The AI built my plan around one $4,800 job. 650 hours of my calendar said no.

"Double down on high-ticket deep cleans."

That was the AI's top strategy pick for a small cleaning company. It came from one glowing email about a $4,800 move-out clean. The rest of the owner's 650-hour quarter told a different story.

AI Prompts, Reviewed takes one prompt or workflow from How I AI, Claire Vo's show where guests share their screens and walk through how they really use AI. We run it on a small-business job, score it, and tell you what we'd change.

Credit: Based on "Generate a Data-Driven Business Strategy with Claude Opus 4.7," a workflow from How I AI (hosted by Claire Vo, written up by the ChatPRD team), from Claire Vo's own episode "My First Impressions of Claude Opus 4.8." Original workflow: https://www.chatprd.ai/how-i-ai/workflows/generate-a-data-driven-business-strategy-with-claude-opus-4-7 . Full episode: https://www.chatprd.ai/how-i-ai/claude-opus-4-8-review

What the original does

Claire gives Claude three months of calendars, projects and performance data. It compares where her time went with her goals, recommends the three biggest changes, then turns them into a 90-day roadmap. The step we liked most is the check at the end: "Reject any claim without a traceable record or calculation."

The test

Composite (made-up) data for a cleaning company: the owner plus 4 cleaners. Quarter: July to September 2026, 13 weeks. Goal: grow commercial contracts from $13,600 a month to $20,000 a month.

Where the owner's 650 hours went:

• Covering cleaning shifts: 247 hours (38%)
• Quoting one-off deep cleans: 156 hours (24%)
• Invoicing and chasing payments: 117 hours (18%)
• Hiring and scheduling: 78 hours (12%)
• Commercial sales, the actual goal: 52 hours (8%)

That's 247 + 156 + 117 + 78 + 52 = 650.

Revenue for the quarter: $38,400 from 64 deep cleans, $40,800 from commercial contracts ($13,600 x 3), and $22,500 from recurring homes. Total: $101,700.

We ran the short prompts from the page. It sounded smart. Two problems:

1. It chased one story. The $4,800 move-out clean had the best email in the pile, so "high-ticket deep cleans" became priority one.
2. It made up a number. It said deep cleans run "about 45% margin." We gave it no cost data. Zero.

The page's own notes warn about both. But the short prompts don't force a source table, so we wrote one in and reran it.

The rerun did the math:

• Deep cleans: 156 hours of quoting for 142 quotes and 64 wins (45%). That's $38,400 / 156 = about $246 per hour, paid once.
• Commercial: 52 hours of sales landed one new contract at $3,200 a month. That's $38,400 a year / 52 = about $738 of yearly revenue per hour, and it repeats.

Same $38,400, a third of the hours.

The gap to the goal is $20,000 - $13,600 = $6,400 a month, or two more $3,200 contracts. At 52 hours per contract, that's 104 hours. Block 8 hours a week and 13 weeks gets you there: 8 x 13 = 104. The rerun said to take them from covering shifts and quoting, not to add deep cleans.

One $4,800 job is a story. 650 hours is a pattern.

Verdict: 7/10

Strengths: the order is right: facts, then strategy, then roadmap. Every claim must name its source and split facts from guesses. The roadmap needs owners and metrics.

Weakness: it assumes the AI can see your whole business. Most owners have a calendar, a bank account and QuickBooks. Without a forced source table, the AI grabs the loudest story.

Founder tweak: build a one-page table first, with hours by bucket and revenue by line. Tell the AI it can't claim a margin unless you gave it costs.

Copy the MCAFax version

This is our own rewrite, not the original prompt. Copy the original from Claire Vo on How I AI: https://www.chatprd.ai/how-i-ai/workflows/generate-a-data-driven-business-strategy-with-claude-opus-4-7

You are a numbers-first strategy advisor for my small business.
I will paste a table of my hours by task and my revenue by line for the last 3 months, plus my one main goal.
Step 1. Rebuild the table and add up every column. If a total doesn't match mine, stop and tell me.
Step 2. For each line, show revenue per hour of my time and say if it repeats or is one-time.
Step 3. Show the math for the gap between where I am and my goal.
Step 4. Recommend 3 changes. For each: the number it's based on, where the hours come from, and the tradeoff.
Step 5. Turn them into a 90-day plan: who owns it, weekly hours, one metric, and the date I decide to keep or drop it.
Rules: Use only numbers I gave you. Never claim a margin or profit unless I gave you costs.
Label each point "Fact" or "Guess."
Ignore single stories, reviews, or emails unless the table backs them up.
My goal: [example: commercial contracts from $13,600 to $20,000 a month by March]
My table: [paste here]

Paste into Claude, ChatGPT, Gemini, or Copilot.

Before you plan next quarter

1. Pull 13 weeks of your calendar and sort it into 5 buckets.
2. Pull revenue by line from your books or bank statements.
3. Write one goal with a number and a date.
4. Run the MCAFax version. Check every total yourself.
5. Find the hours. Ours were in covering shifts and quoting.
6. Win back admin time: https://mcafax.com/notebook/prompts/invoice-payment-reminders (more prompts at https://mcafax.com/notebook/prompts)

SBA on writing a business plan: https://www.sba.gov/business-guide/plan-your-business/write-your-business-plan

SBA on managing your finances: https://www.sba.gov/business-guide/manage-your-business/manage-your-finances