How much is your SMB's monthly Claude or ChatGPT subscription actually bringing in? On October 28, 2026, Anthropic launched Claude Smart Reports, a tool that answers this question head-on for its Enterprise customers. According to TechCrunch (October 28, 2026), the feature combines usage data and employee surveys to show, team by team, what AI produces, what it costs, and where work stalls. For an SMB that has no such tool but still pays for AI licenses, this launch mostly reveals a method that can be copied at low cost. Here is what Claude Smart Reports actually does, the numbers behind it, and how to apply the lesson without waiting for Anthropic to build an equivalent offer for small businesses.
In brief
- Anthropic made three Claude Enterprise features generally available on October 28, 2026: Smart Reports, in-app surveys, and pooled group token budgets (source: TechCrunch, Channel Insider).
- Smart Reports groups a team's activity into workstreams and shows, for each one, the associated cost, completed deliverables, and recurring friction points.
- The tool relies on internal usage data and employee surveys, then suggests improvements, such as turning a frequently repeated task into a reusable "skill."
- Anthropic explicitly states these reports should not be used to evaluate individual employees.
- A Jellyfish report covering 276,000 customers shows Claude tools already account for nearly a third of proposed code changes submitted for review, for about 2% of total engineering cost.
- Anthropic reports a 900% net revenue retention rate over one year among existing customers: a dependent SMB can hardly ignore that level of lock-in.
What Claude Smart Reports actually does
A Smart Report is a document generated by Claude that summarizes a team's activity over a chosen period. In practice, a manager selects a team, a time window, and the relevant Claude products (Claude Code, Claude in Chrome, the API, etc.). The report then groups observed work into workstreams: for example, "billing module migration" or "drafting customer replies." For each workstream, the tool displays the token cost, completed tasks, recurring friction (code sent back in review, abandoned task, missing context), and a concrete recommendation.
Collection
Survey
Grouping
Recommendation
Anthropic adds an important guardrail, cited by Channel Insider (October 30, 2026): Smart Reports are meant to understand adoption and plan investment, not to score individuals. The company explicitly advises against using them for performance reviews or employment decisions. Customer data from these features is not used by default to train Anthropic's models.
Key takeaway
The point isn't the tool itself, which is reserved for Claude Enterprise customers, but the reading grid: cost, deliverable, friction, recommendation. An SMB can replicate this grid by hand, with a spreadsheet and a monthly team check-in.
Why Anthropic is releasing this now
This launch answers a very concrete concern among large accounts: knowing whether money spent on AI produces a real return. TechCrunch cites a report from the firm Jellyfish, based on data from 276,000 customers, which gives a precise measure of Anthropic tool adoption in software engineering.
Nicholas Arcolano, Jellyfish's chief scientist, quoted by TechCrunch, sums up the situation: this spending growth was not anticipated by any CFO, and companies now need to learn to optimize their usage rather than simply spend more. That is exactly the problem Smart Reports tries to solve for large customers. For an SMB, it's an early warning: the AI bill is growing faster than visibility into what it produces, and it is better to set up a tracking method before the gap becomes unmanageable.
What an SMB can take from this without the tool
Claude Smart Reports is only available to Claude Enterprise customers, an offer designed for large organizations. But the logic behind the tool remains accessible to any SMB, with simpler means.
Without a tracking method
AI spending (subscriptions, API) is visible in the accounting, but nobody knows precisely which tasks it actually sped up. Renewal decisions are made "by gut feeling."
With a simple grid
Each month, a manager notes three workstreams where AI was used, the estimated time saved, and one friction point. The decision to keep, extend, or cut a usage rests on facts.
In practice, an SMB can replicate the spirit of Smart Reports with a shared spreadsheet: a "workstream" column (e.g. customer support, quote drafting, contract review), an "estimated cost" column (licenses plus supervision time), a "deliverable produced" column, and a "blocker encountered" column. A 20-minute monthly check-in per team is enough to fill this grid and spot usages worth generalizing, training on, or dropping.
| Criteria | Claude Smart Reports (Enterprise) | DIY grid for SMBs |
|---|---|---|
| Data source | Real Claude usage + employee surveys | Self-reported, monthly team check-in |
| Granularity | Per workstream, automatic | Per workstream, manual |
| Cost | Included in the Claude Enterprise plan | Free (only summary time) |
| Recommendation | Generated by AI | Decided collectively in a meeting |
| Risk of misuse | Explicitly discouraged for individual evaluation | Must be framed upfront by management |
Limits, and the real underlying issue: dependence
This announcement comes as Anthropic reports 900% net revenue retention and revenue driven largely by Claude Code, while preparing for an IPO. Per Jellyfish, companies that have standardized their workflows around Claude have no strong reason to switch providers, and Smart Reports further reinforce this lock-in by making the produced value, and therefore the perceived cost of switching, visible. It is a lighter version of the hands-on support that major AI labs usually offer their biggest customers.
For an SMB, the lesson echoes a point already documented in our piece on AI vendor dependence: the more central a tool becomes in your processes, the more costly it is to replace. Setting up usage and cost tracking, even a homemade one, is exactly what keeps that decision under your control instead of discovering it too late.
FAQ
Is Claude Smart Reports available for SMBs?
No, at this stage the feature is reserved for Claude Enterprise customers, an offer designed for large organizations. An SMB can still replicate the same tracking logic with simple tools (spreadsheet, monthly team check-in).
Is Smart Reports used to evaluate employees?
No. Anthropic explicitly states the tool is meant to understand AI adoption and guide investment decisions, not to judge individual employee performance, and advises against that use.
Why is Anthropic launching this tool now?
Because large enterprise customers are trying to justify sharply rising AI spending: per Jellyfish, that spending has grown tenfold since early 2026, a pace no CFO had planned for.
Is data entered into Smart Reports used to train Claude models?
According to TechCrunch and Channel Insider, Anthropic states that customer data tied to these features is not used by default to train its models.
Conclusion
Claude Smart Reports illustrates a broader trend: after two years of rapid adoption, companies now want proof, not promises, of what AI brings them. An SMB doesn't need to wait for a size-appropriate version to apply the same discipline: define workstreams, track a cost, spot frictions, decide on facts. That tracking, more often than the tool itself, is usually what separates a dormant AI license from an investment that pays off. For more on practical AI management for SMBs, check our Mag resources or our customer success stories.
