Skip to content

What isn't available yet

Most of CloudMonitor reads your own data. Some of it doesn’t yet, and this page says which — because finding out by accident, after quoting a figure to someone, is a bad way to learn it.

Everything here is a gap we intend to close. Nothing here is hidden in the product: each screen says what it’s showing. This page is the list in one place.

Unit Economics shows a Coming soon card. Cost per customer or per request needs business numbers CloudMonitor doesn’t have a way to receive yet. The screen exists so you can see where it will live.

Cost Allocation’s partition and reconciliation panels. Your rules are real and saved. The totals shown beside them — how the whole bill divides across groups — are a worked illustration. For your actual split, use The Cost Groups report and Unallocated Costs, which read your data.

Billing. Plan, seats and cost on the Billing tab are placeholders pending the marketplace integration. Treat your contract as the source of truth, not this screen.

Per-person AI spend. Budgets can be scoped per person and the envelopes are real, but the spend against them is an example allocation. Per-user figures need per-seat feeds from the AI providers that aren’t connected yet.

Some AI token metrics. GitHub Copilot usage is live. Token counts, cache-hit rates and blended cost per million tokens for other providers are examples until those usage APIs are connected.

Some Fabric capacity telemetry. Capacity cost is your real spend. Parts of the fine-grained consumption detail need a feed that isn’t in place.

Budget alerts are not delivered automatically.

Forecasts are a simple run rate. Spend so far divided by days elapsed, multiplied by days in the period. That’s honest and useful for “are we heading over?” It doesn’t model seasonality, a planned migration or anything you know and it doesn’t.

Anomaly root causes are derived. The root cause, next step and severity on an anomaly are worked out from the cost movement itself, not from an attribution feed. They’re a good starting point, occasionally a wrong one, and the owner field is often blank.

WAF Performance Efficiency has no rules behind it. An empty pillar there means “not assessed”, not “all clear”. The other four pillars are live.

Microsoft 365 password expiry and license spend have no backing data and aren’t expected to. The other eight tabs are live.

Some resource details are sparse. Created date, created by and owner on Search Resources show a dash where the underlying data has no value. That’s genuine absence rather than a loading state.

Virtual tag filtering has limits. Tag values are still being wired into the live path, so applying a virtual-tag filter can leave some reports empty rather than filtered. If a report empties when you apply one, that’s why — clear it and the data returns.

Two signals, always:

  • The ⓘ beside a page or panel title says what that number is and where it comes from, including when it’s illustrative.
  • The Demo mode badge in the top bar means the whole app is on sample data. No badge means you’re looking at your own data, an honest empty state, or an error — never sample dressed up as real.

There is no per-panel “sample” label. If something can’t load, CloudMonitor shows an error at the top of the screen rather than quietly substituting made-up numbers.

A cost tool’s only real asset is whether you believe it. Publishing the gaps costs less than having someone discover one in a board meeting.

If a figure matters and this page hasn’t settled it, ask support before you use it.