You started a new diet in January. By March it was gone.

You bought a gym membership in February. By April you'd been three times.

You started meditating in March. You did it for nine days.

This pattern repeats because most people optimize improvement at the wrong layer. They focus on what to start — new habits, new programs, new tools — without ever auditing what's actually working. Without that audit, every new initiative gets layered onto an unexamined foundation.

The Sunday Audit is a 20-minute weekly practice that fixes this. It's the meta-system that makes every other system work. Operators who do this weekly stay consistent for years. Operators who don't, restart their lives every January.

What the audit actually is

20 minutes, Sunday evening. Same time each week. A notebook or document. Five questions, answered honestly.

This is not journaling. This is not goal-setting. This is operational review — the same kind any well-run business does weekly, applied to your own life and performance.

The five questions:

  1. What worked this week?
  2. What didn't?
  3. What's one thing I want to do differently next week?
  4. What's one thing I want to stop doing?
  5. What was the highest-leverage thing I did?

That's it. The whole protocol. The depth comes from answering honestly.

Question 1: What worked this week?

Most people skip this question because they're trained to focus on problems. But knowing what works is more useful than knowing what doesn't — you want to do more of what works.

Be specific. Not "I had a good week." Specifics:

The pattern that emerges over a few weeks tells you what your actual high-performance recipe is. Most people are surprised when they look back at 8 weeks of "what worked" entries. The pattern is usually different than what they thought.

For some people, training before work is non-negotiable. For others, evening training works better. For some, social meals fuel them. For others, social meals drain them.

The audit reveals the pattern. Then you can engineer your weeks around what actually works for you specifically, not what works for somebody else's "morning routine."

Question 2: What didn't?

This is the opposite mirror. Where did the week go sideways? Specifics again:

The trap here is generalizing. "I didn't work out enough" isn't useful. "I planned to work out Tuesday-Thursday-Saturday and only did Tuesday because Thursday I had a hangover from Wednesday drinks and Saturday I was tired from the week" is useful.

Specific failures lead to specific interventions. Generalized failures lead to generalized guilt.

Question 3: What's one thing I want to do differently next week?

ONE thing. Not five. ONE.

The biggest failure mode in self-improvement is trying to change too many things at once. Six new behaviors starting Monday = zero new behaviors by Friday. One new behavior starting Monday = a high chance it sticks.

The one thing should be:

If last week's failure was skipping workouts, the move isn't "work out every day next week." That's ambitious and won't stick. The move is "put gym clothes by the bed Sunday night. When the 5:45 alarm goes off, put them on before anything else." Just that. One change.

Question 4: What's one thing I want to stop doing?

This is the harder question. Most self-improvement focuses on additions. But your week already has too many things in it. Subtractions create more space than additions create capability.

Examples of useful stops:

Again, one thing. Stops are harder than starts because they fight against habit. Pick one stop per week. After 12 weeks you'll have eliminated 12 things that weren't serving you.

The compounding effect of stops is more powerful than the compounding effect of starts. A year of weekly subtractions makes someone unrecognizably more productive than a year of weekly additions.

Question 5: What was the highest-leverage thing I did?

This is the strategic question. Of everything you did this week, what produced the most disproportionate return?

A 90-minute conversation with the right person might be the highest-leverage thing. A 4-hour deep work block might be the highest-leverage thing. Saying no to something might be the highest-leverage thing.

The leverage might not be visible in the same week. The Henderson deal you closed Thursday was actually downstream of the lunch you had four weeks ago. The audit forces you to think in terms of high-leverage moves rather than busy activity.

After 12 weeks of identifying your weekly highest-leverage thing, you'll see a pattern. Most people's high-leverage activities cluster into 2-3 categories. Yours might be relationship-building. Yours might be deep technical work. Yours might be strategic thinking.

Whatever they are, those are the activities you want to protect and expand. Most of your other activities — the meetings, the emails, the calls — should be subordinate to your highest-leverage work.

Most professionals get this backwards. They protect the busy work and squeeze high-leverage work into the gaps.

When to do it

Sunday evening, 7-9pm. Not Monday morning. Not Sunday afternoon while doing laundry.

Sunday evening, with the week's perspective fresh and Monday's reset close enough to plan for, is the sweet spot. Most operators who run this practice do it 7-8pm Sunday with a glass of wine or tea, alone, in 20 minutes.

If Sunday evening is impossible — kids, social obligations, travel — find your equivalent. Friday afternoon at 5pm before logging off works for some people. Saturday morning during coffee works for others. The specific time matters less than the consistency.

The setup

20 minutes. A notebook works. A simple Google Doc works. An Apple Notes file works. Don't over-engineer this. Don't use Notion templates with 17 fields. The friction kills the practice.

Five questions, answered briefly. Maybe a paragraph each. Maybe two sentences each. Whatever's true that week.

Title each entry with the date. "Week of April 28" or whatever. Over time the entries become a record of your operational learning.

What you'll notice after 12 weeks

People who do this audit weekly for 90 days report:

This is the meta-system. It's not impressive in any single week. It compounds.

The thing most people get wrong

The Sunday Audit only works if it's honest.

Most people, when they first start doing this, write answers that are either too positive (humble-bragging about productivity) or too negative (self-flagellating about failures). Neither is useful.

The honesty bar is: would you write the same answers if you knew a smart trusted advisor was going to read them? If yes, you're being honest. If you'd hide the answers from them, you're not.

The audit is for you. It works only if you're truthful with yourself about what's actually happening. The first few weeks may feel uncomfortable. The compounding benefit shows up in week 6-8 when you start seeing patterns you didn't expect.

What to do tomorrow

Put 7:00pm Sunday on your calendar. Recurring. Title: "Sunday Audit."

Pick the medium — notebook, doc, whatever's lowest friction. Have it ready Sunday at 7.

Then for 20 minutes that night, answer the five questions for this past week. Don't overthink it. Two sentences per question is enough. The depth comes from consistency, not from any single entry.

Run this for 4 weeks before deciding whether it's useful. The benefits are not visible in the first week. They emerge around week 3-4 as patterns become clear.

Most operational practices that produce sustained change over a year look unimpressive on any given day. This is one of them. Twenty minutes a week. The compounded return after a year is worth more than almost any other single practice.