Farrells Bookshop: 13 daily automation steps, 0 manual checks.

The problem

Thousands of titles change price and availability every day. Two imports pulled BookNet stock and pricing files uploaded from a PC in the shop, and a third pulled covers and richer metadata from NielsenIQ overnight. Each step assumed the one before it had worked. If the afternoon upload failed, the server still renamed, imported and archived yesterday’s file, and nobody knew until something looked wrong on the website.

Enhance, don’t rebuild

We had built the original integration, so we knew its foundations were sound. A full rebuild on new tooling would have added risk and months of rework for the same result. Farrells agreed to a targeted fix instead: make each step depend on the last one, and make failures visible.

Gatekeeper checks

Before the afternoon and nightly BookNet imports go ahead, a gatekeeper confirms a fresh file has arrived inside its time window. If it hasn’t, every step after it switches off and the chain stops cleanly. The NielsenIQ sync was left out on purpose: it runs independently, so stopping it would add complexity for no benefit.

Controlled restart and alerts

A stopped import stays stopped until someone switches it back on. That’s deliberate: a missed upload leaves a gap in the data, and an automatic restart would hide it. Email alerts now go to the Farrells team and to us the moment a critical step fails, so problems are picked up before customers see them.

What changed for Farrells

• Stale data can no longer be imported as fresh
• Failures are reported by email straight away
• Data gaps are handled on purpose, not papered over
• No new platform, no rebuild, no downtime
• Staff no longer watch the imports by hand

Automation that fails silently can be worse than none. When a manual process breaks, someone notices. We build failure checks into every WooCommerce and data integration from day one, because that’s what makes automation safe to rely on.