Filters isolate spend, usage and variance without changing the evidence underneath.
License chargeback
Investigate why cost changed, which teams are efficient, and where spend is misaligned with usage.
These four close: cost ÷ users = cost per user, over the same 1 department. The estate's whole bill is $126,000 — 3,990 seats wider, on the Overview.
1 filter active Showing: every month Aristo holds (6) Supply Chain All productsChange
1 filter active Showing: every month Aristo holds (6) Supply Chain All products
How the floor holds when the data is cut
There is no per-person view, screen or export for any leader to pull. 1 team is withheld from the counted-utilisation columns for the same reason — they appear in the department table with an em dash rather than a figure. Filtering is how a k-anonymous surface usually leaks: each cut is safe and the intersection of three of them is one person. So the department list offers only the already-merged rows — a small team is not an option here or anywhere else, and selecting the merged bucket shows the aggregate, never a split of it. Where a department and a product are combined, a cell below five reports “fewer than 5”: the count is withheld, the fact is not.
Why this workspace keeps none
This is the illustrative organisation — it is never written to, so there is nothing here to save. In your own tenant, naming a filter set keeps it for everyone who can open this workspace, and answers it with the numbers as they stand each time somebody opens it.
Where the money goes
Consumption month by month, split across the products it was spent on.
Evidence & data — Cost trend by product
Comparison
Microsoft 365 Copilot is the largest band in August 2026 at $126,000, up 17% across the period. Every column adds to the same month on the consumption trend above.
Source & method
Aristo consumption envelope split across your licence records.
Basis: Labelled estimate
Accessible data
| Month | Microsoft 365 Copilot | Microsoft 365 E5 | Power BI Pro | Dynamics 365 | Other paid products | Total |
|---|---|---|---|---|---|---|
| March 2026 | $108,000 | $72,000 | $26,000 | $18,000 | $5,400 | $229,400 |
| April 2026 | $112,500 | $74,000 | $27,000 | $19,000 | $6,150 | $238,650 |
| May 2026 | $117,600 | $76,000 | $28,000 | $19,800 | $5,380 | $246,780 |
| June 2026 | $121,800 | $79,000 | $29,200 | $21,000 | $7,420 | $258,420 |
| July 2026 | $124,500 | $82,000 | $30,000 | $22,200 | $11,060 | $269,760 |
| August 2026 (so far) | $126,000 | $84,000 | $31,000 | $23,000 | $14,540 | $278,540 |
Every figure is a labelled envelope: this product holds a real per-seat price for Copilot alone, so a per-product dollar split is an estimate over the estate and is drawn hatched.
Product breakdown
Every paid product, sized by the seats it holds.
Evidence & data — Paid seats by product
Comparison
Every paid Microsoft product in this estate, sized by the seats bought rather than by what they cost. Money appears in the legend only beside a product this estate holds a real seat price for, which is Copilot. The ring itself is drawn in seats so that every product is in it.
Source & method
Microsoft licence records (Organization.Read.All).
Basis: Counted
Accessible data
| Product | Paid seats | Share | Spend / mo |
|---|---|---|---|
| Microsoft 365 Copilot | 4,200 | 48.0% | $126,000 |
| Microsoft 365 E5 | 2,500 | 28.6% | no seat price held |
| Power BI Pro | 1,050 | 12.0% | no seat price held |
| Dyn365 | 600 | 6.9% | no seat price held |
| Visio Plan 2 | 242 | 2.8% | no seat price held |
| Project Plan 3 | 150 | 1.7% | no seat price held |
The shares are of 8,742 counted paid seats across 6 products, which is the figure in the middle of the ring. A product with no seat price is counted in seats and deliberately never dollarised.
Department details
Every team this page can name, with the step it took since last month.
Scroll sideways for the remaining columns.
| Department | Users | Total cost | Cost / user | Change | Trend |
|---|---|---|---|---|---|
| Supply Chain | 445 | $6,300 | $14 | ↓ 0.9% |
Explore detailed analysisTrends, cost per user, usage alignment, efficiency, products and variance
Spend trend
Metered consumption month by month across March 2026 – August 2026, each month carrying its own basis.
Metered consumption rose 18% across 5 completed months, from $229,400 in March 2026 to $269,760 in July 2026, and 0 of the 6 months shown are measured rather than estimated.
Evidence & data — Metered consumption by month
Comparison
0 of 6 months in this period are measured. A month is solid only once an export covering it has landed — the hatched bars are the honest picture of when measurement began, not a gap in the data.
Source & method
Azure Cost Management export where connected; Aristo consumption envelope otherwise.
Basis: Labelled estimate
Accessible data
Scroll sideways for the remaining columns.
| Month | Metered | Basis |
|---|---|---|
| March 2026 | $229,400 | estimate |
| April 2026 | $238,650 | estimate |
| May 2026 | $246,780 | estimate |
| June 2026 | $258,420 | estimate |
| July 2026 | $269,760 | estimate |
| August 2026 (so far) | $278,540 | estimate |
How this is counted — 2 notes on scope, denominators and what this section cannot answer
This section ignores the department filter. Metered consumption is billed to the tenant, not to a team — there is no per-department series to draw, and apportioning the estate's dollars by headcount would produce a chart that moves when you change the filter and means nothing. The estate series is shown unchanged.
The current month is drawn open because it is a month-to-date figure standing beside completed months. It is excluded from the movement asserted above — comparing a part-month to a whole one is the single easiest way to read a fall in spending that has not happened.
Cost per user
Counted Copilot licence per head, by team, across Supply Chain.
Supply Chain carries $14 of Copilot licence per head — 210 seats across 445 people — and it is the only row here because the department filter is narrowing this page to it, so there is no distribution to compare it against.
Evidence & data — Counted Copilot licence per head, by team
Comparison
A team is expensive per head when more of its people hold a seat, not because its seats cost more — the seat price is the same everywhere in this estate. This ranking is therefore a map of seat coverage in money, and it is a different question from whether those seats are being used.
Source & method
Microsoft licence records joined to your directory.
Basis: Counted
Accessible data
| Department | People | Seats | Coverage | Licence / mo | Per head | Per active seat |
|---|---|---|---|---|---|---|
| Supply Chain | 445 | 210 | 47% | $6,300 | $14 | $53 |
Every row is a team of five people or more. Coverage is seats divided by headcount; per head divides the same bill by the same headcount.
How this is counted — 2 notes on scope, denominators and what this section cannot answer
This is a TEAM total divided by a TEAM headcount, at five people or more. It is not a per-person figure, there is no per-person figure in this product, and no filter combination on this page produces one.
The consumption allocation is deliberately not divided by headcount here. That apportionment multiplies a flat per-user rate by a department size, so dividing it back out returns the same rate for every team — a column that looks like a finding and can never contain one.
Usage vs cost
Seat use against licence cost, one mark per team, sized by counted seats.
Supply Chain uses 57% of its 210 counted seats against $6,300 a month of licence — the only mark here because the department filter is narrowing this page to it, so there is no estate around it to be an outlier of.
Evidence & data — Seat use against licence cost, by team
Comparison
The medians are not drawn: with fewer than three teams a median is the midpoint of a pair, and dividing the estate on it would be a shape, not a finding.
Source & method
Microsoft licence records + Copilot usage signals, joined per team.
Basis: Counted
Accessible data
Scroll sideways for the remaining columns.
| Department | Seats | Seat use | Licence / mo | Position |
|---|---|---|---|---|
| Supply Chain | 210 | 57% | $6,300 | Aligned |
How this is counted — 3 notes on scope, denominators and what this section cannot answer
Every mark is a team of five people or more. The bubble is drawn as an AREA in counted seats, so a team with four times the seats gets four times the ink — scaling the radius instead would quadruple the visual weight of a doubled quantity.
The usage axis runs the full nought to a hundred whatever the data does. Auto-scaling a percentage axis to its own spread makes a five-point difference between teams read as a chasm.
The quadrant guides are not drawn. A median across fewer than three teams is the midpoint of a pair, and splitting two teams into an efficient one and a wasteful one on that basis is not evidence.
Department efficiency
Every team ranked on the share of its counted seats that are working, across Supply Chain.
Supply Chain is the only ranked team because the department filter is narrowing this page to it: 57% of its 210 seats are active, leaving $2,700 a month on 90 seats nobody has touched.
Scroll sideways for the remaining columns.
| # | Department | People | Seats | Active | Seat use | Licence / mo counted |
Per active seat counted |
Allocation consumption · estimate |
Recoverable |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Supply Chain | 445 | 210 | 120 | 57% | $6,300 | $53 | $11,380 | $2,700 |
How this is counted — 2 notes on scope, denominators and what this section cannot answer
Ranked on the share of a team's counted seats that are genuinely active — the one column here with no estimate and no price inside it. The two money columns are on different bases and are never added: licence cost is counted seats at your seat price, the allocation is an apportionment of metered funding.
There is no “change since last month” column and no per-team trend line, and that is a missing fact rather than a missing feature: Aristo stores the estate month by month, but every per-department record it keeps — the utilisation join, the directory headcount, the seat grid — is a single current snapshot overwritten on each refresh. So these teams can be ranked against each other today and cannot yet be ranked against themselves last month. A per-department monthly record is the one upstream change that fills both columns.
Product utilisation
Every paid Microsoft product, in counted seats — bought, assigned, and never handed to anyone.
Microsoft 365 Copilot carries the largest unassigned block: 160 of its 4,200 paid seats have never been handed to anyone, which is $4,800 a month at your seat price.
Scroll sideways for the remaining columns.
| Product | Licensed | Assigned | Never assigned | Assigned share | Active | In grace | Supply Chain seats | Spend / mo | Recoverable |
|---|---|---|---|---|---|---|---|---|---|
| Microsoft 365 Copilot | 4,200 | 4,040 | 160 | 96% | 2,760 | — | 210 | $126,000 | $4,800 |
| Microsoft 365 E5 No seat price held — counted in seats only |
2,500 | 2,440 | 60 | 98% | — | — | — | — | — |
| Power BI Pro No seat price held — counted in seats only |
1,050 | 900 | 150 | 86% | — | — | — | — | — |
| Dyn365 No seat price held — counted in seats only |
600 | 520 | 80 | 87% | — | — | — | — | — |
| Visio Plan 2 No seat price held — counted in seats only |
242 | 120 | 122 | 50% | — | — | — | — | — |
| Project Plan 3 No seat price held — counted in seats only |
150 | 90 | 60 | 60% | — | — | — | — | — |
How this is counted — 3 notes on scope, denominators and what this section cannot answer
Seats are counted for every paid product. Money appears only where this product holds a real price for the product — Copilot today — because an invented price makes a total look precise and be wrong.
5 of the 6 products shown are reported in counted seats and deliberately never dollarised. Their headroom is real; its value is not yet knowable.
The "Supply Chain" column is the counted department × product seat grid, folded at the same five-person floor as everything else: a cell below five reports "fewer than 5" — the count is withheld, the fact is not.
Variance analysis
What moved across March 2026 – August 2026, against the prior month, the prior run, your cap and the labelled projection.
The largest movement across March 2026 – August 2026 is that metered consumption went from $246,780 in May 2026 to $258,420 in June 2026, up $11,640 (4.7%). Alongside it, cold seats fell from 1,283 at the previous run on 2026-07-30 to 1,280 at this one, down 3 seats — $90 a month of licence.
Scroll sideways for the remaining columns.
| What moved | From | To | Change | Basis | Why, where the data says so |
|---|---|---|---|---|---|
| Metered consumption | March 2026 $229,400 |
April 2026 $238,650 |
+$9,250 ↑ 4.0% |
estimate | — |
| Metered consumption | April 2026 $238,650 |
May 2026 $246,780 |
+$8,130 ↑ 3.4% |
estimate | — |
| Metered consumption | May 2026 $246,780 |
June 2026 $258,420 |
+$11,640 ↑ 4.7% |
estimate | — |
| Metered consumption | June 2026 $258,420 |
July 2026 $269,760 |
+$11,340 ↑ 4.4% |
estimate | — |
| Cold seats | previous run, 2026-07-30 1,283 seats |
this run 1,280 seats |
−3 seats ↓ 0.2% |
counted | Cold seats fell by 3 while active seats rose by 3. That is the shape of activation rather than of reclamation — though the previous run did not record the paid-seat total, so the two cannot be reconciled to the seat from here. |
| Active seats | previous run, 2026-07-30 2,757 seats |
this run 2,760 seats |
+3 seats ↑ 0.1% |
counted | — |
| Chargeback-ready cohort | previous run, 2026-07-30 1,118 people |
this run 1,120 people |
+2 people ↑ 0.2% |
estimate | — |
| Next month, projected | July 2026 $269,760 |
projection $289,027 |
+$19,267 ↑ 7.1% |
projection | A projection is what the trailing months imply if nothing changes. It is not a forecast anybody has committed to and it is never counted as spend. |
How this is counted — 4 notes on scope, denominators and what this section cannot answer
A movement between two months on DIFFERENT bases is not purely a movement in spending — one endpoint counts Aristo-mediated consumption and the other counts everything your billing export sees. Every row states the basis of its weaker endpoint, and says so in words where they differ.
Seat and cohort movements are measured between RUNS, not between calendar months, and are labelled that way. A run is when Aristo counted; the calendar is not what moved.
Estimated months are included in the comparison and marked, rather than dropped. Dropping them would leave a tidy series with a hole in it and no sign that the hole was ever there.
This section ignores the department filter for the money rows: metered consumption is billed to the tenant, not to a team. The seat rows are estate-wide counts on both sides for the same reason — the previous run stored an estate total, not a per-team one.
How every number on this page was arrived at, which filters are not offered, and what is not measured yet
Every department figure on this page is the one the Overview states. This page filters, ranks and compares them; it re-derives nothing. That is deliberate: a leader who reads one number on the summary and a different number one click deeper has stopped trusting the product, and no explanation about denominators gets that back.
Counted means read directly from your Microsoft licence records or usage signals. Measured means taken from your own billing export for the month shown. Estimate means an envelope Aristo computed and labelled as one. Assumption means a default this product supplied because you have not set the real value. Projection means a labelled forward figure and is never counted as spend. Every figure carries the weakest of the words that apply to it.
Filters this page does not offer, and why.
- Region and country. Aristo counts seats by region under the same five-person floor, but no money is attributable to a region — the department split is the only k-safe join this product has to dollars. A region filter would change the page and change nothing in it.
- Cost centre, manager and business unit. Aristo reads department from your directory. It does not read cost centres or manager chains, and a manager filter is a per-person view with a different name on it.
- Active / inactive people. That is a per-person cut. Teams are shown at five people or more and there is no per-person view in this product for a filter to reach.
- Licence type. This is the product filter above. Aristo names the SKU as your licence records name it rather than inventing a second taxonomy on top.
What is not measured yet.
- Anomaly alerts stay withheld until enough measured history exists to establish a defensible baseline and threshold; ordinary variation is never promoted into a false incident.
- There is no benchmark to compare this organisation against on this page. Cross-organisation outcome evidence exists in this product only where enough tenants have run the same move, and a spend benchmark is not one of the things it evidences.
- Not every month in this period is measured. Connect a read-only Azure Cost Management export and each estimated month becomes measured, to the cent, from the month it lands onward.
- Every chart opens its exact numbers under "View the data behind this chart". The page-level Export produces the License Report with the same counted seat and licence figures in a finance-ready record.
- A per-team cost TREND is not available. Aristo persists the estate's months and each run's totals, but it does not yet keep a per-department history, so this page can rank teams against each other today and cannot yet rank a team against itself last month.