. AI now has its own place on the desktop. One in five WhatPulse users types into a desktop AI app every month, and that share has grown almost every month for almost three years.
2. The leader can change quickly. Cursor led for most of 2024 and 2025. Claude went from third place to first in about three months.
3. Coding is moving to agents. AI editors grew first; in 2026, coding agents like Claude Code and Codex are growing faster.
4. This is still early. At 1.9% of all keystrokes, AI is a small part of our typing day. The curve says it won't stay small.

See your own AI usage​

Curious how much you type into Claude, ChatGPT or Cursor? WhatPulse tracks keystrokes, clicks and time per application, privately and on your own computer. Your dashboard shows exactly which apps get your time, and how that changes month by month.

You can also look up any app's global stats: Claude, ChatGPT, Cursor.

Download WhatPulse and find out how much of your day goes to AI.

Methodology: hourly per-application data from everyone active on WhatPulse each month. Percentages are shares of that month's active users. At this size, a share like 10.8% has a sampling margin of error of about ±0.5 percentage points. WhatPulse users spend more time at a computer than the average person, so read these numbers as an early signal, not a census. Keystrokes are counted, never recorded. All builds and versions of an app are combined into one product. Only native desktop apps are included. AI used in a browser counts as browser usage and is not part of these numbers. Claude Code runs in the terminal, so we report its reach but not its keystrokes. Some usage on computers shared by multiple WhatPulse installs may be slightly undercounted.

We're making a few changes to WhatPulse Leagues, starting this week. The final sprint is now exactly 48 hours, results now lock on Monday afternoon instead of at midnight, and Pulse Score has a new name: League Score. Here's what's changing and why.

The final sprint is now 48 hours​

The final sprint used to start on Friday at 18:00 CET. It now starts at midnight between Friday and Saturday, so it covers exactly the weekend. As before, rival scores are hidden during the sprint while positions stay live. You'll know where you stand, just not by how much.

Results now lock on Monday at 14:00 CET​

Until now, a round was finalized right after Sunday ended in Europe. That was too early for some players.

The WhatPulse app sends your stats in the background, and they don't always arrive straight away. A computer might have been offline for a while, or a slow connection might delay the upload. Stats from the last hours of the week could reach us after the round was already locked, and they never counted.

To fix this, every app now has until Monday 14:00 CET to send its stats for the week. Then the results lock.

To be clear about what this means:

• Only Monday to Sunday counts. The extra time is only for stats from the week that just ended to arrive. Nothing you do on Monday counts toward the previous round.
• Monday's activity counts toward the new round. That's how it always worked, and it still does, even before your new group appears.
• Your week follows your own computer's clock. It runs from Monday to Sunday in your local time, so every timezone gets the same seven days.

What you'll see on Monday morning​

Between Sunday midnight and Monday 14:00 CET, your league is in its photo finish. Scores stay hidden, the standings are marked as provisional, and a countdown shows when the results lock. Positions can still change as late stats arrive.

At 14:00 CET your results are final. Promotions, relegations and your new group arrive together, followed by your weekly email.

Pulse Score is now League Score​

The name Pulse Score caused confusion. It sounded as if the score came from your pulses, but it doesn't. It's calculated from the hourly stats the WhatPulse app records, the same data behind your overview and productivity dashboards.

So we've renamed it League Score. Only the name changes; the calculation stays the same:

▎ League Score = keys + clicks × 3

The same goes for your existing scores, ranks and history. If you show your score on your profile, you can now use %LeagueScore%.

Why your score doesn't change when you pulse​

This is the question we hear most. Your League Score isn't updated the moment you pulse. Leagues are recalculated once an hour, on the hour, from the stats your app has sent. New activity shows up at the next update. Your league page shows when that is.

In short​

The full rules are on the Leagues page (https://whatpulse.org/leagues). Good luck this weekend.

The best typing website is the one that helps you practice consistently and verify whether your real keyboard habits improve. A one-minute typing test can measure speed, but better practice also tracks accuracy, comfort, mistake patterns, and whether normal computer use changes afterward. Use typing websites for drills, then compare the results with your daily keyboard activity in WhatPulse.

That last part matters. A fast test score feels good, but the useful question is whether you type more comfortably during work, games, school, chat, code, or writing. The sites below cover quick checks, lessons, repetition, custom text, races, and long-form stamina. Pick the site that matches your goal, then use measurable computer activity to keep yourself honest.

What makes a typing website useful?​

A useful typing site gives you a clear exercise, immediate feedback, and enough variety to return tomorrow. The site does not need every feature. It needs to answer a few practical questions.

Can you see words per minute and accuracy separately? Does it show which keys or words cause mistakes? Can you change duration, punctuation, capitalization, or text difficulty? Does it support your keyboard layout?

That is where pairing practice with self-tracking helps. WhatPulse can show keyboard activity over time in the WhatPulse app, while the public stats pages show broader activity patterns. If you are starting from scratch, install the client from WhatPulse downloads, practice normally for a week, and compare daily totals.

Top typing websites by practice goal​

This table is a starting point, not a universal ranking. A programmer practicing punctuation needs a different site than someone learning home-row basics. A gamer may enjoy pressure-based practice. A writer may get more value from long passages that reveal fatigue.

1. Monkeytype for customizable speed tests​

Monkeytype lets you change the test instead of accepting a single default. You can practice common words, punctuation, numbers, quotes, longer durations, and different layouts. That flexibility makes it useful for people who already type regularly and want cleaner measurements.

Use it when you want to compare settings. Run a 30-second, 60-second, and two-minute test. If your short score is high but the longer test drops sharply, the issue may be rhythm, accuracy, or hand tension rather than raw keyboard familiarity.

Track the boring details. In WhatPulse, compare normal keyboard activity on practice days with non-practice days. If a test session adds thousands of keystrokes but your real work still feels clumsy, change the practice text instead of chasing another personal record.

2. Keybr for weak-key repetition​

Keybr focuses on adaptive typing practice. It introduces letters gradually and pushes you toward the keys that need work. This can feel slower than a speed test, which is useful when your goal is to remove friction from specific parts of the keyboard.

Use Keybr if you make repeat mistakes with certain letters, alternate layouts, or newly learned touch-typing habits. Speed tests can hide weak keys because you compensate with familiar words. Letter-focused drills make those gaps harder to ignore.

Pair it with a weekly review. Look at your WhatPulse keyboard activity, then ask whether your practice volume was enough to matter. A few careful sessions beat one giant session that leaves your hands tired. The mouse and keyboard tracker guide explains how input data can support that kind of review.

3. TypingClub for structured lessons​

TypingClub works well for people who want a curriculum. It is useful for beginners, students, or anyone rebuilding fundamentals after years of improvised typing. Lessons give you a sequence, so you spend less time deciding what to practice.

Lesson completion can look like progress even when accuracy suffers, so treat progress bars as one signal and comfort as another.

4. Nitro Type for competitive practice​

Nitro Type turns typing into races. Competition can make practice easier to start and harder to stop. It is a good fit if you need energy, streaks, and a reason to repeat drills.

Pressure reveals habits that calm tests miss. Some people type faster in races. Others make more mistakes because they chase the race instead of the next word. Watch accuracy closely. A lower speed with fewer errors may be more useful than a frantic win.

If competitive practice creates long sessions, use WhatPulse to spot the change. A game-like typing site can blur into regular computer time. Uptime and input totals help separate a useful session from another “one more race.”

5. 10FastFingers for quick benchmarks​

10FastFingers is built for fast benchmark checks. It is useful when you want a simple words-per-minute snapshot without much setup. Short tests make it easy to repeat the same measurement over time.

Treat it like a thermometer, not a training plan. A quick benchmark can show whether you are warmed up, distracted, or improving across weeks. It will not teach form by itself.

For cleaner comparisons, test at the same time of day and use the same language setting. Then compare your typing website score with broader computer activity. If a week contains heavy writing, coding, or chat, your hands may already have done plenty of work before the benchmark starts.

6. TypeRacer for natural text under pressure​

TypeRacer uses real text snippets and races against other people. This helps practice flow, capitalization, punctuation, and sentence rhythm. Natural text feels closer to writing and coding than random word lists.

Use TypeRacer when you want to see how well accuracy survives real sentences. Random word tests reward pattern recognition. Natural passages force you to deal with commas, quotes, names, and uneven word lengths.

If your goal is practical typing, natural text matters. The Microsoft Windows keyboard shortcuts documentation also shows how much keyboard efficiency depends on command habits, not only alphabetic speed.

7. Ratatype for basics and certificates​

Ratatype combines lessons, tests, and certificates. It suits learners who want a clear baseline and a tidy way to show progress. The certificate angle can help if you need a simple external proof of typing speed.

Use it for fundamentals. If you are relearning touch typing, start with form and accuracy before chasing speed. Certificates are useful only when the underlying habit holds up during normal work.

This is another place where WhatPulse can keep the story grounded. A certificate records a test result. Your activity history shows whether keyboard use stayed consistent after the test.

8. Typing.com for beginner and classroom practice​

Typing.com is a broad, lesson-heavy platform with beginner-friendly practice. It works well for schools, families, and people who want guided exercises instead of raw benchmark tests.

The best use case is habit building. Short daily lessons can be easier to maintain than occasional intense practice. If you are helping someone else learn, progress reports and predictable exercises reduce friction.

Use a simple checklist: practice for a set time, stop when accuracy drops, review mistakes, and take a break. Apple’s Mac keyboard shortcuts guide is worth bookmarking because everyday keyboard confidence often comes from navigation as much as text entry.

9. ZType for game-like repetition​

ZType turns typing into an arcade-style game. It is useful when standard drills feel too dry. The visual pressure encourages quick recognition and repeated attempts.

Use it for variety, not your only measurement. Game mechanics can make you faster at that game without improving longer writing sessions. Keep one neutral benchmark in your rotation so you can compare progress.

10. Typelit for long-form typing stamina​

Typelit uses books and long passages for practice. It is useful for writers, students, and anyone who wants to improve sustained typing rather than short bursts.

Long-form practice exposes fatigue. Your first minute may look fine while your tenth minute tells the real story. Watch accuracy, comfort, and repeated corrections.

If your keyboard has seen years of heavy use, hardware also enters the picture. The keyboard lifespan guide explains why total keystrokes are useful context for people who care about switches, durability, and replacement timing.

11. TypeTest.io for a clean modern typing test​

TypeTest.io keeps the loop simple: start a test, watch speed and accuracy, and return often enough to build a baseline. It is a good last stop when you want a focused typing test without turning practice into a setup project.

Use it for repeatable checks beside your WhatPulse keyboard history. One clean test shows the drill result; your WhatPulse activity shows whether practice fits real computer use.

How to choose the right typing site​

Start with the problem you actually have.

• Choose Monkeytype or 10FastFingers if you want fast benchmark checks.
• Choose Keybr if certain keys keep causing errors.
• Choose TypingClub, Typing.com, or Ratatype if you want lessons and structure.
• Choose TypeRacer or Typelit if you want natural text or longer passages.
• Choose TypeTest.io if you want a clean repeatable benchmark.
• Choose Nitro Type or ZType if you need motivation from games or competition.

Rules like that prevent taking fifteen tests, keeping the best score, and calling it science.

How to track whether practice works​

Use typing websites for the drill and WhatPulse for the background trend. Before you start, record a baseline week. Note average daily keystrokes, main computer activities, and any obvious discomfort. After two or three weeks of practice, compare the same signals.

Look for three changes.

1. Your typing test accuracy improves at the same or higher speed.
2. Your normal keyboard-heavy days feel less tiring.
3. Your daily activity pattern stays sustainable instead of spiking during practice and collapsing afterward.

You do not need perfect data. You need comparable data. Keep the same benchmark site, test length, and review schedule. That creates a cleaner picture than switching sites every time a score disappoints you.

Final takeaways​

The top typing websites solve different practice problems. Some measure speed, some build fundamentals, some make repetition less dull, and some test whether you can keep accuracy under pressure. Choose based on your current bottleneck, then track progress with the same benchmark and a simple activity review.

If you want the practice to connect with real computer use, pair your typing site results with WhatPulse. Speed tests show what happened during a drill. Your keyboard activity history shows how typing fits into the rest of your day.

Mouse distance statistics measure how far your pointer travels across your screens during normal computer use. A typical day can include thousands of small pointer movements, but the useful question is not the exact distance. It is what the pattern says about your setup, work style, games, and habits. Mouse travel rises when you switch windows often, work across large displays, edit visual material, browse densely packed tools, or play pointer-heavy games. Tracking it over time gives you a practical baseline for comparing days, devices, and routines.

For WhatPulse users, mouse distance sits beside clicks, scrolls, keys, uptime, application use, and website activity. That combination turns a curiosity metric into a readable picture of how you actually use your computer.

What mouse distance statistics measure​

Mouse distance is the accumulated movement of your pointer. If your cursor moves from one side of the screen to the other, that movement adds to the total. Over a day, those tiny paths accumulate into meters, kilometers, or miles depending on your settings.

The number is shaped by three things: your physical mouse movement, your pointer speed settings, and your screen layout. A user with two large monitors may record more pointer travel than someone on a single laptop display, even if they spend the same amount of time at the computer. A gamer with low sensitivity may move their hand farther on the desk, while the on-screen pointer distance depends on the game and input mode.

That is why mouse distance statistics work best as a personal trend, not a universal ranking. Compare your Monday to your Friday. Compare a coding day to a design day. Compare a normal gaming session to a marathon one. You will learn more from your own baseline than from someone else's total.

WhatPulse added mouse scrolls and distance in version 5.0, including per-application views and historical tracking. The release notes explain the feature in WhatPulse 5.0: Mouse Scrolls, Distance, oh my, and the broader setup advice in Mouse and Keyboard Tracker: A Practical Setup Guide shows how to treat input data as a habit signal.

Why mouse travel varies so much​

Two people can spend eight hours on a computer and produce very different mouse distance totals. The difference often comes from workflow design rather than effort.

A developer who lives in a terminal and editor may produce more keystrokes than mouse movement. A designer moving objects, panels, and timelines may create long pointer paths. A spreadsheet user can generate bursts of precise movement between cells, menus, and filters. A gamer may record rapid movement during matches and almost none while waiting in a lobby.

Screen size also matters. A 4K display gives the pointer more pixels to cross. Multiple monitors add long transitions between work areas. High pointer speed can make the cursor cover more screen distance with less hand movement. Low pointer speed can make a physical desk workout without always increasing on-screen distance in the same way.

Mouse distance also reflects software layout. Dense interfaces with small targets encourage more corrections. The interaction principle behind that is described by Fitts's Law, which explains how target size and distance affect pointing time. In plain terms, small distant buttons cost more movement and attention.

A practical way to read your own numbers​

Start by collecting at least two weeks of data. One day is trivia. Two weeks starts to show rhythm. A month is better if your work changes by weekday, project phase, or gaming schedule.

Use this table to interpret mouse distance without pretending the number explains everything:

The key is to pair mouse distance with context. WhatPulse's mouse click statistics article covers a related input signal, while Computer Activity Baseline: What's Normal? explains why a personal baseline beats a generic benchmark.

Mouse distance and ergonomics​

Mouse travel can hint at friction in your setup, but it should not replace ergonomic judgement. A high total does not automatically mean injury risk. A low total does not mean your workstation is comfortable. The better signal is a pattern that matches discomfort, fatigue, or repetitive strain.

If your mouse distance rises on days when your wrist or shoulder feels worse, look at the physical setup first. The UK Health and Safety Executive's display screen equipment guidance recommends arranging equipment so you can work comfortably and change posture. The Canadian Centre for Occupational Health and Safety also covers mouse placement, reach, and workstation layout.

A few practical checks help:

1. Keep the mouse close enough that your arm is not reaching all day.
2. Adjust pointer speed so common movements do not require exaggerated hand travel.
3. Put frequent controls closer together when software lets you customize panels.
4. Use keyboard shortcuts for repeated actions that do not need pointing.
5. Take short breaks when a high-movement task runs for hours.

Operating system settings matter too. Microsoft documents how to change mouse settings in Windows, including pointer speed and scrolling behavior. Small changes can make large movement patterns feel less clumsy.

What different users can learn​

Mouse distance statistics become more useful when you interpret them by role.

Developers may see low mouse distance during deep coding and higher totals during debugging, browser testing, issue triage, or design review. If mouse movement rises while keyboard activity falls, it may indicate more navigation and less production. That does not make the day worse. It gives the day a label.

Designers, video editors, and content creators often produce high mouse distance because their work depends on visual placement. For them, a high number may simply confirm the nature of the task. The useful comparison is between tools, projects, or days with different layouts.

Gamers can use mouse distance to separate play styles. A shooter session may show different movement from a strategy game, MMO, or idle game. Combine distance with clicks and session length to see whether a game is intensive, casual, or mostly social.

Remote workers can use mouse distance with application and website patterns to understand where attention went. A day full of meetings might show lower mouse travel but high uptime. A day spent hopping between documents, dashboards, and chat tools may show the opposite.

Keyboard enthusiasts can compare mouse distance against key totals. A keyboard-first setup usually lowers pointer use for common actions. If a new shortcut habit reduces mouse distance during repeated work, the data gives you proof beyond vibes. Vibes have a weak audit trail.

How to build a mouse distance baseline​

A baseline is a normal range for you. It should account for workdays, weekends, games, and unusual projects. Do not start with a target. Start with observation.

Use this checklist:

• Track at least 14 days before drawing conclusions.
• Mark unusual days, such as travel, sick days, all-day meetings, game nights, releases, or design deadlines.
• Compare mouse distance with clicks, scrolls, keys, uptime, applications, and websites.
• Split workdays and weekends before averaging.
• Look for repeatable patterns by day of week or activity type.
• Recheck the baseline after changing monitors, pointer speed, mouse hardware, or job role.
• Treat spikes as prompts for review, not automatic problems.

WhatPulse's app helps because it puts input activity beside application, website, network, and uptime stats. You can also install or update the client from WhatPulse downloads if you want fresh data from the computer you actually use.

Common mistakes when comparing mouse distance​

The first mistake is comparing raw totals across people. That turns a personal metric into a scoreboard with bad rules. A triple-monitor designer and a laptop writer do different work on different surfaces.

The second mistake is treating mouse distance as productivity. Movement is activity. It may support output, but it can also come from searching, correcting, browsing, gaming, or fiddling with windows. Pair the number with application and website context before deciding what it means.

The third mistake is ignoring hardware changes. A new mouse, sensitivity setting, monitor, desk, or operating system setting can shift your totals. If your baseline changes overnight, check the setup before inventing a behavioral story.

The fourth mistake is focusing only on daily totals. Hourly patterns often explain more. A day with one intense design block can have the same total as a day of scattered window switching. The total is the headline. The timeline is the evidence.

A simple example​

Suppose your month shows three patterns. Mondays have low mouse distance and high keyboard activity because you write plans and code. Wednesdays have high distance, high clicks, and high website usage because you review dashboards and support tickets. Saturdays have shorter uptime but very high mouse movement because you play games.

None of those days is automatically good or bad. They are different shapes of computer use. Once you know the shapes, you can make targeted changes. You might move repeated support tools into a cleaner layout. You might add shortcuts for common dashboard actions. You might leave gaming alone because it is doing exactly what leisure should do: absorbing time in a measurable way.

The useful question​

Mouse distance statistics answer a surprisingly human question: what kind of computer day did I just have? The raw distance is fun, especially when it turns into kilometers over time, but the pattern matters more.

Use mouse distance as one layer in your activity history. Compare it with clicks, scrolls, keys, uptime, applications, and websites. Watch for changes after new hardware, new work, new games, or new habits. The result is not a perfect measure of effort. It is a practical record of how your hands moved through your digital life.

If you already use WhatPulse, check your mouse distance alongside your other input stats and build a baseline from your own history. If the number surprises you, good. Curiosity is the whole point.