DPD UK automates 55% of customer queries with Dialogflow

Leading parcel carrier DPD UK delivers and collects over 1M parcels across the region every day. The company wanted to improve its customer experience by adopting an AI solution that would resolve queries in its YourDPD app more efficiently. Using a rich text-based conversational experience built with Dialogflow it managed to automate over 55% of customer queries.

About DPD UK
Leading parcel carrier
Part of DPDgroup, founded in 1976
Delivers 1M parcels in the UK every day
Goals
Improve customer experience
Adopt AI solution to quickly resolve issues
Approach
Broke down queries into customer events
Used Dialogflow to build new intents and train the model
Monitored analytics to identify gaps
Activated Small Talk
Results
55% of customer queries resolved with AI

Every day, DPD UK delivers and collects over one million packages across the region, helping small, medium and large businesses to ship online orders for clothes, phones, packaged food, or business supplies safely and on time. More than 10,000 staff at 91 locations handle the logistics that make that happen, while another 500 staff at four call centres deliver customer support at scale.

Focusing on improving its services for customers DPD UK embraces new technology such as YourDPD, its parcel tracking app, launched in 2016, and now with over 3 million users.

“The app has been incredibly successful” says Sinead Croke, Director of Customer Experience at DPD UK. “One of the most popular aspects of it is its chat option. When customers address us through the app, they are immediately identified without having to answer frustrating security questions on a phone call. With 100,000 new users downloading the app every month, it became a challenge to keep pace with demand”

“We found that about 70% of queries could be answered immediately - such as the standard delivery time of a parcel. The chat format is the perfect way to relay this information quickly and effectively”

– Sinead Croke, Director of Customer Experience at DPD UK

In just two months after the chat function was launched, the team handling the chat requests grew exponentially, from 12 to 50 personnel with over 4,000 interactions a week.  To ease pressure on staff, Sinead and her team looked to a rich text-based conversational experience, or chatbot, to manage contacts and help resolve basic queries.

“We found that about 70% of queries could be answered immediately - such as the standard delivery time of a parcel,” says Sinead. “The chat format is the perfect way to relay this information quickly and effectively”.

How Dialogflow and Google Cloud helped?

“We wanted to leverage the investment and research from an AI company, so we could maximise our time building a great user experience for our app users,” says Max Glaisher, Product Innovation Manager at DPD UK. “We looked for a cloud solution that we could learn how to use and operate ourselves and which could keep pace with future developments. Dialogflow has an open approach that makes it very easy for us to take on new channels without expensive integration work, and the platform was self-explanatory, well-documented and easy to use. Choosing Dialogflow was a no-brainer”.

In the beginning, Max and his team ran a proof of concept using Dialogflow on a mobile chat app, using anonymised transcripts of customer conversations to analyse demands. “Usually customers were looking for status information, such as where their parcel was,” says Max. “We sorted those queries into four or five different event types and used Google best practices to design ways of presenting responses to users, which might include maps, images, or prompts, as well as chat responses”.

“People can say thank you in a million different ways. For a decision tree model, that can be a real problem, but the intent model on Dialogflow doesn’t need the wording to be an exact match. That means it understands what customers are looking for much more accurately.”

– Max Glaisher, Product Innovation Manager, DPD UK

From there, DPD UK uses the Dialogflow dashboard to build new intents and train the model. They also monitor analytics in the Dialogflow console and Chatbase for performance and to highlight gaps that need to be addressed. To help make customer experiences more natural, Max and his team enable Small Talk, helping to bridge gaps in conversations without the need to write all of their own informal responses.

“People can say thank you in a million different ways,” says Max Glaisher, Product Innovation Manager at DPD UK. “For a decision tree model, that can be a real problem, but the intent model on Dialogflow doesn’t need the wording to be an exact match. That means it understands what customers are looking for much more accurately”.

Easy integration with multiple channels to leverage learning

“When we set up the experience, we predicted it would serve around 10% of customer queries,” says Sinead. “But when we switched it on, that number was about 25%, and still rising.” Now the team have now moved the YourDPD app onto Firebase, and have ready to hit a resolution rate of 50%, dramatically reducing the need for human advisors to take over on support conversations. At the same time, DPD UK is looking to leverage the learning of their experience with integrations across multiple channels, including email and messaging services.

“To serve our customers, we need a more agile environment than other cloud platforms can provide. Whether using Google Cloud Translation API to identify parcel data in different languages or Dialogflow for conversational experiences, the Google Cloud Platform managed services and infrastructure help us bring innovative products to market faster.”

– Max Glaisher, Product Innovation Manager, DPD UK

Leveraging the latest innovations