Why your billing warehouse is always a step behind, and what to do about the last hour

Warehouses are great for history and terrible for "what's happening right now". Here is how subscription teams cover the gap without building anything.

By Thomas Fitzgerald, Marketing · 2026-10-18

If your business has a data warehouse, Recurly's data probably lands in it on a schedule: every hour, every few hours, or overnight. That is the right design for reporting. Monthly revenue, cohort retention and lifetime value don't need to be up to the minute.

But the questions an operations lead asks in the middle of a working day usually do.

The questions that fall into the gap

• "Did that promotion we launched at 10 actually bring in signups?"
• "Why are support tickets about double charges spiking right now?"
• "Is the decline rate on renewals normal today, or is the gateway having a bad morning?"
• "How many fake signups came in during the last hour?"

Every one of these is about the period since the last refresh. Ask the warehouse and you get yesterday, confidently presented. Ask Recurly's admin and you get one account at a time.

Covering the gap without rebuilding anything

You don't need a faster warehouse. You need the last few hours answered live, and the warehouse left to do what it is good at.

Retrixus reads Recurly when you ask, so it is always current. If you already have a warehouse, it knows where your warehouse stops:

• Questions about today are answered live from Recurly.
• Questions about last year are pointed at your warehouse, and the AI uses it for the long view.
• Questions spanning both get the history from the warehouse and the gap since its last refresh from Recurly.

Already on Recurly's Snowflake share? Keep it for history and reporting. Using BigQuery? Same idea.

No warehouse yet?

Most smaller subscription businesses don't have one, and shouldn't have to build one to get answers. Retrixus answers questions over the last seven days live, and background analysis covers up to 30 days. For longer history without building a pipeline, Warehouse Sync loads your Recurly data into your own BigQuery on a schedule.

Honest when it can't

Reading live has one limit: Recurly doesn't count records for you, so counting means reading them. For a question that needs more records than it can read, Retrixus says how far it got and how to narrow the question. It never presents a partial count as the whole answer.

Start free on one Recurly site, or read about history and warehouses.