Guide
A personal retrospective is a short, scheduled review of your own work: what happened, what helped, what got in the way, and what to change next. A notebook is enough to run one. If memory is the weak spot, Dayflow (free, Mac) writes your day into a timeline. Timing suits Mac users who also bill hours. ActivityWatch is a free local log on any desktop. Rize covers Mac plus Windows.
A personal retrospective is a short, scheduled review of your own work over a fixed period, usually a week. You look at what happened, name what helped and what got in the way, and pick one or two changes to try next.
People get it wrong in two directions. Some treat it as a feelings journal and finish with a vague sense that the week was “busy.” Others copy the full team ceremony from agile software work, with sticky notes, dot voting and a facilitator script, and drop it after a fortnight because reviewing five days takes an hour. The version that lasts sits in between: about twenty to thirty minutes, built on a record of what you did instead of what you remember doing.
A note before we start: Dayflow, one of the tools in the section near the end, is our product. Everything else in this guide works with a pen and a calendar.
The word comes from software teams, who hold a retrospective at the end of each sprint to ask what went well, what did not, and what to change. A personal retrospective applies the same loop to one person and one calendar.
It has five parts:
What it leaves out is just as important. Nobody grades you. There is no score to beat and no audience. The output is a short note, often under a page, that says what happened, what you noticed, and what you will try. If you keep those notes in one file, after a few months you have a record of how you work that is far more useful than any single week’s reflection.
These four get mixed up because they all involve looking at your work. They differ in who does them, how often, and what comes out.
| Personal retrospective | Team retrospective | Time audit | Work journal | |
|---|---|---|---|---|
| Who | You alone | A team with a facilitator | You alone | You alone |
| How often | Weekly or monthly, ongoing | End of each sprint | Once, over one or two weeks | Daily |
| Input | Your records for the window | Shared memory and team metrics | A detailed log kept on purpose | What you write that day |
| Output | One or two experiments | Team action items | A picture of where time goes | Entries to read later |
A time audit is a measurement exercise. You log everything for a short stretch to answer one question: where does the time go? A personal retrospective can use an audit’s results, but it repeats indefinitely and cares about cause and change, not just totals.
A work journal is a raw material. You write in it daily. A retrospective is the moment you read those entries back and decide what they mean. People who keep a journal but never review it have notes and no conclusions.
A team retro and a personal one share the loop but not the mood. In a team setting you soften your words for colleagues. Alone, you can say plainly that you spent Thursday avoiding a hard email, which is often the most useful sentence of the week.
The mechanism is a feedback loop: record, review, decide, test. Most of the value comes from closing that loop every week, so last week’s decision becomes this week’s first item to check.
Pick a fixed slot. Friday afternoon works for people who like to close the week. Monday morning works for people who want the review to feed straight into planning. Either is fine. What fails is “whenever I have time,” because the weeks that most need a review are the weeks with no spare time.
Gather evidence before you think. Open your calendar, your sent folder or chat history, your task tracker, and any commit or document history. Read them in order, day by day. This rebuilding step is what separates a retrospective from a mood check. If you regularly end days unsure what you did, our guide on how to remember what you worked on covers lightweight ways to leave yourself a trail.
Notice before you judge. As you rebuild each day, mark three kinds of moments: where you had energy and momentum, where you got stuck or stalled, and where something surprised you, such as a task taking twice as long as planned.
Decide small. A retro that ends with ten resolutions produces none. One or two experiments, each concrete enough that you will know by next week whether you did it.
Keep everything in one place. A single running document, newest week at the top, makes the monthly pass quick and lets patterns show up across weeks.
Start, Stop, Continue is the simplest retrospective framework, and the easiest to keep doing alone because every observation has to land in one of three action buckets. Here is a five-step version sized for a single person.
If Start, Stop, Continue starts to feel stale, two common variants use the same steps with different buckets. The 4Ls ask what you Liked, Learned, Lacked and Longed for, which suits weeks that were more about learning than shipping. Keep, Problem, Try is a tighter three-bucket version popular with teams that run very short cycles.
This is an invented illustration, not a real person’s data.
Sam is a backend developer who runs a Friday retrospective. This week’s rebuild, from calendar, commits and chat history, looks roughly like this:
Sam marks the moments. Momentum: Monday morning and Thursday morning, both with no meetings before noon. Stalls: Tuesday afternoon, where the chat window kept pulling attention during long test runs, and Wednesday’s fragmented schedule. Surprise: the migration was planned for one day and took most of three.
Sorting them:
Sam picks two experiments: protect three meeting-free mornings, and close chat whenever a test run starts. The surprise goes in the notes too, with a line about breaking migrations into smaller estimates. If underestimating keeps coming up, the time estimation test is a quick way to compare your sense of time against a clock, and our piece on why people underestimate how long things take explains the usual causes.
Next Friday, Sam’s retro opens with those two experiments. Two of the three mornings held, the chat rule lasted two days. That partial result is still useful: it tells Sam the chat rule needs to be more mechanical, perhaps a separate desktop for test runs.
Use the same short set every week so the answers are comparable. These work for most kinds of desk work:
If you want a quick outside view before your first retro, the where does my week go calculator gives a rough split of your week to compare against what your records show.
Working from memory. Memory tends to compress a week into its last day and its worst moment. Without evidence you end up reviewing your feelings about the week, not the week.
Resolving too much. Ten action items feel productive on Friday and are forgotten by Tuesday. Two is a sensible ceiling.
Vague experiments. “Focus more” cannot be checked. “No meetings before noon on Tuesday and Thursday” can.
Never closing the loop. If you do not check last week’s experiments, each retro starts from zero and the habit stops paying off.
Turning it into self-criticism. The point is to find causes, not to assign blame. A week where you avoided a task is a data point about the task, the timing or the context, not a verdict on you.
Skipping bad weeks. The weeks that went badly carry the most information. Keep the retro even if it is ten minutes long.
Letting the format grow. Extra sections, scores and charts creep in. If a retro takes longer than forty-five minutes, cut back to the five steps.
A retrospective needs a record of the week, and the hardest part to reconstruct is unscheduled time. Any time tracker for Mac can supply that record. The details below come from each vendor’s own documentation as checked on October 4, 2026, and Dayflow’s published facts for version 2.2.0. We did not run our own tests.
The three tools besides Dayflow are Timing (see our Timing comparison), Rize (see our Rize comparison) and ActivityWatch (see ActivityWatch vs Dayflow).
| Tool | What it does for this | Price |
|---|---|---|
| Dayflow | Records the screen at one frame every 10 seconds and has AI write the day into a readable timeline; exports a date range as Markdown for your retro notes. Mac only (macOS 14 or later). | $0, MIT licensed; optional Pro at $20 a month |
| Timing | Records time automatically across apps, websites and documents on the Mac, down to specific URLs and Mail threads; AI activity summaries on all plans; data stored locally unless you opt into sync. | Professional $9 per month billed annually or $11 per month billed monthly; Expert $11 billed annually or $14 billed monthly. No free tier, 30-day trial |
| Rize | Tracks apps and websites automatically with AI categorization, AI chat and focus insights; runs on Mac and Windows; data hosted in the cloud on AWS. | Basic $9.99/mo billed annually or $12.99/month billed monthly; Pro $23.99/mo billed annually; Max $39.99/mo billed annually. No free tier, 7-day trial |
| ActivityWatch | Logs the active app, window title and away-from-keyboard time on your own device; browser extensions add tab titles and URLs. Open source (MPL-2.0), no AI summaries listed. | Free |
Dayflow is our product. It fits the evidence step of a retrospective when the gap in your week is “I was busy but cannot say on what,” because the timeline describes what was on screen rather than only which app was open. You can export the week as Markdown and paste it into your retro note, or follow the export work history workflow. With local AI through Ollama or LM Studio, screenshot analysis stays on your Mac; with your own cloud provider or Dayflow Pro, selected screenshots are processed off-device. AI summaries can miss or misinterpret work, so treat the timeline as a draft to check, not a verdict.
It does not fit if you need invoices, payroll approvals, team reports or website blocking, because Dayflow has none of those.
A personal retrospective is a short, repeated review of your own work that ends in a small, testable change. Keep the window fixed, the time box tight and the questions the same every week. Rebuild the week from evidence before you judge it, sort what you find with a simple framework like Start, Stop, Continue, and pick no more than two experiments.
The habit is what makes it work. A thirty-minute retro every Friday for a few months will tell you more about how you work than any single long reflection, because each week checks the last. Start with your calendar and a blank document this Friday. Add a tool later, if and when the gaps in your record become the thing you most want to fix.
Weekly works for most people. A week is long enough to show patterns and short enough that you still remember the context behind them. Add a lighter monthly pass that only reads your four weekly notes and asks which experiments stuck. Daily retrospectives tend to collapse into to-do planning, and quarterly ones rely too much on memory to be useful on their own.
Plan on twenty to thirty minutes for a weekly retro. About half of that goes to rebuilding the week from evidence, the rest to sorting what you noticed and choosing one or two changes. If it regularly runs past forty-five minutes, you are probably rebuilding the week from memory or trying to fix too many things at once. Set a timer and stop when it rings.
The weekly review, as David Allen describes it in Getting Things Done, is mainly about getting your task lists and commitments current so next week starts clean. A personal retrospective looks at how the past week actually went and asks why. Many people run both in one sitting: the retrospective first, so its decisions shape the planning that follows.
Yes. Your calendar, sent messages, commit history, closed tickets and browser history together give a reasonable outline of most weeks. The gap is unscheduled time: the afternoon that disappeared into small fixes or context switching. If that gap is where your questions sit, some kind of passive record helps, but it is an upgrade, not a requirement for starting.
It is the easiest one to begin with because it forces every observation toward an action. Some people prefer the 4Ls (Liked, Learned, Lacked, Longed for) for weeks that were more about learning than output, or Keep, Problem, Try for a tighter loop. The format matters less than ending every session with one specific experiment you can check next week.
Treat it as a sign the experiment was too vague or too large. Rewrite it smaller and more mechanical, for example closing the chat app before a work block instead of resolving to get distracted less. If the problem still repeats after three or four honest attempts, it may sit outside your control, such as a meeting load, and the retro's output becomes a conversation with someone else.
ActivityWatch is the better fit there, and Dayflow cannot help because it runs only on macOS 14 or later. ActivityWatch is free, open source under MPL-2.0, and runs on Windows, Linux, Mac and Android. It logs the active app, window title and away-from-keyboard time on your own device. It does not list AI summaries, so you read the raw activity yourself during the retro.
Usually not the raw notes. A personal retrospective works because you can be blunt about what went badly, and that honesty fades if you write for an audience. Share the outputs instead: the change you are trying, or a blocker you want help removing. If you owe a regular update anyway, draft it from the retro notes as a separate, edited document.
Full screen recordings and the local timeline are stored on your Mac. With local AI through Ollama or LM Studio, screenshot analysis stays on-device too. If you choose your own cloud provider or Dayflow Pro, selected screenshots are processed off-device. Under the privacy policy, Pro deletes screenshots after processing and may retain activity data for sync. Your own provider has its own terms.
Timing stores tracked time locally unless you opt into sync and starts at $9 per month billed annually, with a 30-day trial. Rize hosts data in the cloud on Amazon Web Services, runs on Mac and Windows, and starts at $9.99/mo billed annually, with a 7-day trial. Choose Timing if you also bill hours, Rize if you switch between Mac and Windows.
Dayflow is $0 and open source. Full recordings stay on your Mac, with local AI available for on-device analysis.
Download for MacOS macOS 14+ · MIT licensed · no account required