Coming at this from the merchant side rather than the routing side, which changes what "profitable" means.
Measure payment success rate, not capacity. The only number I track is: of the invoices I hand a buyer, what share settle first try, and which routes fail. Capacity charts are vanity. Log every failure and its error, and the same two or three bad hops show up over and over.
Buy inbound liquidity only after proven demand. Don't pre-buy a channel because a calculator said so. Wait until real invoices are getting paid and you're actually hitting the inbound ceiling, then size to the demand you measured, not the demand you hope for.
Don't race fees to zero. Zero-fee channels look great on a dashboard and lose money. Set a fee budget you can defend, adjust slowly, and let unprofitable flow go elsewhere. Cheap routing is not profitable routing.
Prune monthly. Once a month, list peers with no forwards and no settled payments in 30 days and close them. Dead channels are locked capital plus on-chain risk you're carrying for nothing.
Merchant checklist if you sell anything:
Reliable receive beats everything. If the invoice fails, the sale is gone and the digital delivery never happens.
Backups tested before you need them, not after.
Don't overbuild channels for a quiet shop. A few well-chosen channels beat a sprawling node you can't keep balanced.
Context: I run a small Lightning digital shop (Shopstr) - https://shopstr.market/marketplace/clipper - so all of this comes from selling files for sats and watching payments fail, not from farming routing revenue. Different incentives, but the failure data is real.
Coming at this from the merchant side rather than the routing side, which changes what "profitable" means.
Merchant checklist if you sell anything:
Context: I run a small Lightning digital shop (Shopstr) - https://shopstr.market/marketplace/clipper - so all of this comes from selling files for sats and watching payments fail, not from farming routing revenue. Different incentives, but the failure data is real.