A Large Account Without an Operating Model
L'Oreal was one of Amazon's largest advertising clients, with a $400M book spanning multiple brands across beauty, haircare, and skincare. The relationship was commercially significant but operationally inconsistent. Fill rates varied. Data quality issues created downstream reporting problems. The account team was reactive rather than systematic responsive to L'Oreal's requests, but not operating from a proactive service architecture.
The opportunity was clear: build a service operating model that treated the Amazon account team as a genuine extension of L'Oreal's own organization with the infrastructure, cadence, and accountability that implied.
What a 99% Fill Rate Actually Requires
Getting a $400M book to 99% fill rate is not a talent problem. It is an architecture problem. Three structural gaps were driving the inconsistency:
No shared operating cadence
L'Oreal's planning cycles and Amazon's execution cycles were not synchronized. Campaigns were being handed off at points in the process where errors were hardest to catch and most expensive to fix. A shared operating cadence needed to be designed and adopted by both sides.
Data quality without ownership
IDQ issues were being discovered late after campaigns had launched or inventory had been committed. The fix was not more quality checking. It was moving the quality gate earlier in the process and assigning explicit ownership for each check.
Reactive account management
The account team was skilled but operating reactively. L'Oreal surfaced problems; the team solved them. What the relationship needed was an account team that identified problems before L'Oreal did and brought solutions, not just responses.
The Brand Concierge Model
The Brand Concierge operating model was designed around a single principle: the Amazon account team should be able to operate as if they were inside L'Oreal's organization, with the same context, the same priorities, and the same accountability for outcomes.
That required four structural changes:
Synchronized planning cadence
A shared planning calendar aligned Amazon's campaign execution timeline to L'Oreal's internal brand planning cycles. Handoffs were redesigned to occur at points where quality could be verified before commitment, not after.
Front-loaded quality architecture
IDQ checks were moved to the earliest viable point in the workflow. Ownership was assigned by role, not by team. Issues that had previously been discovered at launch were now caught at intake. 100% IDQ was the result of a process change, not a headcount increase.
Proactive signal reporting
A weekly reporting package was built that surfaced performance signals, anomalies, and recommendations before L'Oreal asked for them. The account team became the party that identified problems first which changed the nature of the client relationship fundamentally.
Portfolio growth infrastructure
The operational reliability created space for commercial conversations that had not previously been possible. When fill rates are inconsistent and data quality is a persistent issue, growth conversations are difficult. When both are solved, the conversation shifts to opportunity. The 110% YoY growth was a commercial outcome enabled by operational credibility.
What the Model Produced
Sustained across the full book. Achieved through process redesign, not headcount increase.
Data quality issues moved from a persistent operational drag to a non-issue. Front-loaded quality gates were the mechanism.
Commercial growth enabled by operational credibility. When execution is reliable, growth conversations become possible.
The team shifted from reactive problem-solving to proactive signal identification. L'Oreal's internal teams treated the Amazon account team as an extension of their own organization.
The Pattern Behind the Result
The 110% growth and the 99% fill rate are commercial and operational outcomes. The more durable signal is what produced them: treating an account relationship as an operating system problem rather than a talent or effort problem. When the architecture is right synchronized cadences, front-loaded quality controls, proactive reporting performance at this level is not exceptional. It is the expected output of a well-designed system.
This is the same principle that produced the results in every other engagement. The context changes. The approach does not.