A team that never looks back keeps making the same mistake, just faster. A person who never looks back does the same, only no one is in the room to notice.
Software teams figured this out years ago. At the end of every sprint they stop shipping for an hour, sit in a room, and ask three blunt questions: what went well, what went badly, what should we change. No blame, no theatre, just a structured look in the mirror before the next two weeks begin. It is one of the oldest practices in agile, and it survives because it works. The strange thing is that almost nobody runs it on themselves. We will happily improve a product through relentless reflection while leaving our own week completely unexamined. This piece is about closing that gap: how to take the team retrospective, shrink it to a party of one, and run it every week as a small, durable habit.
Why your life deserves a retro
You are already a project, you just never scheduled the review. Your week has scope, deadlines, stakeholders, a budget of time and energy, and outcomes that drift from the plan. Every one of those is the kind of thing a project would inspect on a cadence. Yet most people's only feedback loop is a vague Sunday-night dread or a once-a-year resolution that evaporates by February. The retrospective fixes the cadence problem. It puts a small, regular checkpoint between you and the long stretches of autopilot where bad habits quietly compound.
The magic is not insight, it is frequency. A single grand annual reflection is almost useless, because the signal is twelve months stale by the time you act. You cannot remember why Tuesday in March felt awful, so you cannot fix it. A weekly retro catches the lesson while it is still warm. The Toyota Production System, which gave agile much of its DNA through Taiichi Ohno's thinking, treated continuous small improvement as the whole game: lots of tiny corrections beat one heroic overhaul. Your week works the same way. Twenty minutes every seven days outperforms a four-hour soul-search every December, and it does so by a wide margin.
Solo removes the one thing that ruins team retros: politics. In a group, people soften the truth to protect feelings or reputations. Alone, there is no audience to perform for. The only person who benefits from your honesty is you, and the only person who pays for your evasion is also you. That alignment is rare and valuable. Use it.
The three questions, kept simple
Start with the classic trio and resist the urge to complicate it. What went well. What did not. What will I try next. That is the entire framework, and its plainness is a feature. The moment you add a fourteen-point template with mood scores and gratitude logs and a colour-coded matrix, you have built something you will abandon by week three. The retro you actually run beats the elegant one you admire and skip.
What went well is not a victory lap; it is reconnaissance. The point of naming good things is to find what to repeat. If your best work happened on the morning you closed your inbox until noon, that is not a happy accident to smile about, it is a pattern to protect. Most of us are far better at cataloguing failure than success, so we keep fixing the negative while accidentally dismantling the positive. Spend real attention here. Ask what conditions produced the good moments, because those conditions are the most actionable thing in the whole review.
What did not go well needs a soft entry and a hard look. The trap is either brutal self-flagellation or comfortable excuse-making, and both kill the ritual. The fix is to describe events, not character. "I skipped the gym four times" is a useful observation. "I am lazy and hopeless" is a mood, and moods do not generate experiments. Kahneman and Tversky spent careers showing how badly the mind narrates its own behaviour, leaning on stories instead of facts. Your retro is a small countermeasure: write down what happened, in concrete terms, before your inner storyteller gets to spin it.
What will I try is the only part that changes the future. A retrospective that ends in observations is just journaling. The deliverable is a single, specific experiment for next week. Not five. One. "Try a hard stop at 18:00 on weekdays" is an experiment. "Be more balanced" is a wish. Treat each change as a hypothesis you are testing, which means next week's retro opens by checking whether it worked. That feedback loop, the trying then the reviewing, is what separates this from yet another well-meaning list.
Start, Stop, Continue: a sharper variant
When the three questions feel stale, switch to Start, Stop, Continue. It is the same spirit in a more action-shaped frame, and many people find it easier because every entry is already a verb. You are not describing the week, you are issuing instructions to next week's version of yourself. Start names a behaviour to add. Stop names one to drop. Continue names one to defend, which is the column we forget and the one that quietly carries most of the load.
Keep each column to one or two items. The instinct is to dump everything, which produces a wall of intentions you cannot possibly honour. A retro that asks for ten changes gets zero. Cap yourself, force the prioritisation, and let next week's review surface the rest. Anything important that you drop will simply reappear, more loudly, seven days later, which is exactly the feedback the system is meant to give you.
Make it measurable, lightly
A retro improves when it has a little data to argue with. Memory is a flattering liar, especially about how a whole week felt. So track one or two cheap numbers and glance at the trend during your review. Not a forensic time-log, just a single daily reading you can capture in five seconds. The quantified-self idea is at its best here: not surveillance, just enough signal to keep your stories honest.
The reading I would suggest is a daily energy or focus score, one to ten, jotted before bed. Over a few weeks it forms a line, and a line is far harder to misremember than a feeling. You start seeing the real shape of your weeks: the Wednesday slump, the lift after a good night's sleep, the slow drain of an overbooked Thursday.
Let the number guide the question, not replace it. A dip on the chart is not a verdict, it is a prompt: what happened that week. The data points you at the right conversation; the retro is still where the thinking happens. Numbers without reflection are trivia, and reflection without numbers is guesswork. You want the pair.
Keeping the ritual alive
Pick a fixed slot and defend it like a meeting with someone important. Friday afternoon and Sunday evening are the popular choices, one to close the work week and one to set up the next. It barely matters which you choose; it matters enormously that it is the same time every week. Parkinson's law tells us work expands to fill the time available, so give the retro a hard twenty-minute box and it will fit neatly. Give it an open afternoon and it will either bloat into self-indulgence or, more likely, never start.
Write it down somewhere you will actually reread. A note app, a paper journal, a single running document, anything you will open next week. The rereading is the whole point: last week's "what I will try" is this week's first agenda item. Tiago Forte's idea of a second brain applies in miniature here. You are building a tiny external memory of your own experiments so that you stop relitigating the same lessons from scratch.
Forgive the misses and keep the cadence. You will skip weeks. You will run a lazy retro that produces nothing. That is fine, and it is not a reason to quit; a retro you do badly this week still beats the one you abandon forever. The value is in the streak of attention, not the polish of any single session. Miss one, shrug, show up next Friday.
A team improves because it pauses to look. You can give yourself the same gift, alone, in less time than a coffee break. Three questions, one experiment, one small number, every single week. Do it for a couple of months and the change is not that you suddenly become disciplined; it is that you finally start learning from the one project you can never quit.