How Anthropic’s latest Claude Enterprise updates help organizations track adoption, understand ROI, and manage AI spend with confidence
Introduction
As AI moves from an experimental tool to a core part of how organizations get work done, a new challenge has emerged for IT and finance leaders: how do you actually manage something like Claude at scale? Unlike a traditional SaaS seat license, an AI assistant’s usage and cost patterns shift constantly as employees discover new use cases, adopt new models, and lean on Claude for increasingly complex, agentic work.
On July 2, 2026, Anthropic announced a meaningful step forward on this front — a set of new analytics and cost-control features for Claude Enterprise. The update, published on the official Claude blog, introduces richer admin analytics, model-level entitlements, and spend alerts designed to give administrators both a clearer picture of how Claude is being used across their organization and finer-grained tools to keep costs in check.
This isn’t Anthropic’s first pass at admin controls — it builds on an existing foundation of spend caps, access and model routing, a usage analytics dashboard, and effort controls. What’s new is the depth and precision of the visibility now available, plus the ability to act on that visibility automatically rather than reactively.
Why This Matters Now
Enterprises adopting Claude aren’t just running one-off chat sessions anymore. Employees are creating artifacts, editing files, using skills, connecting third-party services, and running Claude Code across development teams. Each of these activities carries a different cost profile, and without the right tooling, it becomes very difficult for an admin to answer basic but important questions: Which teams are getting the most value? Where is spend concentrated? Are certain models being overused for routine tasks that don’t need them?
Anthropic’s answer is to treat cost visibility as an ongoing operational discipline rather than a monthly surprise. As Product Manager Kyra Abbu put it in the announcement, granular spend data paired with regular alerts pushes teams toward continuously reassessing how they use Claude, rather than discovering a problem only when the bill arrives.
What’s New: Tracking Adoption and Cost
The centerpiece of this release is a substantially upgraded admin analytics dashboard. Administrators can now see usage and cost broken down by both group and individual user, with concrete output metrics — artifacts created, files edited, skills used, connectors invoked — displayed right next to their associated cost. Because the dashboard can be filtered using the SCIM groups that IT teams already manage, the cost breakdown maps naturally onto the organization’s existing structure rather than requiring a separate reporting hierarchy.
For engineering organizations specifically, Claude Code now gets two dedicated tabs inside the admin console:
- Usage, which shows active developers, session counts, and the most common commands across the organization, refreshed daily.
- Value, which estimates things like productivity lift, cost per commit, and projected annual value — with every underlying formula visible and adjustable, so admins aren’t stuck trusting a black box.
This kind of transparency matters. It’s one thing to know how much an organization is spending on Claude Code; it’s another to be able to defend that spend to a CFO with a formula you can inspect and tune yourself.
Beyond the dashboard, Anthropic has also expanded Analytics Chat, a feature that lets admins ask usage questions in plain language — for example, asking which teams have doubled their usage in the past month, or which teams are getting the most value per seat — and receive shareable charts in response. This turns what used to require a data analyst or a custom export into something any admin can do conversationally.
For teams that want to integrate Claude’s usage data into their broader financial operations stack, the Analytics API now exposes this data programmatically. That means finance and IT teams can pull Claude spend into tools like Datadog Cloud Cost Management or CloudZero, sitting it alongside the rest of their cloud and AI infrastructure costs rather than tracking it in isolation. The API supports filtering by date range, team, product, or model, and — notably — skills and plugins now report their own usage and cost data, with new endpoints specifically tracking plugin adoption and artifact creation.
Finally, this visibility isn’t limited to admins. Individual users can now see their own cost, product, and model breakdowns, along with progress against their personal spend limits. That means fewer surprise cutoffs mid-task, and it gives employees a better sense of how their own habits — which models they reach for, which skills they lean on — translate into actual spend.
What’s New: Controls for Managing Spend
Visibility is only half the story. The other half is being able to act on what you see, and this release adds several concrete levers for that.
Model defaults and entitlements let admins decide which Claude model a new conversation starts with, across chat, Cowork, and Claude Code. This is a subtle but important control — it means routine, low-stakes work doesn’t automatically default to the most expensive model available. Admins can also restrict which models are available at all, either for specific roles or organization-wide, giving them a way to align model access with actual need.
Spend-threshold alerts now notify admins automatically at 75% and 90% of an organization’s spend limit, giving them a runway to raise the cap before anyone actually gets blocked mid-task. On the user side, individuals get their own in-app notifications at 75% and 95% of their personal limits, and can request a limit increase directly from their admin without needing to leave Claude to do it. This closes a loop that used to require manual back-and-forth — a developer hitting a wall mid-project, pinging IT, and waiting for a ticket to be resolved.
For larger organizations managing limits across many teams, the Admin API now supports moving these cost-control workflows into scripts. That means organizations can automate the review of increase requests, flag users approaching their spend limits, and catch rapidly changing usage patterns — all without manual, one-by-one oversight.
Real-World Signal
The announcement includes a few perspectives from customers that help ground these features in practical terms. CIO Carter Busse framed the challenge succinctly: the goal isn’t to slow down high-performing teams, but to be able to show cost next to business impact so that ROI conversations with a CFO have real substance behind them. Meanwhile, Product Director Ciro Yamada made a point that cuts against a common assumption — that raw token counts aren’t actually the most useful metric. What matters more, in that view, is which skills get used repeatedly across an organization, since repeat usage is a much stronger signal of genuine value than volume alone.
These perspectives underscore something important about this release: it’s not just about restricting spend, it’s about connecting spend to value. A dashboard that only shows “this team spent $X” is far less useful than one that shows spend alongside what that spend actually produced — commits shipped, artifacts created, skills reused.
Getting Started
For organizations already on Claude Enterprise, none of this requires a new contract or a lengthy rollout. Admins can start by exploring the new usage and cost breakdowns directly in the admin console, then move on to setting model defaults and spend limits by group, and configuring spend-threshold alerts so they’re not caught off guard by rising usage. For teams that want this data flowing into existing financial reporting systems, the Analytics API is the natural next step, with getting-started documentation available through Anthropic’s support center.
The Bigger Picture
This update reflects a broader trend in enterprise AI adoption: as tools like Claude become embedded in daily workflows — writing code, building documents, running agentic tasks across connected systems — the old model of flat licensing and occasional spot-checks no longer fits. Organizations need the same kind of financial rigor around AI spend that they already apply to cloud infrastructure: real-time visibility, threshold-based alerting, programmatic access to the underlying data, and controls that scale without requiring a human to manually review every case.
By pairing richer analytics with more granular spend controls, Anthropic is signaling that it sees this operational maturity as a prerequisite for continued enterprise adoption — not an afterthought. For IT and finance teams currently trying to make sense of a rapidly growing AI budget line, that’s a welcome development.
References
- Anthropic. “Giving Admins More Visibility and Control Over Claude Spend.” Claude Blog, July 2, 2026. https://claude.com/blog/giving-admins-more-visibility-and-control-over-claude-usage-and-spend
- Anthropic. “View Usage Analytics for Team and Enterprise Plans.” Claude Support Center. https://support.claude.com/en/articles/12883420-view-usage-analytics-for-team-and-enterprise-plans
- Anthropic. “Use Analytics Chat to Ask Claude About Usage.” Claude Support Center. https://support.claude.com/en/articles/14729354-use-analytics-chat-to-ask-claude-about-usage
- Anthropic. “Analytics API Documentation.” Claude Developer Platform. https://platform.claude.com/docs/en/manage-claude/analytics-api
- Anthropic. “Manage Model Access for Your Organization.” Claude Support Center. https://support.claude.com/en/articles/15694740-manage-model-access-for-your-organization
- Anthropic. “Spend Limits API — Example Workflows.” Claude Developer Platform. https://platform.claude.com/docs/en/manage-claude/spend-limits-api#example-workflows
- Anthropic. “Get Started with the Claude Enterprise Analytics API.” Claude Support Center. https://support.claude.com/en/articles/13694757-get-started-with-the-claude-enterprise-analytics-api