Creating a better courier onboarding experience for Evri using GenAI


Evri_case study illustration_hero large
The client


From their roots as a mail order company nearly 50 years ago, Evri has grown considerably over the decades. Evri launched in March 2022 following a successful rebrand from Hermes UK. Now, with a team of 18,000+ couriers, 8,500+ local one-stop ParcelShops and Lockers, and a growing network of state-of-the-art hubs and depots, Evri provides delivery solutions for anyone who wants to send a parcel in the UK and to more than 220 international destinations.

Evri_case study illustration_client

Key Outcomes

We developed a robust Ai-powered chatbot at pace, upskilling the Evri Innovation Team in tandem, and providing them with the data needed to build a business case for further investment.

  • Pace A robust chatbot capability was developed at pace, enabling Evri to test the business value
  • Upskilling The Evri Innovation Team was upskilled in both Generative AI technology and agile delivery practices
  • Data Clear metrics and measurements were created to quantify the success of the trial, allowing Evri to build a data-driven business case for further investment
The Challenge

Seasonal spikes in demand made it challenging to onboard new couriers

Evri experiences a significant spike in demand every year in the last quarter, which is driven by Black Friday/Cyber Monday and Christmas. To meet demand, Evri recruits and retains around 7k new couriers annually. Evri was keen to understand why some couriers leave sooner than expected, and make changes to increase courier retention.

This year, Evri looked to meet the peak in demand by scaling their courier network without needing to scale their courier support teams by the same ratio. A chatbot called Elmo was envisioned to help address the routine questions from new couriers and speed up the response rate for couriers leading to a better onboarding experience.

Evri_case study illustration

An AI-powered chatbot PoC to establish business value

We were brought in to work in a blended team with the Evri Innovation Team to turn their proof of concept (PoC) into a working prototype at a speed that would allow Elmo to be tested in the real world ahead of peak season, and validate if Elmo will realise the intended business value.

We created a three-phase approach to taking the existing PoC and creating a working prototype to test the core hypothesis. The three phases were: ‘productionise’, ‘increase quality’, and ‘feature enhancements’.

The ‘productionise’ phase focused on making the application fit for trial by embedding best coding practices, carrying out performance testing, defining and implementing success metrics, improving stability, ensuring consistent and correct answers to frequently asked questions, introducing standard disclaimers, improving the user interface and blocking sensitive data. Additionally, the team ensured that answers returned by Elmo were only from content within the provided documents and not other sources.

The ‘increase quality’ phase focused on improving the quality of Elmo's responses by introducing context-aware splitting, refining the prompt passed into the solution for further context, and introducing a feedback loop to allow users to provide feedback on the quality of the answer provided by Elmo.

The ‘feature enhancements’ phase focused on introducing features that add a "wow" factor, such as allowing couriers to use their voice to ask questions of Elmo and have the answer read out to them. This functionality reduces the need for couriers to interact via the screen of their device and speeds up the interaction time.

An illustration depicting an AI chatbot

The building blocks for an effective AI-powered chatbot

  • Open_AI_logo
  • AWS_Logo
  • google-cloud-logo-300x150
The Results

A chatbot capability that enables Evri to test their value hypothesis

A robust chatbot capability was achieved quickly, enabling Evri to test their value hypothesis and validate if scaling their courier support capability without scaling the size of support teams was achievable.

Elmo has three core capabilities: Ingestion, Chat, and Analytics & Reporting:

  • Ingestion provides the ability to convert source documentation into a format which can be used by the large language model (LLM).

  • Chat provides a user interface (both visual and audial) to allow users to ask a question and receive a response.

  • Analytics & Reporting provides the ability to understand the performance of the solution in terms of both user behaviour and quality of the responses provided. 

Through working in a blended team with AND, the Evri Innovation team was upskilled in both the technology used and agile delivery practices.

The introduction of clear metrics and measurements allows for the success of the trial to be clearly quantified and articulated, and ensures that Evri can make a data-driven business decision whether to further invest in the solution.

An illustration depicting clear and measurable business analytics
Case studies

See our impact elsewhere

More successful digital transformation projects...

Want to learn more about the work we do with clients? Take a look below.

Talk to us.

Get in touch to discuss your digital ambitions.

Contact us