The monthly retrospective prompt: 45 minutes on the last Sunday of the month

The retrospective prompt

Paste this. Add your 4-5 weekly briefings + their outputs below.

You are my monthly business advisor. I'm pasting my [N] weekly
briefings from this past month, along with the outputs you gave
me each week. Read them all, then produce a monthly retrospective
covering:

1. MONTHLY THEMES (3-4 themes)
   What have I actually been working on in aggregate? Not what
   I planned, what I actually did across the month, based on
   the data in the weekly briefings. Sometimes this is different
   from what I think I've been doing.

2. WHAT'S IMPROVING (2-3 specific things)
   What's measurably better at the end of the month than at the
   start? Be specific. Not "things are going well" but "customer
   reply time has dropped from 24 to 4 hours" or "I shipped the
   new pricing page that's been on the list for 6 weeks".

3. WHAT'S REGRESSING (2-3 specific things)
   What's measurably worse? Or what was supposed to be improving
   and isn't? Be honest. The retro is useless if it only sees
   green flags.

4. THE AVOIDANCE PATTERN
   What decision have I now mentioned across multiple weekly
   briefings without acting on? Name it specifically. If there
   are several, pick the most important one and explain why I'm
   avoiding it. If you're not sure, ask me one clarifying
   question.

5. THE BET FOR NEXT MONTH
   What's the one big bet I should make next month, given the
   themes + improvements + regressions? Be specific about the
   bet, the expected outcome, and the cost of being wrong.

6. THE ONE THING TO STOP
   What should I stop doing, based on what's regressing or what
   the data shows isn't producing value? Be specific. Most
   operators are great at adding new things and terrible at
   stopping old ones. Name the thing.

Honest, useful, not sycophantic. Tell me the truth even if it
stings. Australian English, no exclamation marks. If something
doesn't have a clear answer from the data, say "not visible from
this data; here's what additional information would help".

Weekly briefings for this month:

=== Week 1 ===
[paste week 1 briefing + Claude output]

=== Week 2 ===
[paste week 2 briefing + Claude output]

...

Output: structured retrospective. Read it twice. Push back where Claude is wrong (always one or two things). Save the final output for next month’s retro.

Why this is the single most underrated workflow

Most operators run some version of weekly review. Few run a real monthly retro. The monthly retro is where insight compounds:

• Patterns invisible at the weekly tier become visible at the monthly tier. “I keep saying I’ll deal with the supplier issue” is a pattern across 4 weeks. You don’t see it in any single week.
• Stop-doing decisions only surface from longitudinal data. The thing that felt important in week 1 and produced nothing by week 4 needs to stop. You can’t decide that from week 4 alone; you need weeks 1-4 in view together.
• Quarterly + annual planning gets easier. End of each quarter you have 3 monthly retros + 12 weekly briefings. The quarterly retro is half-written before you start.

We’ve watched solo operators run this for 6+ months and become noticeably sharper in their decision-making. Not because they’re working harder, because the longitudinal pattern recognition closes the gap between “the decision I should make” and “the decision I do make”.

The bet for next month: how to do this well

The “bet for next month” section is the highest-action part of the retro. Most operators get it wrong by either over-betting (committing to too many bets) or under-betting (committing to none).

The shape that works:

• One bet per month, not three. Three bets = no priority.
• Specific outcome, not activity. “Launch the new pricing page” is activity. “Test the new pricing page against the current one and decide which converts better by end of month” is an outcome.
• Cost-of-being-wrong stated. If you’re wrong, what’s the cost? If the cost is small, the bet is worth taking. If the cost is large, you need more information before committing.
• Reviewable inside the month. The bet must be decidable from inside the month’s data, not from outcomes that take a quarter.

Good monthly bet examples we’ve seen:

• “Run the inbox-cleanup AI workflow daily for 30 days. By end of month, decide if it’s saving 5+ hours/week. If yes, commit. If no, drop.”
• “Reach out to the 12 dormant customers from 6+ months ago with a personal email. By end of month, measure response rate. Re-engage those who respond.”
• “Stop posting to LinkedIn for the month. By end of month, check if there’s any measurable drop in inbound enquiries. If not, the monthly time spent on LinkedIn can move to something else.”

Bad monthly bet examples:

• “Grow the business 20% next month” (no path)
• “Try to do less email” (no decision criterion)
• “Hire someone” (probably needs longer than a month)

The stop list: usually the bigger win

Most monthly retros surface 1-3 things that could stop. Most operators don’t stop them. The pattern:

• The thing you’ve been doing since the business started that isn’t producing value any more (a service line, a marketing channel, a recurring meeting)
• The thing you started because someone advised it 6 months ago that hasn’t paid off
• The thing that consumes daily time but you can’t articulate the outcome

When the retro surfaces a clear stop-doing candidate, the bet for the month is often “stop X, watch for 30 days, see if anything bad happens”. Usually nothing bad happens.

We’ve watched DotVA clients drop entire service lines, marketing channels, and recurring meetings via the monthly stop-list discipline. The reclaimed time + attention always finds a higher-use use.

Pre-requisite check

Before you run this, you should have:

• Been doing the Sunday-night weekly briefing for at least 4 consecutive weeks
• A Claude Project with your business voice + context (see Projects setup guide)
• Calendar slot booked for the last Sunday of the month, 7-8pm AEST, recurring

If any of those are missing, do them first. The monthly retro requires the weekly input data to be useful.

What this doesn’t solve

Be honest about limits.

• Strategy. Direction-setting is a quarterly / annual exercise, not monthly.
• Personal therapy. AI is not a therapist. If the patterns surfacing are emotional / personal / stress-laden, those need a human.
• Team decisions. This is your personal retro. Team retros are a different workflow (collaborative, in-person ideally).
• Crisis response. When something is on fire, you don’t need a retro; you need to fix the fire.

For the operating cadence layer between weekly action and quarterly direction, this is the workflow.

What’s next

• Your Sunday-night business briefing prompt, the weekly companion that feeds this monthly retro.
• Claude Projects setup in 30 minutes, for the Project context this workflow assumes.
• The first 10 prompts every Australian SMB owner should run, broader starter prompts.
• Book a free audit if you want help designing a quarterly operating cadence on top of weekly + monthly.