How WPP Won a Competitive 3-year Renewal by Proving 30-40% Efficiency Gains with Devin

WPP Enterprise Solutions is the business transformation operating unit of WPP, the global creative company headquartered in London. One of WPP's four strategic operating units, it brings together over 12,000 experts across 40+ markets, designing, building, and operating the growth systems powering global brands like IKEA, Ford, L'Oréal, and Nestlé. The company has partnered with Cognition to make AI-driven software delivery the foundation of how it builds for clients, with engineers across the UK and India working alongside Devin today.

For a business whose primary offering is engineering delivery, clients' demands for faster delivery at lower cost and competition from offshore providers put pressure on every renewal. WPP Enterprise Solutions rebuilt its software development lifecycle around Devin and baked the resulting efficiency gains directly into its contracts: full scope at a fixed price with a contracted delivery velocity, fully backed by sprint-level data.

Evaluating solutions

Prior to Devin, WPP Enterprise Solutions had already invested heavily in AI tooling. They previously rolled out an AI coding assistant to 500 engineers yet saw none of their tracked delivery metrics improve.

The team went back to the drawing board, ran side-by-side evaluations of different leading platforms, and chose Devin Desktop because of its ability to quickly and comprehensively build context on large, complex client codebases better than any alternative.

The Halfords proof

The proving ground for Devin was Halfords, the UK's leading provider of motoring and cycling products and services and a client of more than ten years. Facing a competitive renewal against a low-cost provider, WPP Enterprise Solutions put Devin to the test: rebuild Halfords' e-commerce platform on headless Salesforce Commerce Cloud. The project was fixed price with a twelve-week timeline, and the team deliberately estimated it the traditional way, without AI, to establish a clean baseline for measuring productivity gains.

The build delivered 110% of the contracted scope in 25% less time. Unit test coverage rose from 80% to 99%. Efficiency gains, tracked as estimated versus actual effort per discipline per sprint in Jira, averaged 32% across the lifecycle, with the largest gains realized in business analysis and technical architecture.

The results allowed WPP Enterprise Solutions to secure a three-year renewal that commits the team to a baseline delivery velocity per sprint, scoped using WPP's standard pre-AI estimates. WPP made a bet to staff the engagement with a leaner team while still overdelivering on contracted output, trusting Devin to fill the gap.

Eight months in, the smaller team consistently delivers 20-35% more output per sprint than initially contracted, according to Halfords' Delivery Director. Automated test coverage has maintained near 100%, no post-delivery work has been required, and the rebuilt platform ran at 99.97% uptime through Halfords' busiest trading period of the year.

The same methodology powered by Devin is now delivering results across the client portfolio. On a global sportswear brand engagement, the team reverse-engineered a legacy platform and produced more than 40 technical documents and 332 test cases in four weeks, work that typically takes around 10 weeks using traditional techniques. The client's prior test coverage in this area was effectively nil, with no E2E or accessibility coverage and only fragmented Cypress snippets, so the new test cases represent a net-new quality capability. On another client engagement with a global beverages brand, a delivery team took on a test-framework migration estimated at 80 person-days of effort and finished it in eight.

How WPP deploys Devin

Client delivery. Devin Desktop is the default IDE across the engineering community. Since WPP's engineers land in a different client's codebase on every engagement, they heavily lean on Devin's ability to rapidly build context on large, unfamiliar systems, then put it to work on code generation, test creation, refactoring, and legacy code analysis. Devin understands the client-specific rules files that encode that client's coding standards, security requirements, and architectural patterns, allowing it to behave correctly inside every engagement.

Repeatable enablement. To scale the productivity gains across the engineering organization, WPP built its enablement layer on the platform's customization primitives: a shared skills library, reusable prompt and context patterns, and structured onboarding for every engineer, creating a self-improving playbook to apply to each new client engagement.

Full-lifecycle coverage. Some of the largest measured gains sit left of engineering: business analysts and architects use Devin to turn unstructured requirements into user stories, non-functional requirements, and architecture diagrams as code.

Model flexibility. Engineers plan with frontier models and execute with more cost-effective options, toggling between models for each stage of their workflow within one enterprise-grade platform. On fixed-price engagements, that choice becomes a significant margin lever.

"An engineer might plan their work with a frontier model, then execute with SWE-1.6 because it feels like they're still in the code, working with a copilot… Giving engineers that choice is huge."

— Shahil Bhimji, Chief Engineering Officer, WPP Enterprise Solutions

Autonomous execution. The current Devin Cloud pilot targets necessary but time-consuming work that gets deferred sprint after sprint, starting with the migration of a legacy suite of hundreds of Selenium tests at Halfords to a modern Playwright framework. Devin Cloud agents run in their own environments with a shell, an editor, a browser, and a build environment, enabling them to execute the full write-test-verify loop and return finished, verified work ready to be merged.

The partnership

The partnership goes far beyond a platform license: Cognition engineers work hands-on with WPP's engineers to scope each new use case and drive org-wide adoption. A pooled consumption model means any WPP engineer can pick up Devin on any client engagement the day it starts. The two companies are now co-building a joint go-to-market motion on a guiding principle: WPP wins client work, centrally powered by Cognition.

"With new clients it's always AI-assisted software delivery, never tool-first. We've completely reshaped how we deliver software, augmenting every part of the SDLC so the end-to-end gets the 30-40% efficiency gains we've proven in live environments. And we regularly talk about our partnership with Cognition as the foundation of how we deliver that."

— Shahil Bhimji, Chief Engineering Officer, WPP Enterprise Solutions

What's next

WPP Enterprise Solutions plans to expand Devin usage to its broader global engineering community, extend agentic workflows across the full lifecycle from business analysis through operations, and bring the methodology to new sectors.

We're excited to continue working with WPP Enterprise Solutions as they define what AI-native services delivery looks like at enterprise scale. Organizations interested in deploying Devin at scale can contact our enterprise team to learn more.