There is no single correct priority fee for every situation. The optimal fee depends on the current state of the network, the urgency of your transaction, and the accounts it touches. A good strategy adapts dynamically rather than using a fixed value.

‍

The best priority fee strategy monitors local fee markets and adjusts in real time — not one that sets a number and forgets it.

‍

Monitor Local Fee Markets

Solana uses local fee markets, meaning fees can spike for transactions that touch congested accounts (like popular DEX programs) while remaining low for transactions that touch uncongested accounts. Always query fee data using the specific account addresses your transaction will interact with, rather than using a global fee estimate.

‍

Tiered Fee Strategy

A practical tiered approach: start with a baseline fee at the 50th percentile for your target accounts. If the transaction fails to confirm within a target time window, retry with the 75th percentile. For critical operations, jump directly to the 95th percentile or higher. Many wallets implement this automatically through retry logic.

‍

Spam Avoidance

Sending the same transaction with the same fee repeatedly — a practice common in MEV bot competition — is generally counterproductive for regular users and increases network load. Instead, use exponential backoff on retries and consider whether the transaction is still valid before resubmitting. A transaction that has become stale (its recent blockhash has expired) should be rebuilt, not resent.

‍

Compute Unit Optimization

Accurate compute unit limit setting is arguably more impactful than fee level selection. Use simulateTransaction to measure actual CU consumption, then set your limit to 110–120% of that value as a safety margin. This reduces the fee you pay and improves your transaction's efficiency score in validators' scheduling logic.