Automated Order Processing for a Local E-Commerce Shop

NaturSchön is an online shop based in Basel that sells Swiss natural cosmetics products directly to end customers. The range includes 180 products from 12 local manufacturers. The shop processes an average of 400 orders per month through its own WooCommerce shop and additionally through a Galaxus marketplace account. A two-person office team was primarily occupied with order processing and had barely any time for marketing, customer care, or range expansion.

Managing director Maria Keller founded NaturSchön five years ago as a side project and grew faster than expected. Success brought a capacity problem: every incoming order had to be manually reviewed in the WooCommerce dashboard, recorded as an invoice in Bexio, and created as a shipping order with Swiss Post. For Galaxus orders, an additional manual step was required. The average processing time per order was 8 minutes, at 400 orders per month, an enormous time burden. Data entry errors led to 5-10 wrong deliveries monthly, resulting in costly correction rounds.

The main problem was triple manual data entry: every order had to be confirmed in WooCommerce, recorded as an invoice in Bexio, and created as a shipping order with Swiss Post. Customer data (name, address, order items, amounts) was manually typed three times. Additionally, inventory had to be manually updated, regularly leading to overselling: products were shown as available even when physical stock was already zero. Returns processing was another bottleneck: each return required manual review, credit note in Bexio, and inventory update. At 30-40 returns per month, this consumed additional work hours.

We built an AI-powered workflow automation that automates the entire order process from receipt to shipping label. The solution consists of three core modules: The order processing module monitors new orders in WooCommerce and Galaxus, validates customer data (address validation, postal code check), automatically creates the invoice in Bexio, and generates the shipping label via Swiss Post API. Only orders with irregularities (missing data, unusually high order values, new wholesale customers) are forwarded for manual review. The inventory module synchronizes stock levels in real-time between WooCommerce, Galaxus, and the physical warehouse. When minimum stock is reached, a reorder request is automatically sent to the respective manufacturer. The returns module processes returns automatically: categorize reason, create credit note, update inventory, and send customer email.

Implementation took 6 weeks. Particularly challenging was the Galaxus API integration, which uses different data formats than WooCommerce. We developed an AI-powered data formatter that automatically normalizes the different formats. Another highlight was address validation: the AI recognizes typos in Swiss addresses (e.g., 'Bahnhofstr.' vs. 'Bahnhofstrasse') and corrects them automatically, significantly improving first-attempt delivery rates. For returns processing, the AI trained a classification model based on historical return data that automatically recognizes the reason for the return with 90% accuracy.

Results after 4 months of production operation exceeded expectations: 95% of all orders are now processed fully automatically, from order to printed shipping label takes an average of 45 seconds instead of 8 minutes. Wrong deliveries dropped from 5-10 per month to 0-1, as AI-based address validation and order verification eliminates human entry errors. Monthly savings total CHF 2,500: CHF 1,800 from saved staff hours, CHF 400 from reduced wrong deliveries, and CHF 300 from avoided overselling. The office team now has 80% of its working time free for marketing and customer care. Revenue increased by 18% in the following quarter, partly due to the time gained for marketing activities.

The most important insight: the 5% of orders that still require manual review are actually the cases where human judgment adds real value: unusual special requests, first orders from wholesale customers, potential fraud cases. Automating the routine improved the quality of manual work because the team can now fully focus on demanding cases. Second: real-time inventory synchronization had a surprisingly large effect on customer satisfaction: no more 'unfortunately sold out' emails after ordering.