Earning Money With AI

Here’s what that usually looks like in the real world. You try an AI workflow that writes content, summarizes messages, or generates images. It feels productive for a day or two. Then the output doesn’t match your brand voice, your staff doesn’t trust it, your customers don’t notice, or you can’t deliver it consistently. You end up with an AI project that never plugs into your sales process, your delivery process, or your pricing.

You need a way to filter ideas fast. Not “Is AI useful?” but “Will this use case make money for my business, with my constraints?” That’s what this chapter solves.

Your reader avatar (the kind of business owner this chapter is built for) You run a small business with real delivery pressure. You might be the owner, the manager, and the person answering customer questions. You don’t have time to experiment for months. You care about cash flow, you need a clear offer, and you need something you can run on a normal week.

To anchor this chapter, I’ll use one primary case study: Talia, 34, restaurant owner. She wants to use AI, but she refuses to bet her operation on something that breaks service or creates extra work.

The transformation promise After you work through this chapter, you will be able to: - take 5-10 AI use case ideas and score them quickly, - pick the one most likely to earn money first, - outline exactly how you’ll deliver it (not just what the AI produces), - and spot the “looks good on paper” failures before you invest.

---

Why The Opportunity Fit Scorecard Works (And Why It Prevents Waste)

The Opportunity Fit Scorecard forces you to connect AI to three real-world systems: your offer, your customer need, and your capacity. Instead of asking whether the AI can do something, you ask whether the business can profit from it.

This works because profitable AI use cases share a pattern: they reduce time or increase conversion in a place where you already have money moving. The AI should either (1) help you sell more of what you already offer, (2) help you deliver faster or more consistently, or (3) help you capture leads you currently miss. If it doesn’t touch one of those, you’ll struggle to measure impact.

When you score an idea, you stop debating opinions and start making tradeoffs. Talia didn’t need a “marketing AI” or a “content AI.” She needed a use case that fit her restaurant rhythm, her staffing reality, and her customer pain points - like last-minute reservations, menu confusion, and slow responses during busy hours.

Here’s a concrete way to think about it. A chatbot that answers “What’s on the menu?” might sound helpful. But if your team already answers that in under a minute and people still don’t convert, the chatbot won’t move revenue. A better use case might help customers decide faster - like suggesting the right dish for dietary needs using your actual menu and current specials. That touches conversion, not just “engagement.”

The scorecard: how to measure fit in plain terms You’ll score each idea from 0 to 5 on each category below. You can do this in a spreadsheet or a notebook. The goal isn’t perfect accuracy. The goal is fast, honest selection.

1. Offer Match (0-5): Does this use case directly support something you already sell (or a clear extension of it)? - Example: AI that drafts reservation follow-ups fits a restaurant’s offer; AI that generates random blog posts usually doesn’t.

2. Audience Need Fit (0-5): Does it solve a specific customer problem they feel this week, not “someday”? - Example: busy diners need fast answers about allergens and portions, not generic descriptions.

3. Delivery Capacity (0-5): Can you deliver it reliably with your current team and time? - Example: if it requires staff to review 200 messages manually, you’ll drown.

4. Measurable Impact (0-5): Can you track a result tied to money or operations? - Example: track reservation conversion, average response time, or fewer missed calls.

5. Risk and Compliance (0-5): Will it create brand damage, wrong info, or safety issues? - Example: allergen claims require careful setup; “AI guesses ingredients” creates risk.

6. Time-to-Value (0-5): Can you test and see something useful within 7-21 days? - Example: menu Q&A improvements should show up quickly if you build it with your real menu data.

Total the points. The highest score isn’t always the best idea, but it usually becomes the best first test.

---

Applying The Opportunity Fit Scorecard to Pick Talia’s First AI Use Case

Let’s walk through a realistic scenario with Talia and pick the use case she should test first.

Talia runs a restaurant. She already posts daily specials, but she notices two recurring issues: - People ask the same questions repeatedly: “Is this gluten-free?” “Do you have spicy options?” “What’s the portion size?” - During dinner rush, her staff gets overloaded answering messages and phone questions that slow down service.

She writes down 5 AI use case ideas she’s heard about online. Then she scores them using the scorecard.

Step-by-step: score, choose, and define delivery

1. List 5 candidate use cases (keep them specific). Example ideas Talia considers: - AI menu Q&A for common questions (allergens, spice level, portion sizes) - AI-assisted reservation follow-up messages (confirmation + reminders) - AI draft captions for daily specials - AI lead capture form that qualifies catering inquiries - AI summary of daily customer feedback to spot issues

Expected outcome: you reduce “random AI ideas” into testable tasks.

2. Score each idea using the 6 categories. Talia scores each idea from 0 to 5:

• Menu Q&A: Offer Match (5), Audience Need Fit (5), Delivery Capacity (3), Measurable Impact (4), Risk (2), Time-to-Value (4) Total: 23 - Reservation follow-up: Offer Match (5), Audience Need Fit (4), Delivery Capacity (4), Measurable Impact (4), Risk (4), Time-to-Value (5) Total: 26 - Caption drafts: Offer Match (3), Audience Need Fit (2), Delivery Capacity (5), Measurable Impact (2), Risk (4), Time-to-Value (5) Total: 21 - Catering lead qualification: Offer Match (4), Audience Need Fit (4), Delivery Capacity (3), Measurable Impact (4), Risk (3), Time-to-Value (3) Total: 21 - Feedback summaries: Offer Match (3), Audience Need Fit (3), Delivery Capacity (4), Measurable Impact (2), Risk (4), Time-to-Value (4) Total: 20

Expected outcome: you see which idea wins on fit, not on hype.

3. Pick the top-scoring idea that also fits your next 7-21 days. Talia’s top total is reservation follow-up. That matters because she can test it without touching complex claims like allergens. She chooses: AI-assisted reservation follow-up messages.

Expected outcome: you protect time and reduce risk.

4. Write a delivery definition that your staff can follow without guessing. Talia defines exactly what the system will do: - Trigger: a reservation gets confirmed - Message timing: send one reminder 24 hours before, one 2 hours before - Content rules: include reservation time, address, parking note, and a “reply for changes” line - Human control: staff reviews and approves the message templates weekly, not per message

Expected outcome: you build a workflow that doesn’t create extra work during rush.

5. Set one measurable target and one operational check. She tracks: - Reservation no-show rate (or a simple proxy: “reservations that don’t arrive”) - Average time to respond to reservation questions in the hour after reminders go out

Expected outcome: you connect AI activity to outcomes you can see.

Quick checklist (Talia’s version) - Score each idea 0-5 on Offer Match, Audience Need Fit, Delivery Capacity, Measurable Impact, Risk, and Time-to-Value - Pick the highest score that you can test within 7-21 days - Define triggers, timing, message rules, and who approves templates - Track one money-linked result and one operations metric

---

What to Watch For: Common Mistakes That Kill Profits

Even with a scorecard, you can still mess up. These are the mistakes I see most often when owners try to “do AI” without turning it into a money machine.

Mistake 1: You score “cool” higher than “deliverable” Do this: Use Delivery Capacity as a hard gate. If you can’t run the workflow during your busiest week without extra stress, your score doesn’t matter. Not this: “We’ll figure it out later” when your team needs to manually correct outputs every time.

Fix: Force yourself to write the workflow in staff-level steps. If the steps sound vague, you didn’t define delivery.

Mistake 2: You pick an idea with measurable impact… but you measure the wrong thing Do this: Track outcomes tied to revenue or operational savings. For a restaurant, you don’t measure “likes” from AI captions first. You measure reservations turning into seats, or you measure fewer missed calls. Not this: Measuring vanity metrics because they feel easier.

Fix: Pick one metric you already track weekly, then add one AI-related check on top.

Mistake 3: You ignore risk until the first bad output Do this: Treat anything that can cause safety or legal issues - like allergens, pricing, or policy statements - as high-risk. Put human approval where it matters. Not this: Letting the AI freestyle claims because it “usually sounds right.”

Fix: Use strict templates and grounded data. For example, Talia limits follow-up messages to confirmed reservation details and standard house info, not “AI guesses” about menu ingredients.

---

Your Chapter Roadmap: Pick One AI Use Case This Week

You don’t need to master AI to start earning with it. You need to choose one use case that fits your business and can ship fast.

Here’s the progression from today: - Today: Write down 5-10 AI use case ideas you’ve seen or heard about. Make them specific to your offers. - This session: Score them with The Opportunity Fit Scorecard (0-5 on the 6 categories). - Next: Pick the best one and write your delivery definition: trigger, timing, message rules, approval steps. - This week: Set your measurement plan using one money-linked metric and one operational check. - Within 7-21 days: Test with real customers, then adjust templates or rules based on what actually happens.

One practical rule to keep you honest: if you can’t describe the workflow in 10 sentences, you don’t have a usable use case - you have an idea.

As you move into the next step of the book, you’ll learn how to build and refine these AI workflows without turning your business into a science project. The win starts with selecting the right target, and the scorecard gives you the fastest path to that choice.

End of chapter one. 4 more chapters in the full book.