Case study

How Power Digital governs Claude usage with a 51-control review

phData reviewed Power Digital’s Claude Enterprise configuration, aligned access and model governance with Claude usage, and established controls for an organization-wide Claude deployment.

51
controls reviewed across Power Digital's Claude Enterprise organization
15
governance decisions documented for Power Digital's owners
6
platform deliverables shared for ongoing visibility and control

At a glance

Industry Professional services
Scale A performance marketing agency scaling Claude Enterprise across a growing user base
Challenge Access, spend, and model governance had not kept pace with organization-wide Claude usage.
Technology Claude Enterprise, SSO, Okta, Cowork, Code, Compliance and Analytics APIs, Streamlit
Results 51 controls reviewed: 31 passed, 5 gaps identified, 15 decisions documented
Phdata Service AI Governance & Enablement, Anthropic
The problem

Claude usage outpaced governance

Power Digital, a data-driven growth marketing agency, needed a shared Claude foundation before Claude usage outpaced governance. The company needed a shared Claude foundation before Claude usage outpaced governance. The company had moved beyond an early-adopter footprint, but access and spend needed clear limits and named owners as the user base grew. 

The platform review found practical gaps. These included:

  • Access control gaps: role enforcement was not in place, allowing users access to models or capabilities they were not meant to. 

  • Spend management and attribution: spend limits had drifted from actual usage. There was no monthly organization-wide limit to prevent runaway spend. 

  • Gated access: there was a desire to provide users access to higher capabilities and spend limits based on demonstrated experience through learning. The existing org structure did not allow for this though.

The team needed a way to see usage and cost without adding a separate platform dependency, so the goal was deliberate Claude access, visible spending, and a repeatable way for owners to adjust controls as usage changed.

Claude Enterprise is Anthropic’s enterprise deployment of Claude, which includes centralized administration for access, spend, and model governance across an organization.

Spend-tier roles are access levels, such as Foundation through Super User, that pair a set of Claude capabilities with a corresponding spend limit.

Compliance and Analytics APIs are Claude Enterprise interfaces that expose usage and governance data so admins can review access and spend without a separate reporting platform.

What phData did

Four decisions shaped the delivery.

phData checked 51 controls across Power Digital’s Claude Enterprise organization. The review covered SSO, connector restrictions, Okta groups, spend-tier roles, model governance, Cowork and Code access, and usage visibility through the Compliance and Analytics APIs. Thirty-one controls passed, five gaps were identified, and 15 decisions were documented for Power Digital.

phData compared actual usage with tier assignments to ground policy decisions in employees’ use of Claude. The FinOps / Value Optimization work clarified the Foundation through Super User progression, gave owners a measured basis for model and access decisions, and created a path for ongoing review without publishing sensitive thresholds.

phData mapped role-based model guidance to the existing tiers, reviewed managed permissions, and documented default and routing guidance across available models. Cowork and Code access controls were tied to tier ownership instead of individual exceptions.

phData shared six platform artifacts: a post-remediation report, model-governance guidance, a model-selection guide, FinOps / Value Optimization guidance, a cost-and-context optimization guide, and a Streamlit app for usage attribution.

The result

51 controls, 15 decisions, Claude ownership made visible

phData reviewed 51 controls across Power Digital’s Claude Enterprise organization: 31 passed, 5 gaps were identified, and 15 decisions were documented for Power Digital’s owners. Access, model choice, governance gaps, and follow-up decisions are now visible and owned, giving Power Digital a Claude foundation it can build on as usage grows.

  • Reorganized users into new tier-aligned groups so they can progress to higher capabilities and spend limits based on required learning.

  • Chargeback attributed Claude spend to individual departments through the FinOps / Value Optimization work.

  • Reference six shared platform deliverables, model governance guidance, a model-selection guide, FinOps / Value Optimization guidance, a cost-and-context optimization guide, a post-remediation report, and a Streamlit usage-attribution app, to manage the platform going forward.

  • Review usage and cost directly through the Compliance and Analytics APIs, without adding a separate platform dependency.

51

controls reviewed across Power Digital’s Claude Enterprise organization

15

governance decisions documented for Power Digital’s owners

6

platform deliverables shared for ongoing visibility and control

Ready to see what a similar engagement could look like for your organization?

Get a scoping conversation with phData’s AI governance team.

Why this matters beyond this project

This work provides a repeatable pattern: review the platform against one standard, tie access and spend to measured Claude usage, and make ownership visible. It gives Power Digital a Claude foundation that can grow while governance evolves with usage. Organizations moving from early Claude experimentation to wider adoption need a working control model with named owners, role-based access, model guidance, spend visibility, and a reporting path that can adapt as usage changes.

Ready to see what a similar engagement could look like for your organization?

Related customer stories

How Order.co used agentic AI in procurement to automate vendor ordering
Procurement / Operations

How Order.co used agentic AI in procurement to automate vendor ordering

See how phData used agentic AI in procurement to automate Order.co's vendor ordering in 6 weeks, with a 100% success rate using Anthropic Claude on AWS.

Read more
Financial services

FSI Giant Uses RAG AI Chatbot for Contract Inquiries

Explore how a large financial organization worked with phData to create an AI chatbot to retrieve contract language quickly and accurately.

Read more
How a K-12 edtech company built a personalized learning platform for million of students
Education technology

How a K-12 edtech company built a personalized learning platform for millions of students

Discover how one of the largest K-12 edtech companies built a production-grade personalized learning platform with Claude on AWS Bedrock and Snowflake, saving $1M+ annually.

Read more

Frequently asked questions

A Claude Enterprise governance review is an audit of an organization’s access, spend, and model controls against a defined standard to close gaps before Claude usage scales further. For Power Digital, phData checked 51 controls covering SSO, Okta groups, spend-tier roles, model governance, and Cowork and Code access.

Gated access ties higher-capability models and spend limits to demonstrated experience, so users progress through tiers such as Foundation through Super User as they complete required learning. phData reorganized Power Digital’s users into tier-aligned groups to support this progression.

Organizations can attribute and chargeback Claude spend by department using the usage and cost data exposed through Claude’s Compliance and Analytics APIs. phData built Power Digital a Streamlit app on top of this data specifically for usage attribution.

A company should review its Claude governance model once usage moves beyond an early-adopter footprint and before spend and access drift from what leaders intended. Power Digital commissioned its review for this reason, checking 51 controls before scaling further.

Get started

Ready to bring order to your Claude Enterprise deployment?

Meet with phData to determine your AI governance approach.Meet with phData to determine your AI governance approach.