Order Processing Without Staff: Our Setup at nano
At nano. we get 50-150 orders a day. That used to mean hours of order processing every day. Today it takes 15 minutes, and only for the edge cases. At mate, where we run the same setup, the workload went from 4 hours a day down to 15 minutes.
Where we started
Before (manual):
1. Check the order in Shopify
2. Copy the data over to the fulfilment partner
3. Wait for the tracking number
4. Enter the tracking number in Shopify
5. Email the customer
6. Chase problems by hand
Time per order: 3-5 minutes. At 100 orders: 5-8 hours.
The automated flow
Today:
Order comes in, everything runs automatically, customer gets tracking
Let me walk you through the setup:
Trigger: Shopify webhook
Every new order fires a webhook to our n8n server. It carries all the order data. If webhooks are new to you, we explain them here without the jargon.
Step 1: Validation
The workflow checks:
- Are all required fields there?
- Does the address look plausible? (postcode format, etc.)
- Any special requests in the comment field?
- Was express shipping selected?
If something is off: Slack notification to us, and the order gets flagged.
Step 2: Fulfilment API
Order data goes to our fulfilment partner automatically. We use the REST API directly, no manual portal.
Transformations the workflow handles:
- SKU mapping (Shopify to fulfilment partner)
- Address formatting to the DHL spec
- Shipping method based on order value and customer choice
Step 3: Status monitoring
n8n polls the fulfilment status on a schedule. As soon as it flips to "shipped", the tracking number gets pulled.
Step 4: Shopify update
The tracking number goes back into Shopify via the API. Order status is set to "fulfilled".
Step 5: Customer notification
An automatic email to the customer with:
- Tracking number and link
- Expected delivery date
- Support contact in case of problems
Nice side effect: because customers have their tracking immediately, we get far fewer "where is my parcel?" messages. Our WISMO automation catches the rest.
The tech stack
- Shopify as the store system
- n8n (self-hosted on a Hetzner VPS, our reasons for self-hosting are here) for orchestration
- Fulfilment API (connected directly)
- Gmail API for transactional email
- Slack for alerts
Total cost: around 15 euros a month for the server.
Error handling
The most important question in automation: what happens when something breaks?
Our rules:
1. Every error triggers a Slack alert with the details
2. Retry logic: 3 attempts with exponential backoff
3. After 3 failures: move it into a "manual queue"
4. Daily: an automatic report of all anomalies
In 6 months we had 4 cases that needed manual intervention. All of them came from problems on the fulfilment side, none from our automation.
The results
| Metric | Before | After |
|---|---|---|
| Time per order | 3-5 min | 0 min (auto) |
| Time to shipment | 24-48h | 4-8h |
| Daily workload | several hours | 15 min |
There is one more effect no table shows: orders shipped to the wrong place have become rare. Those 15 minutes now consist of checking Slack alerts and handling edge cases.
Limits of this setup
So you know what you are getting into, three constraints:
- Your fulfilment partner needs an API. Without one you are back to clicking through a portal or uploading CSVs, and both break the chain. It was a selection criterion for us when we switched partners.
- Special requests stay manual. The workflow spots "please gift wrap this" in the comment field and flags the order, but a human at the warehouse still has to do it.
- This is not an afternoon of work. Our first working setup took about a week, and today's version has matured over months. If you want a realistic view of the cost: what automation really costs.
What I learned
1. Invest in monitoring. Automation without monitoring is driving blindfolded.
2. Build for failure. Most of the development time went into "what if X goes wrong?"
3. Start simple. Our first setup was far more basic. We iterated.
4. APIs beat portals. Any fulfilment partner without an API is out for us.
Want the same thing?
Two options:
Build it yourself: Write to me and I will send you more detail. benedikt@flowhouse.ai
Have it built: This is exactly what we do at Flowhouse. Talk to us.