--- title: Cowork reporting canonical: "https://cloudmonitor.ai/docs/using-cloudmonitor/reports/cowork/" description: The current availability of Cowork usage and cost reporting in CloudMonitor. --- :::note[Coming soon] This screen isn't finished yet. If you open it you'll see a **Coming soon** card. This page describes what it will do, so you know whether to wait for it. ::: :::caution[The figures on this screen are examples, not your data] Choosing **Preview anyway** on that card reveals the full report running on a fixed example month. Those numbers describe nobody's real tenant, including yours, and the page says so above every panel. Don't quote, export or budget from them. ::: :::note[Not everyone has this] **Microsoft 365** is an optional module. If you can't see it in the menu, it isn't switched on for your workspace: ask an admin to turn it on under Settings ▸ Features. ::: **Cowork** sits under the **Microsoft 365** menu section, alongside Usage, Security, Licensing and Copilot. ## What you see today CloudMonitor does not ingest Cowork usage or cost data, so there is nothing real to show yet. Rather than an empty page, the screen offers a worked example of the finished report. Opening it puts a **Coming soon** card over the page. Choosing **Preview anyway** reveals the report, with a standing notice above every panel saying the figures are examples and not your tenant's. That notice cannot be dismissed, and it stays there for as long as the figures are examples. The example runs on a fixed month that reconciles with itself: the credits match the costs, the task splits match the totals, the cost groups add up to the period, so the magnitudes are realistic. None of it is a measurement. It exists so the shape of the report, and the questions it answers, can be agreed before the pipeline is built. No delivery date is promised. The global filter bar is hidden on this screen. Period, Provider, Cost group, Virtual Tags and Cost basis would each move nothing while the report runs on a fixed sample period, and a filter that changes nothing is worse than no filter at all. ## What the report will show ### Adoption Whether anyone is using Cowork, and whether that is holding. Daily active users, the split between tasks people start themselves and tasks Cowork runs on a schedule, retention: people active in the last seven days *and* in the seven before them, and the day-of-week pattern. ### Cost & credits What it costs, and how long the prepaid credits last. The credit burndown with a projected exhaustion date, cost per task, per conversation, per scheduled run and per active user, weekly spend, and the charge lines underneath. ### People Who is driving the spend. Each person's task counts, active days and last activity, joined to their credit spend and cost group, alongside the people who set up a schedule and then stopped using Cowork. ### Allocation Whose budget it came out of. Cowork spend by cost group, the per-person and per-task rate for each, and an unallocated bucket for people who belong to no cost group. ### Access & policy Who is waiting for access and what the limits are. Pending requests with an estimated cost, the billing model and purchased capacity, hard cap and per-user limits, and credit consumption by plugin. ## Where the figures come from Two Microsoft surfaces, and neither one serves the report alone. The **Cowork usage report** supplies the people and the tasks: active users, total tasks, retained users, the daily series, the user-initiated and scheduled split, and the per-person detail. It carries no money at all. The **Copilot Credits** cost-management surface supplies the credits, the prepaid balance and the spending policy. It breaks consumption down by user, group, service or agent. Anything about money per task is CloudMonitor putting those together: credit charges for the period divided by task counts for the period. That makes every per-task figure a **period average** rather than a measurement of one task, because Microsoft never attributes credits to a single task, and the screen says so wherever it shows one. Cost-group allocation is a second join, onto CloudMonitor's own cost groups, which Microsoft has no equivalent for. Some things no source provides, and which the report will not claim: - Whether anyone read a scheduled run's output. The abandoned-schedule panel infers this from the absence of any conversation, which is weaker evidence than reading it directly. - The credits consumed by one individual task. - Anything before 1 April 2026, which is as far back as per-person Cowork detail exists.