Score each 1 to 5. Higher favours building.
| # | Question | 1 (buy) | 5 (build) | Score |
|---|---|---|---|---|
| 1 | Is this capability core to how we win? | Commodity | Our edge | |
| 2 | Do we have proprietary data a vendor cannot? | No | Years of it | |
| 3 | How fast must it change? | Stable | Weekly | |
| 4 | Do we have engineers who have shipped this before? | No | Yes | |
| 5 | Can a vendor meet our guardrail and audit requirements? | Easily | Not at all | |
| 6 | Total cost over 3 years, vendor vs build and run? | Vendor far cheaper | Build far cheaper | |
| 7 | Lock-in risk if the vendor changes terms or dies? | Low | Severe | |
| 8 | Time to value? | Vendor: weeks | Similar | |
| 9 | Integration depth with our systems? | Shallow | Deep | |
| 10 | Will we learn something strategic by building? | No | Yes |
Reading the score
- 10 to 24: buy, and spend your effort on the guardrails around it.
- 25 to 37: pilot both for 90 days with the same evaluation set, then decide.
- 38 to 50: build, starting in suggest-only mode.