Orders and after-sales without manual data entry

The challenge

The team checked every order across three different dashboards: the store, the warehouse spreadsheet and the support inbox. Every status change was copied by hand, and customer notifications depended on someone remembering.

How it was done before

Every morning, one person exported the new orders, pasted them into the warehouse spreadsheet and combed the support inbox for related issues. Customer notifications were written one by one.

What we built

• A central flow picks up every new order and hands it to the warehouse with the data already normalized.
• Status changes propagate in both directions, without duplicating records or overwriting information.
• Issues are classified and routed to the right person, with an automatic heads-up to the customer.
• When something fails, the flow retries and alerts the team before the buyer notices.

Human oversight

Issues the classifier cannot identify with confidence go to a queue reviewed by a person; refunds and address changes are always approved by someone on the team before they run.

What the client takes away

• Flow in production with error handling
• Monitoring dashboard
• Documentation and training

Outcome (expected)

The specific metrics — hours recovered, errors avoided, response times — are agreed at the start of each project and measured afterwards. This scenario is illustrative: we do not publish figures we cannot back up.

Limits

• It does not decide refunds or commercial exceptions: it prepares them and leaves them ready to approve.
• If the store or the warehouse changes systems, the corresponding connector needs adapting.
• With only a handful of orders a day, the savings do not justify the project.