The use case: how Deliveroo deploys Rossum
Deliveroo uses Rossum to capture and validate data from supplier invoices, including very large grocery invoices. Rossum receives invoices by central email, classifies and routes them, matches them against POs, checks for duplicates and posts them in Coupa.
More reported results
Figures published in Rossum’s case study. They are not independently measured by usedby.
HOP invoice processing time dropped from 15 minutes to 45 seconds, with an average automation of 35% and 97.6% data extraction accuracy.
The implementation received overwhelmingly positive feedback across the organization, with over 500 daily users logging into Coupa.
The final “big bang” switch to Coupa and Rossum occurred in April 2025, bringing all 11 countries online.
In the customer’s words
We have rebuilt procurement from scratch at Deliveroo, creating a new governed and efficient S2P process. While Coupa acts as the backbone for that process, Rossum delivers massive efficiency gains to our AP teams by automating invoice data capture for our complex invoices and delivering accurate, validated data back to Coupa.Other AI tools used by Deliveroo
Other public relationships recorded on usedby.
Explore Deliveroo’s AI stackRossum users in Logistics & Supply Chain
Companies in the same sector, with public evidence.
See all Rossum customersGo straight to the source
Based on Rossum’s published case study. We check the public evidence; the results are reported by the vendor.
rossum.ai- Source published
- Sep 16, 2025
- Last checked
- Nov 15, 2026
Open source capture What the story claims, and what we checked
What we compared with the page.
- The publication dateChecked
Read from the page’s own metadata, never guessed.
- Deliveroo uses RossumChecked
Confirmed line.
- The result itselfNot checked
We quote it; we did not measure it.
Invoice processing efficiency increased by 23x, with individual transactions jumping from 2 to 45 per hour per user. HOP invoice processing time dropped from 15 minutes to 45 seconds, with an average automation of 35% and 97.6% data extraction accuracy.







