DHL Parcel Benelux is part of the Deutsche Post DHL Group: the world's largest, leading logistics company.
Automation, digitisation, and data analysis enable DHL to improve not only the entire transport and delivery process but also the communication around it. Customer Service receives inquiries from both shippers and recipients; from businesses and consumers. For an easier and faster service, DHL launched a chatbot in 2019.
Read the interview with Natasja Wientjes, Customer Service Director DHL Parcel Benelux, and discover their conversational strategy and their experience with Conversational AI Cloud.
"At the end of 2019, we started with a chatbot. We started small to gain experience and expanded the project quickly. Today the bot is already being used a lot, alongside the traditional telephony and email channels.
Now we want to take the next step, to make the bot more intelligent. We've set the goal to make more digital channels available, to easily answer the more simple questions quickly. First, we want to focus on our consumer recipients.
By starting small, you can take time to analyse the data. What questions are coming in? But also: how can we make the bot more intelligent? So not just a scripted bot, not just a standard answer, but include context, such as time. This way we can provide recipients with an answer in the right context. If you expect your package between 1:00 and 3:00 p.m., you'll get a different answer at 12:00 p.m. than when you chat at 1:30 p.m."
"Yes. At the beginning of this project, we put a lot of energy into establishing relationships with our core systems. This way the bot can retrieve the right information and we can extend this with more context. This way we can help everyone with the right answer."
"For me, it's about two things: to use the right channels, but also to set up the backbone. Otherwise, a channel won't be effective.
As a customer or partner, you want to get an immediate answer via the channel you choose. It is annoying if you don't get an answer via WhatsApp and then still have to call. At the same time, if someone contacts you via another channel, you want to have all the information at hand. What was asked and discussed before?
Our strategy is to launch channel by channel. We arrange it properly on the backend and the frontend to prevent channel hopping."
"Do you believe Artificial Intelligence is going to take over humans? To me, it's much more important to ask: do we want that?
We can answer that with a resounding "no". Why? A bot gives support, it can answer questions quickly and easily. A human can do much more: show emotions, think outside the box, act outside processes. I think humans always need human contact.
We are certainly looking to automate and digitise more. For example, we are thinking about voice. At the same time, I think it's the human touch that is the beauty of customer service."
"The first and the last mile are the most important for DHL. How we can improve the combination of experience, shadowing, and listening in on the delivery person? Why does he/she do things the way he does? And then combine this with data; not only from our systems but also from external systems such as Google reviews etc.
We'll try to become smarter through analysis and improve based on data. Can a device become even smarter with Artificial Intelligence? Can it then identify the main issues on its own?
There is still a whole world open to us to provide even better services."
"That's different for every organisation. You can start using it at any time. For us, it was the moment we saw the need for other digital channels. It's really important to get the backend right, though.
A bot with only standard answers, not linked to systems, has less added value. Our tip is to have a good backbone so that you can quickly and easily provide the correct answer to the user.
Make it more than a FAQ bot, but create a bot that gives really customer-specific, personal answers. That's why we use context. That is a success factor of our collaboration."
"For us, after 'starting small', it became clear that we wanted to scale up. From the consumer market to the business market, but also expand to multiple languages.
How neat would it be to use different languages? However, you don't want to rewrite every bot. You want to work from one source, for all languages.
A third important factor was that we wanted to be in control, including over the intelligence of the bot. Our people have a lot of experience with customer service. They write the scripts and do the analyses themselves. This gives them all the freedom, for example, to respond to current events. Think of problems with a truck, or weather conditions like snow. It's very nice that with Conversational AI Cloud, we are able to include these changes directly in the conversations with the bot."
There are actually a number of them. For example, during the tender process, we learned that some requirements, which we thought were logical, were not necessary after all. To summarise three learnings:
Dare to ask what you don't really know
Start small, gain experience
Ensure a seamless handover to a live employee
Select a region to show relevant information. This may change the language.
Is this region a better fit for you?