The reliability guarantee your billing platform won't provide

I talked to an engineering manager today about monitoring issues in production. He mentioned an open-source platform his company uses to increase reliability: Temporal. I went to check it out and realised how hugely popular it's become. Long story short, it lets you define workflows that automatically survive crashes, restarts, and network failures: without you writing retry or recovery logic yourself.

It made me realise how easy it is to take reliability for granted when dealing with pricing / billing complexities.

Every major billing platform - Stripe, Chargebee, Metronome, Polar etc. is reliable on its own. None of them, however, owns the layer between your CRM, your usage events, invoices, and your warehouse in a way that survives failed event delivery, events arriving out of order, and missed reconciliations.

That layer is yours to build. If you or your team have never explicitly built this layer, it's most likely not there. This is most likely quietly costing you money in ways your dashboard won't show.

PS: You don't have to use Temporal. Inngest, Restate, AWS Step Functions, Kafka, or even a well-behaved queue with idempotency keys can get you there. The pattern matters more than the tool.

PPS: This is the work I do. Happy to nerd out in the comments or DMs.

4 comments