Entitlements Management

As DataRobot scaled, feature access was controlled through a mix of hard-coded logic, feature flags, manual provisioning, and tribal knowledge spread across teams.

Over-provisioned accounts retained access to increased compute capacity and paid features long after contracts ended, contributing to hundreds of thousands in annual leakage.

Customers occasionally lacked access to features they had purchased, leading to confusion and support escalations.

We could not safely run trials, demos, or upsell experiments at scale because the whole setup was so brittle.

I made the decision to build an entitlements management service that would remove the error-prone manual processes and improve governance. I believed that this would eliminate the revenue leakage and reduce the burden on support, all while laying the foundation for future growth motions.

I identified accounts that were in violation and began laying plans to ensure it couldn't happen again. In tandem, I started exploring ways that automated, governed access could be used to drive growth.

One source of truth for feature access provides clarity to internal teams and customers.

Entitlements should be modular and can be added/removed independently in designated quantities.

A time component ensures that access is removed when no longer valid, eliminating manual cleanup.

Teams can hook into triggers in the customer’s lifecycle to layer in additional actions.

1. Defined the Entitlements Model

• Authored the long-term vision and core schema mapping for customers, phases, features, limits, and time-bound access.
• Partnered closely with Engineering, Security, Support, and Analytics to clean up feature flag tech debt and ensure enforcement was auditable.

2. Proved It with Trials

• Replaced manual trial provisioning with governed entitlements.
• Validated that features turned on and off correctly under real usage and cost constraints.

3. Extended to Sales-Led Demo Environments and Paid Customers

• Adopted entitlements for sales-led demo environments, reducing operational drag on support.
• Extended the model to paid customers and used to provision 24 separate product SKUs.

4. Unlocked Reverse Trials

• Paying customers could now self-serve temporary access to premium features like Generative AI.
• Added hooks to CRM to allow account teams to follow up.