How to deal with damaged item queries, automatically

15 Jul 2024

5.5 min read

A common issue that ecommerce brands have to face is that items get damaged in transit, or arrive in a poor condition at the customer’s address. This is obviously a negative situation that the customer service team need to work to rectify as quickly as possible. 

To speed up the process, this is something that can be done automatically through AI. It can be partially or even completely automated depending on the situation and what you have found. 

Here’s how to automatically handle damaged item queries using AI. 

First steps - map out your manual process

When a customer complains about an item being damaged in their order, what is your normal process?

In most cases it will look something like this: 

Step 1 - Get the customer’s order number and pull it up to confirm that it is a genuine order, and also that it is a recent order. 

Step 2 - Ask the customer for visual proof of damage, if they haven’t already provided it, so either a photo of visible damage, or a video of the item not functioning. 

Step 3 - With that, then offer a customer the option of a refund or a replacement.

Step 4 - Optionally, ask them to send back the damaged item if there is a chance you can repair or recycle it. Send them a return label or QR code for a drop off. 

Obviously there may be other smaller steps in between but this would cover most of the big steps involved. 

The key here is to look at the process and see the steps that can be automated through AI. All of these steps can be done through chat, or email or anything text-based channel the customer is using.

Collecting the necessary information

Getting all the information you need to verify that this is a genuine order and that an item is damaged is the first part that can be done via AI. 

This is actually a relatively simple application of AI. Through Intent Detection, DigitalGenius’s AI can understand what the customer is asking. Once it’s identified that it is about a damaged item, then it needs to check which order it relates to, and then set about collecting proof.

So the AI can set about asking for the order number, and then asking which item(s) within the order is affected if it hasn’t been provided. It can then do a basic verification that the order is a genuine recent order (i.e. was the order made in the last X days). 

Gathering proof, and verifying the information

At this point, the AI has established which order this is and which item(s) is damaged. Then it can ask the customer to send proof. So it can ask the customer to submit a photo or video.

Once it has that, you could at this point decide to pass all this information over to an agent. Where AI has helped is that it’s gone through the basic process steps that can take some time. This is better than having an agent sitting and waiting for all the information to come back, because now they have everything they need to get started with the next steps.


However, if you want to fully automate the process, you have other options here. 

If the customer is a loyal customer with plenty of previous orders that didn’t have issues, you could choose at this point to take them at their word and just refund or replace the item without checking.

Or if the cost of the item is so low that it’s not worth the agents’ time, then you would want to bypass them entirely. 

That is a business decision for you, but if you have rules set up in this way, you can create a flow that if certain conditions are met then the customer is cleared for an automatic refund. 

Using Visual AI to detect damage 

Another option is to use Visual AI to assess the images or videos and detect damage. DigitalGenius’s Visual AI can be trained to look for common types of damage. Whether that’s holes in the fabric of shoes, cracked make up, or wilted plants, with enough examples the AI can be trained.

When it looks at a new image it can then give a confidence score based on what it sees. Once you have tested it on your own products, you can then decide what confidence score you would be happy for the AI to approve. You may want to add that to the business criteria set out above. 

Measures to reduce fraud

Of course, it’s always possible for someone to fool an automated system. That’s why it’s important to set rules about when to automate a refund or replacement. For example, you may choose to handle first time orders manually and check everything that has been sent, but for loyal customers you may allow automation. 

One thing that some retailers do is to ask customers to sign or authorise a legal declaration that what they are submitting is accurate and which spells out any consequences for fraud. This has been quite effective when we’ve seen it done, and companies who start implementing it do see a reduction in refunds given. 

This declaration can be sent out automatically as part of the conversation flow process.

Issuing the refund or replacement

Once you are satisfied that your business rules have been met, you can then issue the refund or replacement - whichever the customer has picked. 

You could issue this as an internal note for an agent to process it. However, through deep integrations with your payments system and your order management system, you could do either automatically. 

In both cases, the AI could take the relevant information (address, card details etc.) and pass it back to the appropriate system to enable a quick resolution. 

Returning the item

Finally, AI can help if you want the damaged item to be returned. Generating return labels is time-consuming and frustrating for an agent. But AI can once again take the relevant information and through an API integration, submit that data to a carrier and generate a pre-paid label.

Then all the customer has to do is drop it off wherever it’s most convenient, and you can possibly reuse or recycle the item. 

An end-to-end automation for damaged items

Here we’ve walked through the steps of automating a “damaged item” process and how AI can do the whole journey.

Even if you do not want to do the whole process automatically, you can clearly see that AI has a role to play at various stages in speeding up the process and ensuring that customers get the right resolution fast.

To find out more about how DigitalGenius works, watch our WISMO (Where Is My Order) demo here. Or meet the team for your own demo.