The Question Every CTO Gets
"Should we use blockchain for this?"
After building multiple production blockchain systems, here's my framework for answering that question.
The Decision Matrix
Use blockchain when all three are true:
- Multiple parties need to verify the same data
- Trust boundaries exist (parties don't fully trust each other)
- Audit trail is critical (immutability matters)
If any of these is false, use a database.
Real Examples
✅ Good Use Case: Multi-Party Settlement
Problem: Three companies need to track shared transactions without a central authority.
Company A <-> Shared Ledger <-> Company B
|
Company C
- Multiple parties: ✅
- Trust boundaries: ✅
- Audit trail critical: ✅
Solution: Private blockchain (Hyperledger, Polygon CDK, or consortium chain)
❌ Bad Use Case: User Authentication
Problem: Need to store user login credentials.
- Multiple parties: ❌ (just your system)
- Trust boundaries: ❌ (you control everything)
- Audit trail: Not primary concern
Solution: PostgreSQL + proper security. Don't overcomplicate.
Cost Analysis
Blockchain is expensive. Here's what you're paying for:
| Operation | Traditional DB | Blockchain | Multiplier | |-----------|---------------|------------|------------| | Write | $0.001 | $0.50 | 500x | | Read | $0.0001 | $0.01 | 100x | | Storage/GB/mo | $0.10 | $50+ | 500x+ |
These costs are not academic. At scale, they're brutal.
The 3-Question Filter
Before implementing blockchain, ask:
1. Can I use a signed database log instead?
// Often this is enough
interface SignedEntry {
data: any;
timestamp: number;
signature: string; // ECDSA signature
previousHash: string;
}
If a signed, append-only log solves your problem, do that instead.
2. Who actually needs decentralization?
Be honest: Do your users care? Or is this "blockchain for blockchain's sake"?
3. What happens when gas fees spike?
In 2021, Ethereum gas hit $200+ per transaction. If your business model breaks at high gas prices, rethink the approach.
When to Actually Use It
Treasury Management — Multiple signers, high-value transactions, full audit trail.
Cross-Organization Data Sharing — No trusted central party exists.
Supply Chain with Adversarial Parties — Need verifiable provenance across untrusted actors.
Tokenized Assets — When you actually need composability with DeFi or secondary markets.
Implementation Reality Check
Building production blockchain systems requires:
- 3x longer development time vs traditional systems
- Specialized security review (smart contract audits aren't optional)
- Operational complexity (key management, gas price monitoring, node infrastructure)
- Different failure modes (immutability can be a bug when you need to fix things)
Final Framework
START -> Multiple parties? -> No -> Use PostgreSQL
| Yes
Trust boundaries? -> No -> Use PostgreSQL
| Yes
Audit critical? -> No -> Use PostgreSQL
| Yes
Can justify 10-500x cost? -> No -> Use signed logs
| Yes
-> Blockchain might be right
Conclusion
Blockchain is a powerful tool when you need it. Most systems don't.
The best engineering decision is often the simplest one that meets requirements.