An AI colleague for your web shop: order status, returns and repeat questions
Most tickets in a web shop are the same question: where is my order. Nova looks up the status, includes the tracking link and prepares a return label where relevant. That is work with a lot of repetition and few exceptions — exactly what automation is best at. What is left for your team are the conversations that genuinely matter.
Published on · Last updated on
The problem in this sector
"Where is my package?" keeps coming back at customer service, alongside return questions. Same question, every day.
| Aspect | For this sector |
|---|---|
| Channels | E-mail · WhatsApp |
| Own knowledge | Product info + live link to PostNL, DHL, GLS |
| Deployment | Nova Air |
| Expected outcome | An order-status question gets a track-link from the integration back. The team keeps the rest. |
How far your sector has come
For trade, CBS measures something specific that applies directly here: almost half of businesses in trade use AI for marketing or sales. That is the highest application within that sector. Customer contact sits right alongside it, because an order-status question is the final step of that same sale — and it is exactly the volume absorbing your support team today.
Where the limit sits
The constraint here is money and rights. A customer requesting a refund is invoking a right — and exactly what applies depends on when they bought, what they bought and whether it has been used. Nova prepares the complete file with date, order and reason, but the decision to refund or grant an exception is one that directly affects the customer. That stays with a human.
Concretely, these are the things Nova does not do on its own here:
- Issuing refunds, granting discounts or promising exceptions on its own.
- Judging whether a return falls within the period or meets the conditions.
- Handling a complaint that concerns damage or liability.
That limit does not live in a setting someone can flip by accident; it is established up front. That is the difference between a system you know the behaviour of and one you find out about afterwards.
Consumer law is a party to the chat
Whatever the assistant says in a sales conversation, you say as the trader. That brings consumer law into a place nobody looks for it: the chat.
The mandatory information from article 6:230m of the Dutch Civil Code has to be there before the customer is bound. The fourteen-day cooling-off period from article 6:230o you name as a legal right, not as a courtesy of the house. And a statement that is factually incorrect amounts to a misleading trading practice under article 6:193c — intent is not required.
That shapes the setup more than the GDPR does. The assistant picks from fixed texts and pulls facts from your systems; it does not phrase a delivery time, condition or exception itself.
How that plays out in a sales conversation is covered in AI in your webshop and consumer law, down to what has to be on screen before the order button.
The general rules on automated decision-making and transparency are set out in AI and the GDPR for SMEs and the AI Act for SMEs.
What we need from you
This differs per sector, so it sits here rather than in a generic checklist:
- Your shop platform and whether Nova may look up order status or only refer on.
- Your returns policy as it actually is, not as it appears on the page — those two often diverge.
- Which cases always go to a human: damage, quality complaints, and anything involving money.
How a project runs — intake, analysis of your existing communication and a shadow week in which nothing is sent yet — is set out on the approach page.