Start Lesson
Your AI Opportunity Audit surfaced three strong candidates. Your CTO says "let's build it." Your marketing lead says "just use ChatGPT." Your CFO forwards a pitch deck from a vendor promising 10x productivity. All three are right -- for different situations. You need a framework that tells you which path fits which problem, in under 20 minutes per decision.
This lesson gives you a weighted scoring matrix. By the end, you will have a filled-in decision for each of your top AI opportunities: prompt it, buy a tool, or build custom.
A completed Build vs Buy vs Prompt Decision Matrix for your top 3 AI opportunities from Lesson 1, with weighted scores that make the choice defensible to your leadership team.
| | Prompt | Buy | Build | |---|---|---|---| | What it means | Your team uses ChatGPT, Claude, or Gemini directly. Copy-paste into workflows. | You purchase a vertical AI product built for your use case (Jasper, Intercom AI, Harvey). | You develop custom AI using APIs, frameworks, and your own data. | | Cost | $0-20/user/month | $50-500/month | $5,000-100,000+ upfront, plus ongoing | | Time to value | Hours | Days to weeks | Weeks to months | | Customization | Limited to prompt quality | Moderate -- configure within their product | Full control over everything | | Integration | None -- manual copy-paste | Limited to vendor's connectors | Whatever you build | | Data ownership | Low -- data goes to provider | Varies by vendor | Full | | Maintenance | None | Vendor handles | You handle |
For each AI opportunity, score these six criteria on a 1-5 scale. Each criterion has a weight reflecting its importance. Multiply score by weight, then add up the totals for each path.
| Criterion | Weight | Prompt (1-5) | Buy (1-5) | Build (1-5) | |-----------|--------|-------------|-----------|-------------| | Budget fit -- Does the cost match your available budget? | 3x | | | | | Speed to value -- How fast do you need results? | 2x | | | | | Customization need -- Does the task require your specific data, tone, or logic? | 3x | | | | | Integration need -- Must it connect to your CRM, helpdesk, or other systems? | 2x | | | | | Data sensitivity -- How critical is it that your data stays in your control? | 2x | | | | | Scale -- Will usage grow 5-10x in the next year? | 1x | | | | | Weighted Total | | ___ | ___ | ___ |
How to interpret: The path with the highest weighted total wins. If two paths are within 5 points, default to the simpler option (Prompt beats Buy, Buy beats Build).
The firm's top AI opportunity from their audit: candidate sourcing and outreach (scored 14 on the Opportunity Scorecard).
| Criterion | Weight | Prompt | Buy | Build | |-----------|--------|--------|-----|-------| | Budget fit (tight -- small firm) | 3x | 5 (15) | 3 (9) | 1 (3) | | Speed to value (need results this month) | 2x | 5 (10) | 3 (6) | 1 (2) | | Customization need (standard outreach) | 3x | 3 (9) | 4 (12) | 5 (15) | | Integration need (must connect to ATS) | 2x | 1 (2) | 4 (8) | 5 (10) | | Data sensitivity (candidate PII) | 2x | 2 (4) | 3 (6) | 5 (10) | | Scale (doubling recruiters next year) | 1x | 2 (2) | 4 (4) | 4 (4) | | Weighted Total | | 42 | 45 | 44 |
Buy and Build are close, but Buy wins on speed and budget. The firm adopts a recruiting AI tool at $300/month. If they outgrow it after scaling, they revisit Build.
Key insight: Build scored highest on customization and data sensitivity, but the budget and speed weights dragged it down. That is the whole point of weighting -- it forces you to prioritize what actually matters for your business right now.
The right approach is almost never to jump straight to Build. Think of it as a ladder:
Start with Prompt. Give your team access to ChatGPT or Claude. Let them experiment for 2-4 weeks. See which tasks naturally attract AI usage.
Graduate to Buy when you notice a pattern: multiple people doing the same AI-assisted task repeatedly, and a commercial product exists that does it better than raw prompting. The signal is usually that the task needs tool integration or shared templates.
Escalate to Build only when one of these is true:
Most businesses will never need to Build. That is a perfectly good outcome.
Before committing to Buy, answer these three questions:
1. Can I export my data? If you spend six months building prompt templates and custom configurations inside a vendor's platform, can you take them with you? If not, factor in switching costs.
2. What happens to my data? Does the vendor train on your inputs? Store your customer data? Where are the servers? Read the data policy, not the marketing page.
3. What is the pricing at 10x scale? A tool costing $100/month for 1,000 queries might cost $5,000/month for 10,000. AI tool pricing is volatile. Know the curve before you sign, and get pricing commitments in writing if possible.
The best Buy decisions come with an exit plan. Before you sign, know how you would migrate if you had to.
Take your top 3 AI opportunities from the Lesson 1 scorecard. For each one, fill in the weighted decision matrix:
Opportunity 1: _________________________ (Score: ____ from Lesson 1)
| Criterion | Weight | Prompt | Buy | Build | |-----------|--------|--------|-----|-------| | Budget fit | 3x | | | | | Speed to value | 2x | | | | | Customization need | 3x | | | | | Integration need | 2x | | | | | Data sensitivity | 2x | | | | | Scale | 1x | | | | | Weighted Total | | ___ | ___ | ___ |
Repeat for Opportunities 2 and 3. Your output: a clear Prompt/Buy/Build recommendation for each opportunity, backed by numbers you can show your leadership team.
Reality check: If all three opportunities land on "Prompt," that is not a failure. It means you can start generating value this week with near-zero cost. Most businesses should be prompting for months before they spend a dollar on buying or building.
You now know what to automate and how to implement it. But your CFO wants a number. "What is this actually worth in dollars?" Lesson 3 gives you the ROI calculator -- a formula you can fill in for each opportunity to project monthly and annual returns. Bring your top-scoring opportunity and its Prompt/Buy/Build recommendation with you.
