Split shipments: How to handle them with AI
14 Mar 2025
5 min read
Having your customers’ orders shipped across two or more shipments is common for a lot of ecommerce brands. But they can create a headache if you are trying to deal with a WISMO (where is my order) customer service inquiry. Customer service agents have to dig in to see the whole order and find out what is happening with individual parts of the order.
Automating these kinds of WISMO tickets can be a real nuisance. If you are pulling the headline tracking status in order to form a response, then it will likely either say “Delivered” or “In Transit” but it won’t tell you that some of it is delivered, and some of it isn’t, and so on.
Split shipments is something that we can handle at DigitalGenius. Here’s how it works.
When would you have split shipments?
Broadly speaking orders get split into multiple packages for two major reasons:
Different items are being shipped from two different locations
One item is being shipped in multiple packages
We see the first scenario quite often regardless of the type of brand, but the second one mostly affects brands shipping large items such as furniture, where there is some self-assembly required. Occasionally we see both situations occur.
Let’s take each in turn.
Scenario 1: Items shipped from different locations
In this instance, most ecommerce platforms would break down an order into its constituent parts, each of which should have its own tracking code. Being able to track each part is fairly straightforward for a human.
If a customer was to ask a human agent about their package, the agent could go into each order and check the status, the journey and come back with an answer. For an AI agent, it needs to be able to go through the data line by line, and go deeper.
The pitfall that a lot of AI providers fall into is that they may just read the first tracking order or the master status, and may not go that level deeper to find out what’s really going on.
Scenario 2: One item shipped in multiple packages
This scenario is a little bit more challenging. Certain carriers may not show the information that would be required to see what was happening to each individual box.
But carriers like Fedex do show it. In this situation there is an overall master tracking, as well as child tracking for each individual box. Once again, for a human agent, they would need to check each one to see what the status is, and this is the behaviour that an AI agent would have to mimic.
Why is being able to handle split shipments important?
Quite simply, if you are not able to see each part of the shipment then you can’t give customers a helpful response if part of the package has not arrived.
Imagine you are a customer and you’ve ordered a bed to assemble at home. But you’ve only received the base, and not the headboard. So you go to ask customer service what’s happening to your order.
A less advanced AI Agent might just look at the master tracking, or the tracking information at the top. This might say that the item has been delivered, so the agent gives a response to that effect.
It could just as easily say that the item is still in transit, which would be useful, but it’s basically a lucky guess.
That’s better than some chatbots, which wouldn’t even check the status but would rather just send you the tracking link, meaning that you have to find a way to speak to a human.
This obviously creates a bad experience for the customer because they could be getting wrong or useless information back from the bot, causing understandable frustration.
So if you regularly do split shipments, handling WISMO queries automatically is beyond you. Unless you use DigitalGenius.
How it works in practice
At DigitalGenius we have built deep integrations with ecommerce platforms in order to access this level of information and pull the relevant data to provide the best customer experience possible.
Our AI Agent would access the ecommerce platform through an API and check each item individually. Assuming that one part has been delivered, and one part is still in transit, it could then craft a message using that context. Something like:
I can see that part of your order has already arrived, but the other part is still in transit and expected to be delivered tomorrow. Can I confirm that you have received the first package?
The agent has consolidated all of the information from the platform and provided a summary message, rather than listing out each individual item and where it is. You can then apply business logic to flag an issue, for example if a package has been stuck for more than X days, you could take action. Or if the customer says they didn’t receive the first package, you could trigger a response to deal with that.
Something similar happens when it has to be done through a carrier. Assuming that the carrier shows “master” tracking and “child” tracking and it’s accessible through an API, then our AI Agent can pull the information for each part and then summarise it.
This saves time compared to a human agent, because the agent isn’t having to go to multiple platforms and enter multiple tracking links in order to see what’s happening. And the consolidated message means that the customer is only getting one response rather than an individual response per tracking link.
Some caveats
Being able to provide this does depend on the tech stack you are using.
The first scenario – different items from different places – requires your ecommerce platform to split the order into different tracking links. Most platforms should do this by default, or allow you to configure it. Shopify is one that does for example.
The second scenario – one item in multiple packages – requires the carrier to have individual tracking for each box. This is not as common across the carriers we work with. It’s also fair to say that we haven’t checked this for every single carrier. One that does do it is Fedex.
So if you are handling split shipments through Shopify and Fedex, then we absolutely can help you. If you have another combination, do speak to us and we’ll do some digging.
