For eighteen months, my Monday morning ritual was a small, recurring act of self-harm. I produce a metrics summary for our leadership team, and producing it meant opening five dashboard tabs, exporting four CSVs, copy-pasting numbers into a Google Doc, and then staring at the pile until a narrative grudgingly emerged from the spreadsheet fog. The prose I wrote sounded exactly like what it was: a tired person reading numbers off a screen and gluing the words 'which is up' onto the end of each one. It ate about two hours every single week, and I resented every minute, because not one of those two hours was actually thinking. It was assembly. It was data data-entry for a robot I had to be.
The worst part was that the output was forgettable. Two hours of my life, every Monday, producing a document that read like a phone book and that my CEO almost certainly skimmed in forty seconds.
The new Monday
Now I export our main dashboard as one CSV, paste the whole thing into the metrics_data field of the Turn a Metrics Dump into a Skimmable Executive Summary prompt, set the business context (B2B SaaS, around 500 customers, north star is weekly active users), and type in the time period. GPT-4o returns a three-finding summary in about five seconds. I read it once, sanity-check the numbers against the dashboard, add one sentence of context my leadership team will want that the data alone cannot know, and I am done in twenty minutes. The two hours became twenty minutes, and — this is the important bit — the twenty minutes is now the thinking part, not the assembly part. I deleted the assembly entirely.
The discipline it enforces that I never could
Left to my own devices, I forget to say whether a number is actually good. I would write 'WAU was 14,200' and leave the reader to wonder, on their own time, whether that was a triumph, a crisis, or a Tuesday. This prompt has a hard rule baked in that every single metric arrives with its comparison baseline, so the output always reads 'WAU was 14,200, up 6 percent from last week and 2 points ahead of the quarterly goal.' That one enforced habit, which I could never reliably enforce on myself at 8am, made my reports roughly twice as useful overnight. A number without a baseline is just trivia.
- Every metric arrives with a baseline, so my CEO never has to ask the dreaded 'compared to what?' in the all-hands.
- The 'What This Data Can't Tell You' section flags the gap before leadership does — it has saved me from at least three 'but what about...' ambushes in meetings.
- The recommended actions are assignable, not vague: not 'investigate churn' but 'interview the 12 customers who churned this month with fewer than 3 logins in their last 30 days.'
- It once caught a comma-formatted number it suspected was a parsing artifact and flagged it instead of reporting it as fact — exactly the kind of silent error that quietly destroys trust in a recurring report.
The honest caveat
It is not magic and I will not pretend it is. About once a month it misreads a number that has a thousands separator, and one memorable week it confidently described a flat, slightly-declining metric as 'roughly stable' when I would have personally called it 'flat and quietly concerning.' So I read every output before it ships, and that twenty minutes is real review time, not zero. But here is the thing: reviewing a draft is a fundamentally different and easier job than generating one from a blank page, and my brain is fresh for the only part that genuinely needs me — the single sentence of human judgment about what this means for us specifically, which no model has the context to write.
The other thing I did not expect was how it changed the meeting itself. Because the summary is under 400 words and front-loads the three findings that matter, leadership now actually reads the whole thing before the call instead of skimming the first chart and forming a half-opinion. We spend the meeting discussing the recommended actions rather than re-explaining what the numbers were. The report stopped being a wall I built and started being a conversation starter, which is what a metrics summary was always supposed to be and never was when I wrote it by hand at 8am on a Monday.
Three months in, my CEO mentioned offhand that my Monday reports had become the most useful they had ever been. I did not mention the workflow change. I did quietly grab a few more prompts that same afternoon. If your weekly report is currently two hours of assembly and twenty minutes of thought, flip the ratio — grab the Turn a Metrics Dump into a Skimmable Executive Summary prompt on Prompt Dock and paste in your own messy export. And when a finding needs a deeper cut into who is actually driving it, I send the same data through my Segment a Customer List into Actionable Cohorts prompt and bring the answer to the same meeting.