How to Track How Much Time You Spend Coding (2026 Guide)

Install WakaTime if you only want time inside your editor, sorted by project and language. Its free plan keeps one week of history. Use ActivityWatch for a free, local log of active apps and windows. On a Mac, use Dayflow to see the whole day, including docs, reviews and browser debugging, written up as a timeline. Timing is the better choice when you bill the hours.

You want an honest number: how many hours this week went into building software, and where the rest went. Most attempts fail in one of two ways. You install an editor plugin, see a daily total and treat it as your coding day, even though the debugging in browser devtools, the pull request review and the hour of reading docs never reached the editor. Or you start a timer, forget to stop it at lunch, and give up by Thursday.

This guide gives you a method that works with a notebook, an editor plugin or an automatic tracker. Disclosure: Dayflow is our product, and it is one option for one of the steps below. Competitor details come from each vendor’s own documentation as read on October 4, 2026. This guide compares what vendors publish. Nothing here comes from running the tools side by side.

The short version

1. Write down what counts as coding for you.
2. Pick a capture method that can actually see those activities.
3. Set up project names so time lands in the right bucket.
4. Record one ordinary week without changing how you work.
5. Write a two-line log at the end of each day.
6. Review the week against your guess and change one thing.

What you need before you start

• A list of your active projects. Use repo names, client names or product areas, whatever you would naturally say out loud.
• Ten minutes at the end of each workday. Every approach, automated or not, needs a short review.
• A decision about where data may live. If you work on an employer’s code or under a client NDA, decide whether activity data can go to a vendor’s cloud. This rules some tools in or out before you install anything.
• A guess. Before you measure, write down how many hours a day you think you code. You will compare against it in step 6.

Step 1: What counts as coding time for you?

“Coding” is not one activity. Before choosing a tool, write a short definition. A developer’s work usually falls into a handful of buckets:

• Writing new code
• Reading and reviewing code, including pull requests in a browser
• Debugging, which often happens in a browser, a terminal or a log viewer
• Working with an AI agent: prompting, reading diffs, steering
• Research: docs, issues, Stack Overflow, internal wikis
• Talking about code: design meetings, pairing calls, chat threads

Decide which of these count as “coding” in your numbers and which count as “around coding.” There is no right answer. There is only one you apply consistently.

Illustration: A backend developer decides that writing, debugging and agent sessions count as coding. Review and research go in a separate bucket called “engineering support.” Meetings stay out of both.

Mistake to avoid: Letting a tool’s default become your definition. An editor plugin measures editor time, and a window tracker measures app focus. Neither one knows what you meant by coding.

Step 2: Which tool can actually see that work?

Tools capture work at three levels, and each level has a different blind spot. Match the level to the buckets you defined in step 1. For a wider survey, see our roundup of Mac time trackers.

Editor-level: WakaTime

WakaTime uses open source plugins for more than 100 editors, IDEs and terminals. They record coding time per project, language and editor and send it to a web dashboard. It also tracks AI coding usage across tools such as Claude Code, Cursor, Copilot and Codex. That includes adoption, model usage and spend, and AI-generated versus human-written lines.

Its limit is built into the design: it only captures activity inside supported editors and terminals. The data goes to WakaTime’s hosted dashboard. The free plan keeps 1 week of dashboard history, and full history requires Premium.

App and window level: ActivityWatch, Timing, Rize

ActivityWatch records the active application, the window title and whether you are away from the keyboard. Browser extensions add the tab title and URL, and editor plugins add coding activity. It is free, open source under MPL-2.0, and stores everything on your device. We compare it directly in ActivityWatch vs Dayflow.

Timing is a Mac-only automatic tracker. It records apps, websites and documents, down to specific URLs. It lists AI activity summaries and stores data locally unless you opt into sync. See our Timing comparison.

Rize tracks apps and websites on Mac and Windows and uses AI to categorize and tag them. Its data is hosted on Amazon Web Services. See our Rize comparison.

These tools see the browser tab where you review a pull request. What they record is app names, titles and URLs, so you still have to work out which project a given hour belonged to.

Screen level: Dayflow

Dayflow records the screen at one frame every 10 seconds and has AI write the day into a readable timeline. Because it works from what was on screen, it can describe a debugging session that moved between editor, terminal and browser as one block of work. Recordings and the timeline stay on your Mac. Analysis can run locally through Ollama or LM Studio, through a cloud provider you configure, or through Dayflow Pro. The details are in local and cloud processing.

Illustration: The backend developer from step 1 spends a lot of time debugging in the browser. An editor-only tool would miss that bucket entirely, so they choose an app-level or screen-level tool, or run one alongside WakaTime.

Mistake to avoid: Choosing a tool whose blind spot is your biggest bucket. If most of your week is review and debugging, an editor plugin will undercount it every day.

Step 3: How do you make the time land in the right project?

Raw time with no project attached tells you very little. Set up naming before the week starts.

• WakaTime groups time per project automatically from the files you work in, so clear repo names do most of the work.
• ActivityWatch records window titles. If your editor puts the project name in its window title, those titles become your project signal.
• Timing and Rize let you organize tracked activity into projects and categories inside each app. Rize uses AI to auto-categorize and tag.
• Dayflow writes its timeline from what is on screen, so descriptive repo, branch and ticket names make the summaries more specific.

Illustration: A freelancer has two clients with repos named api and web. They rename the local folders to acme-api and northwind-web so every tool, and their own end-of-day notes, can tell the clients apart.

Mistake to avoid: Putting everything into one “Work” bucket. If you lump it together, you cannot learn which project is quietly eating your week.

Step 4: Record one ordinary week

Install on Monday morning and leave it alone until Friday evening. Do not try to look productive. Work the way you normally work, because the point is to see your real pattern.

Pick a representative week. A week with an offsite, an on-call rotation or a launch will give you numbers that describe that week and nothing else.

If you are using a screen-based tool on a laptop, it is reasonable to wonder about power. We cover that in screen recording and battery impact.

Illustration: A developer starts recording on a Monday after a holiday. Half the week is catch-up email and planning, so they discard that week and start again the next Monday.

Mistake to avoid: Changing behavior mid-week because the numbers look bad. That is the experiment working. Let it finish before you react.

Step 5: Write a two-line log at the end of each day

Automatic data tells you where time went. A short note tells you why. At the end of each day, write two lines:

1. What moved forward today.
2. What took more time than it should have.

The daily work log template gives you a fixed place for these notes. Look at your tool’s view for the day while you write. That might be the WakaTime dashboard, the ActivityWatch timeline, or Dayflow’s written timeline. The log takes a minute when the data is in front of you and much longer when you work from memory. The same notes also help with standups, which we cover in automatic standup updates.

Illustration: “Moved: payment retry logic merged. Ate time: a flaky CI test, about two hours across three attempts, all in the browser and terminal.” An editor-only tool would show very little of that second line.

Mistake to avoid: Skipping days and filling them in on Friday. Memory compresses a whole afternoon of debugging into “fixed a test,” and the log loses its value.

Step 6: Review the week and change one thing

On Friday, put three things side by side: your original guess, the measured totals per bucket, and your five daily notes.

Look for three patterns:

• The gap between the guess and the measurement. If your guess was far off, try the time estimation test. It is a short exercise comparing your sense of time with a clock, not a diagnostic.
• Where the non-coding time went. The where does my week go calculator helps you add up meetings, chat and admin around the coding.
• Which bucket grew without anyone deciding it should. Review, CI babysitting and context switches are the usual candidates.

Save the week before it disappears. WakaTime’s free plan keeps only one week of dashboard history. Dayflow can export a date range as Markdown from Settings → Export, and we walk through that in export your work history. If you report upward, the weekly status report template turns the review into something you can send.

Illustration: A developer guessed they coded most of each day. The week shows long stretches of review and CI waiting. They change one thing: they batch reviews into two fixed slots a day.

Mistake to avoid: Treating the total as a score. More coding hours is not automatically better. The goal is to find time that went somewhere you did not choose.

Doing it by hand vs with a tool

Approach	Effort	What you get	Cost
Manual log or start/stop timer	High. You must remember to start, stop and label.	Exactly the buckets you define, with gaps wherever you forgot	Free
WakaTime (editor plugin)	Low after installing plugins	Coding time per project, language and editor, AI usage metrics, goals	Free plan with 1 week of history. Premium for full history is $14/mo billed monthly, or $12.83/mo billed $153.96 yearly
ActivityWatch (automatic, local)	Low to set up, plus time to interpret window titles	Apps, window titles, AFK status, tab URLs with extensions, all stored locally	Free
Timing (automatic, Mac)	Low, plus project setup	Apps, sites and documents, AI activity summaries, billing-oriented reports	Professional is $9 per month billed annually, or $11 per month billed monthly. 30-day trial
Rize (automatic, cloud)	Low	AI-categorized app and website time, focus insights	Basic is $9.99/mo billed annually, or $12.99/month billed monthly. 7-day trial
Dayflow (screen-based, Mac)	Low, plus a daily check of the AI timeline	A written timeline of whatever was on screen, across editor, browser and terminal	$0, MIT licensed. Optional Pro at $20 a month

As noted at the top, Dayflow is our product. It requires macOS 14 or later. It has no invoicing, payroll approvals, team reports or website blocking, and its AI summaries can miss or misread work. Check them before you share anything.

Common mistakes

• Measuring only the editor and calling it your workday. Editor time is real, but it is one slice. Review, debugging and research often happen elsewhere.
• Running two trackers and adding their totals. If you pair WakaTime with an app-level tracker, use one for editor detail and the other for the day’s shape. Their hours overlap, so do not add them.
• Letting free-plan history expire. On a one-week history limit, last month’s pattern is gone unless you saved it.
• Sending raw numbers to a manager without context. A low coding total in a heavy review week looks bad and means nothing. If tracking is being discussed at the team level, read time tracking without surveillance first.
• Trusting AI categorization blindly. Rize, Timing and Dayflow all use AI in some form, and any of them can mislabel a block. Spot-check during your daily log.
• Tracking forever with no question in mind. One deliberate week with a review beats months of data you never open.

Where Dayflow is the wrong pick

• You want precise coding metrics. Time per language, per editor and per project, programming goals and private leaderboards are what WakaTime does. Dayflow has none of these. WakaTime is the better tool.
• You want to measure AI coding adoption or spend. WakaTime tracks model usage, spend and AI-generated versus human-written lines across Claude Code, Cursor, Copilot and Codex. Dayflow does not.
• You are not on macOS 14 or later. ActivityWatch runs on Windows, Mac, Linux and Android for free. Rize covers Mac and Windows.
• You bill clients by the hour. Timing lists billing status tracking, report rounding and a GrandTotal invoicing integration depending on plan. Dayflow produces no invoices.
• A team needs shared dashboards. WakaTime’s Team and Business plans, Timing’s Connect plan and Rize’s Team plan are built for this. Dayflow has no team reports.
• You do not want your screen recorded at all. ActivityWatch logs app and window metadata locally, and WakaTime only sees editor activity. Either avoids screen capture entirely.

The bottom line

Decide what coding means for you before you pick a tool, then choose a tool that can see all of it. If coding means editor time, WakaTime is the direct answer, and its free plan is enough to start as long as you save your weeks. If you want a free, local record of every app and window, use ActivityWatch. If you bill clients from a Mac, Timing is built for that. If you want the whole day written up, including the debugging, review and research that never touch the editor, Dayflow does that at $0. Whichever you choose, record one ordinary week, write two lines a day, and change one thing on Friday.

Frequently asked questions

Does WakaTime track time spent outside my code editor?

WakaTime works through plugins for more than 100 editors, IDEs and terminals, and it only records activity inside the tools those plugins support. Reading documentation in a browser, reviewing a pull request on the web or talking a bug through in a meeting does not show up. To capture that work, pair it with an app-level tracker such as ActivityWatch or a screen-based tool such as Dayflow.

Is WakaTime free?

WakaTime has a free plan with 1 week of dashboard history, unlimited projects and 1 programming goal. Full history requires Premium, which costs $14/mo billed monthly, or $12.83/mo billed $153.96 yearly. Basic costs $9/mo billed monthly, or $8.25/mo billed $99 yearly. The Team plan has a 14-day trial with no credit card required.

What is the best free way to track coding time on a Mac?

Three options cost nothing. WakaTime's free plan covers editor time, but you only see one week of dashboard history. ActivityWatch is free and open source under MPL-2.0. It logs apps and window titles locally, and editor plugins add coding activity. Dayflow is $0 and MIT licensed. It records the screen at one frame every 10 seconds and writes the day into a timeline. Dayflow runs on macOS 14 or later only.

Is WakaTime or Dayflow better for tracking coding time?

WakaTime is better if you want precise coding metrics: time per project, per language and per editor, plus programming goals, team dashboards and tracking of AI tool usage and spend. Dayflow has none of those breakdowns. Dayflow fits when the question is what the whole working day looked like, including browser research and reviews. If editor metrics are all you want, WakaTime alone is the better choice.

Can I track time spent working with AI coding agents?

WakaTime lists AI coding usage tracking across tools such as Claude Code, Cursor, Copilot and Codex. It covers adoption, model usage and spend, and AI-generated versus human-written lines. Dayflow does not count lines or spend. Because it works from screen frames, its timeline can show when you were in a terminal or editor session with an agent. AI summaries can misread work, so check them.

How can I track coding time without sending data to the cloud?

ActivityWatch states that all data stays on your device. Timing stores tracked time locally unless you opt into sync. Dayflow keeps recordings and the timeline on your Mac, and its Local AI mode runs analysis on-device through Ollama or LM Studio. WakaTime sends activity to its hosted dashboard. Rize says its data is hosted on Amazon Web Services.

Can I use tracked coding time to bill clients?

Dayflow does not produce invoices or billing records. Timing is built for this. It lists billing status tracking, customizable reports with rounding, and a GrandTotal invoicing integration depending on plan. Timing starts at $9 per month billed annually, or $11 per month billed monthly, with a 30-day trial. Rize offers client reports in PDF and CSV on Pro and above.

Does Dayflow work on Windows or Linux?

No. Dayflow requires macOS 14 or later. ActivityWatch runs on Windows, Mac, Linux and Android and is free. Rize runs on Mac and Windows, but its vendor says Linux is not yet available. WakaTime works anywhere one of its editor or terminal plugins runs, with a web dashboard.

How much of a developer's day is actually coding?

There is no single honest number that applies to you, and figures quoted online rarely define coding the same way. The useful answer is your own. Write down what you count as coding, record one ordinary week with a tool that can see all of that work, and compare the result with your guess. That comparison is what tells you something.