Case study · Engagement in progress

Making AI stick inside a €3.8bn consumer brand

The UK arm of a €3.8bn global consumer goods business already had the platform — Google Workspace, Gemini licences, leadership backing. What it didn't have yet was adoption. This is how the programme is built, and what it measures.

Client: €3.8bn global brand
Functions: Five teams
Platform: Workspace + Gemini
Model: Embedded, workflow-first

The situation

A Gemini licence that gets used twice is not an AI capability. It is a line item on a P&L waiting for someone to notice.

This business wasn't starting from zero — the Google stack was already in place, and the brief was never "introduce AI". It was: make the investment pay. Across Sales, Finance, Marketing, Supply Chain and Service, the same pattern showed up — middle managers spending too much of their week on manual data work. Reconciling payments. Compiling reports. Checking forecasts. Chasing information across systems.

AI's job here is not to replace those people. It is to handle the plumbing so they can focus on the commercial decisions only humans can make.

The brief, read back

Leadership asked four questions. The whole programme is organised around them.

Sell more

Promotion insight and account planning that separates real growth from noise.

Reduce cost

Retailer payments, financial trackers, and the admin that eats finance time.

Move faster

Forecast checks, new line forms, and market analysis without the manual lag.

Adopt daily

Coaching, workflow playbooks, and use cases validated by the people who run them.

The approach

The unit of adoption is not the tool. It is the weekly workflow.

Most rollouts fade because people are shown a tool instead of a better way through the job in front of them. The launch creates curiosity — then the account manager goes back to pulling reports by hand and the finance team goes back to reconciling exceptions line by line.

So the programme starts with the task that costs each team time every week, builds a Google-native pattern around it alongside the people who own it, and lets adoption spread from visible relief rather than generic instruction. One constraint is respected throughout: the core ERP and planning systems stay exactly where they are. AI wraps around the systems of record — it doesn't replace them.

Six workflows, not a transformation deck

Each workflow was chosen because a team loses real hours to it every week — ranked by impact, effort, and risk before any build started.

Promotional analysis

Connect sell-in, sell-out and margin data so commercial teams can separate category growth from cross-retailer cannibalisation.

Retailer payments

Match debit notes, terms and allowances automatically — surfacing only the exceptions, with enough context for Finance to approve or challenge.

Account manager admin

Summarise account performance and draft routine updates, so account managers spend selling time selling instead of assembling the story.

Market trackers

Turn external market research exports into internal trackers and presentations without the manual reformatting.

New line forms

Pre-fill retailer listing templates from product data, asset records, approved copy and pricing — so launches don't start from a blank form.

Forecast checks

Sense-check forecasts across hundreds of SKUs, by retailer, by month — with the planning system staying the forecast of record.

Results are measured, not assumed

Every workflow is baselined at week zero — hours spent, cycle times, error rates — and tracked weekly against that baseline. The numbers publish here, with the client's sign-off, as the programme completes.

Hours returned, per team, per week: —
Licensed users active weekly: —
Workflows live in daily use: —

Same pattern, your workflows.

Fifteen minutes, no slides, no pitch. Tell me where your team is now and what's blocking progress — I'll tell you honestly whether there's a quick win worth chasing.

Book a Call with Javan