nowRolling out
Viva Glint - Copilot admin configuration assistance
Microsoft public roadmap
Review source
Aristo's read Grounded in page evidence
The Frontier report connects the executive position, next moves and supporting proof.
Prepared exclusively for Contoso
Aristo · Contoso · Live view
The Frontier Firm Report
Are you a Frontier Firm? You are Advancing (55–84 of 100).
Frontier Firm: Microsoft’s name for an organization rebuilt around AI — not just using it, but delegating real work to it.
What is a Frontier Firm?
Microsoft’s definition · the three pillars · the four bands
A Frontier Firm (Microsoft, 2025 Work Trend Index) is an organization rebuilt around AI — intelligence on tap, human-agent teams — not just using AI but delegating real work to it.
Aristo does not re-define the term — it measures it, for your tenant, from your own counted signals. Here is the arithmetic, in full.
Fig · The three pillars and their fixed weights. A pillar that is not available is dropped and the remaining weights re-normalized — never silently scored zero.
Adoption · 50 pts
How far AI has actually spread — your readiness score: real activation, real engagement, real outcomes.
Activation · 25 pts
How much of the assigned Copilot estate is genuinely working — one minus the share of assigned seats sitting cold.
Security · 25 pts
Your Microsoft Secure Score — counted only once you have consented to share it, never assumed.
Fig · The four bands, at their true widths on the 0–100 scale. Versioned with every run, so a month-over-month comparison always states the scale it was scored on.
Emerging
0–29 of 100
The licences are in place; everyday AI use is just beginning.
Developing
30–54 of 100
Adoption is climbing, but genuine activity is still the minority.
Advancing · you
55–84 of 100
Counted adoption taking hold, proven lift under way.
Frontier
85–100 of 100
AI is how the everyday work runs.
Counted, never self-reported. Every pillar above is read from your tenant’s own signals. Where a number can only be estimated, this report says so on the number itself.
Aristo
I scored you 78 — and found a $38,400/mo activation opportunity in cold assigned seats.
78 / 100
Frontier score
▲ 4 · Run 3 → 4
68%
Seats active
2,760 of 4,040 assigned seats
$38,400
Activation opportunity
1,280 assigned-but-cold seats / month
180
Made
skills & agents, by your people
Next: Fund the ready cohort — capped (credits only for the proven-ready people)
Show everything ▾
Every capability is reachable now — nothing is locked. Jump to stand · money · usage · next move · proof · what’s coming · methodology. Or open Make / Ask Aristo below.
Share, delivery and source
decks · Teams · Copilot · full context
Take this to your meeting
BoardFinanceITSellerPrint
One deck per room, or the whole document as paper — the same counted numbers as this page, never re-typed. A spoken brief is one press away below.
Aristo comes to you
In Teams. When Teams delivery is configured, sponsor cards use this same counted summary.
In Copilot. Aristo Spark is a Microsoft 365 Copilot agent — ask it in Copilot Chat or Word.
In your deck. Every report leaves as a PowerPoint for the room you are walking into.
You’re in a solid but uneven place. Activation is 68%, with 2,760 active users, but 1,280 people are still cold — and that same 1,280 are now at risk of staying there.
94
Days to renewal
Live view composed 2026-08-29 · Last counted Run #4 2026-07-30 · assessment overdue 3 days
Question 1 · Where do we stand?
As of Jul 30, 20264 points up across completed runsReaches Frontier in ~53 days
Last counted Run #4 2026-07-30
Current view: 78, Advancing (55–84 of 100). Completed-run record: 4 points closer across completed runs — Run #3 at 70, Run #4 at 74.
Fig · The Frontier Scale. Four bands, fixed at Emerging 0–29 · Developing 30–54 · Advancing 55–84 · Frontier 85–100, versioned with every run so a month-over-month comparison always states the scale it was scored on. Your position is counted, never self-reported. The dashed flag is your organization’s own approved target.
Our AI strategy
Approved by your organization
You are 78, heading for 85 by January 2027 — at your current rate you arrive October 2026, on track.
Every team ships one AI-run routine this year.
We fund from recovered licence waste, not new budget.
Nobody is measured on AI use — only on what it returns.
Activation %the seats you already pay forAdoptionthe heaviest pillar — half the scoreSecurity readinessthe pillar a CISO is asked about
Approved by your organization · version 1. Every run re-reads your position against this, and the Frontier Scale above carries your target. Change it any time — a new version is written, and the old one stays, so a past run always states the strategy it was measured against.
Blind
Counted
Steered
Proven
Compounding
| The move | Worth | Owner | Lands | Running total |
|---|---|---|---|---|
| Raise readiness — real engagement and counted outcomes, not sign-ins worth 11.0 in full — more than the distance left to your target |
+7.0 | Adoption lead | Oct 21, 2026 | 85 |
| Reaches your target | +7.0 | 85 |
Fig · The path to your own approved target of 85. The remaining points are decomposed, not apportioned: each step is worth exactly what closing that pillar is worth on the weighted score, and the four steps sum to the distance. Landing dates are projected from your own fitted line across your last counted runs — re-drawn as each run lands, never asserted.
How we counted this
what the band means · the trend · the peer range
Advancing means: counted adoption is taking hold, proven lift is under way and spend is steered — but the everyday isn’t AI-run yet.
📈 on pace to cross into Frontier next quarter — a projection from your own last runs; the fitted line reaches 85 in about 53 days. Re-drawn as each run lands, never asserted.
So what: when the board asks “what’s our AI strategy?”, this page is the answer — position, trend, and the distance that remains, counted, not felt.
Defined: Frontier Score — 0–100, computed monthly from your tenant’s counted signals (seats, genuine activity, proven lift, spend steering); methodology in §7. Band — the four ranges (Emerging 0–29 · Developing 30–54 · Advancing 55–84 · Frontier 85–100), versioned per run so they stay comparable month to month. The path — the points still available, decomposed across the three pillars at their own weights; the steps sum to the distance by construction, never by judgement.
See how the score is built
the three pillars · your own history
Activation 68%2,760 active1,280 cold
1 move runningActivate the cold Copilot cohort
0 unseated people engaging
180 agents live
Question 2 · Are we wasting money?
As of Jul 30, 202690 dollars idle down across completed runs
Last counted Run #4 2026-07-30
$38,400 a month is the activation opportunity in 1,280 assigned-but-cold seats, counted from usage. Total recoverable exposure is $43,200 a month: it also includes 160 bought-but-unassigned seats.
Fig · Every assigned Copilot seat, on one scale. The amber bar is money you are already spending on seats nobody used this period — counted from Microsoft's own usage report, at your own $30 seat cost. Recovery here means reroute to people waiting, never removal.
Question 2 · Are we wasting money?
$38,400 / month activation opportunity
1,280 assigned Copilot seats are quiet this period — counted from Microsoft's own usage report, and reroutable to people already asking. Total recoverable exposure is $43,200 a month after 160 bought-but-unassigned seats are included.
Aristo · Contosoaristo.phoenixhalo.com/example
How we counted this
how idle is counted · what recovery means here
So what: pay for your AI strategy with the savings from your unused licences — this page funds the work. Recovery here means reroute to people waiting — never removal, never a shame list.
Defined: genuinely active — used Copilot within the last 28 days per Microsoft’s own usage report, not merely signed in. assigned-but-cold — assigned a paid seat, no genuine activity this period; this is the activation opportunity. total recoverable exposure — assigned-but-cold plus priced bought-but-unassigned seats. credits — metered agent capacity (Cowork), billed per use.
See where the money is, seat by seat
idle seats · credits · the heat map
The whole AI estate — counted, never estimated
$38,400
1,280 cold seats
Idle / mo
1,120
$33,600/mo cap
Credit-ready
160
$4,800/mo unassigned
Idle licenses
$246
re-forge cost avoided (envelope)
Never re-burned
Where the credits go — and what they make
Earning its credits
$320,000
envelope estimate — measured when the org connects the Cost Management export
Steer
—
no credits steered yet this month
Idle burn
$38,400
1,280 cold seats — counted · +$4,800 idle licenses
Your credits are earning their keep — $320,000/mo in planned productive Cowork.
■ solid = measured from your Cost Management export · ▩ hatched = envelope estimate (measured when the org connects the export)
Named threshold (dashed): your funded Cowork cap · $320,000/mo.
Fixed bands, never re-normalized: Earning its credits = productive Cowork spend · Steer = credits steered to the included free seat · Idle burn = cold Copilot seats. Comparable across every view.
Where the credits went — over time
$320,000
Aug '26
estimate
This month's Cowork is a $320,000 envelope estimate — it becomes measured the moment the org connects its Cost Management export.
◆ What the credits built, counted to date: 8 skills forged · 180 agents built.
▩ hatched = envelope estimate (measured when the org connects the export)
Named threshold (dashed): your funded Cowork cap · $320,000/mo.
One estimated month so far — this is the honest single-month state, not a projected trend. The line fills in as each month is measured.
Consumption cost · three planes
Copilot seats
$82,800
2,760 active seats × $30/seat — counted · assumed $30 seat price (set yours for exact ROI)
Agents & credits
$320,000
envelope estimate — measured when the org connects the Cost Management export
Azure AI · Foundry
—
one read-only Reader consent away — never estimated
Two consumption lines to grow — and the one read-only Reader consent that lights the third. The Azure axis is dark by design, never estimated.
License truth: 4,200 bought · 4,040 assigned · 2,760 genuinely active
698 Microsoft 365 Copilot seats rescue-resistant 45+ days · 1 raised hand▷ Reroute
The whole licence story as its own document — every paid Microsoft subscription, what sits idle and where, who should have those seats instead, and whether that pays for the AI strategy.The License Report →
Credit ladder: Expand — holdout proves +14% causal lift on 79 proven seats; the budget is earning its keep — expand to the next 1,120 ready.
Funded Cowork budget $320,000/mo cap · next cohort ~$33,600/mo — envelope estimate; measured billing arrives when the org connects the Cost Management export
Cowork spend at 87% of your $320,000 safe budget — envelope estimate — may under-count
Copilot seat utilization — Seats Seats org-wide: 65.7% of 4,200 paid Copilot seats are genuinely active against your 75% floor — 1,440 paid seats nobody is using (160 never assigned to anyone) · counted
Finance pre-purchase pool — Pre-purchase pool Pre-purchase pool Finance team pacing 1.31× — about 20,815,120 of 21,500,000 credits drawn, projected 28,128,540 by Dec 1, 2026, so the pool runs dry before the term does · estimate run-rate
Sales pay-as-you-go — Pay-as-you-go Pay-as-you-go Sales team pacing 1.06× budget — $41,531 spent, projected $44,580 vs $42,000 by Aug 31 · estimate
Engineering credit packs — Credit packs Credit packs Engineering team pacing 0.78× allotment — 4,741,902 of 6,500,000 credits used, projected 5,090,000 by Aug 31 · estimate
MS seller upsell: not configured — name a Microsoft account contact and opt in (Settings) to send them the 90% expansion signal
Built once, reused across the org — credits we never spent twice. ~$246 protected across 41 reuses — envelope estimate; measured when the org connects the Cost Management export.
416 staff in the labelled illustrative Shadow-AI scenario — no Microsoft sign-in log was measured.
ChatGPT (268) · Claude (148) + 9 smaller apps withheld · aggregate sample, never a name
ChatGPT (268) · Claude (148) + 9 smaller apps withheld · aggregate sample, never a name
Question 3 · What is actually being used?
As of Jul 30, 20263 people active up across completed runs
Last counted Run #4 2026-07-30
2,760 people work with AI weekly — and 180 agents already live in your tenant.
Microsoft Teams · Aristo Spark
Real capture
Fig · Aristo at work, in the place your people already are. Not a mock-up and not a render — a capture from a live tenant.
1Somebody asks in their own words. No portal to learn, no syntax, no training.
2Aristo answers with a card rather than a link — the play itself, ready to run.
3The exact prompt, written for them and scoped to their own week. One tap to copy.
4And it says what it costs, up front: free, on the Copilot seat they already hold.
Every card is one job, one tap, one honest price — and every tap is counted, which is where the usage numbers on this page come from.
How we counted this
the working, the words, the sources
So what: this is your real AI estate — usage where it actually happens, and every agent (ours or not) belongs on one register. No more guessing what exists.
Defined: agent — a scoped AI assistant that does one job. declarative agent — the safe kind: instructions + knowledge, no code, runs inside your Copilot. register — the single list of every agent, whoever built it.
See the whole estate — every team, every agent
teams · skills · agents · reach
13 teams counted — champions lift the rest with their own proven plays
Engineering1,550 peopleready for credits · ~$50,900/mo
Sales1,370 peopleready for credits · ~$44,580/mo
Operations1,020 peopleready for credits · ~$30,380/mo
Customer Success885 peopleready for credits · ~$27,200/mo
Finance650 peopleready for credits · ~$23,460/mo
Marketing540 peoplemedium Cowork fit
IT495 peoplemedium Cowork fit
Product455 peopleready for credits · ~$16,010/mo
Human Resources370 peoplemedium Cowork fit
Manufacturing537 peoplemedium Cowork fit
Copilot seats by slice COUNTED · k≥5
Engineering680 seatsMicrosoft 365 Copilot: 680
Sales600 seatsMicrosoft 365 Copilot: 600
Operations420 seatsMicrosoft 365 Copilot: 420
Customer Success390 seatsMicrosoft 365 Copilot: 390
Finance340 seatsMicrosoft 365 Copilot: 340
Marketing260 seatsMicrosoft 365 Copilot: 260
Product260 seatsMicrosoft 365 Copilot: 260
IT250 seatsMicrosoft 365 Copilot: 250
Data & Analytics240 seatsMicrosoft 365 Copilot: 240
Manufacturing210 seatsMicrosoft 365 Copilot: 210
Supply Chain210 seatsMicrosoft 365 Copilot: 210
Human Resources180 seatsMicrosoft 365 Copilot: 180
Knowledge worker1,700 seatsMicrosoft 365 Copilot: 1,700
Engineer950 seatsMicrosoft 365 Copilot: 950
Manager850 seatsMicrosoft 365 Copilot: 850
Executive540 seatsMicrosoft 365 Copilot: 540
United States1,850 seatsMicrosoft 365 Copilot: 1,850
India1,000 seatsMicrosoft 365 Copilot: 1,000
United Kingdom650 seatsMicrosoft 365 Copilot: 650
Canada540 seatsMicrosoft 365 Copilot: 540
North America2,390 seatsMicrosoft 365 Copilot: 2,390
India delivery1,000 seatsMicrosoft 365 Copilot: 1,000
EMEA650 seatsMicrosoft 365 Copilot: 650
Most-reused recipe: Turn a messy sheet into a clear chartrun 4 times
Shelf skills reused across the org — built once, credits never re-burned41 reuses
180 agents live — every one owned by your people, forged from a sentence
8 skills on the shelf — reused 41 times, credits never re-burned
0 Business Dev Environments — stage your forged agents in a real ALM dev box →
Aristo — Frontier Firm (Copilot agent)ask the numbers in chat
Executive Score (personal agent)the exec’s own week, scored
🛰️ Fresh from the roadmap — just shipped in Copilot, ready to try
Viva Glint - Copilot admin configuration assistance — ▶ try in Copilot → · what’s new →
Unified Agent and App Installation Management Across Microsoft 365 and Teams Adm — ▶ try in Copilot → · what’s new →
Copilot Notebook in OneNote on Mac — ▶ try in Copilot → · what’s new →
Aristo answers your Frontier Firm — the score, the idle spend, and who is credit-ready — here in the authenticated Report, from counted signals, never estimates. The answer stays attached to the evidence, actions and report context leaders can verify.
Download the Copilot agentAnswers from Aristo’s illustrative scenario until your org connects · in Microsoft preview
Contoso Travel Desk · Contoso IT · not built by Aristousage: not yet exposed by Microsoft
Benefits Answers · Contoso People Team · not built by Aristousage: not yet exposed by Microsoft
Warehouse Stock Check · Contoso Operations · not built by Aristousage: not yet exposed by Microsoft
Every agent in your tenant, whoever built it — read from your Copilot catalog. Per-agent usage isn’t exposed by any Microsoft API yet, so no run counts are shown.
Question 4 · What should we do next?
As of Jul 30, 2026
Last counted Run #4 2026-07-30
One thing: Activate 1280 cold Copilot seats. Everything else can wait.
| The move | Worth | Owner | Lands | Running total |
|---|---|---|---|---|
| Raise readiness — real engagement and counted outcomes, not sign-ins worth 11.0 in full — more than the distance left to your target |
+7.0 | Adoption lead | Oct 21, 2026 | 85 |
| Reaches your target | +7.0 | 85 |
Decision details · immutable view path-6f355a0d9891daf84cf3c7a1Export this exact atomic board (CSV) →
Raise readiness — real engagement and counted outcomes, not sign-ins · +7.0 points · aligned to approved strategy
Current
Self-reported readiness 78 of 100
Pre-consent organization self-report; not measured from product evidence · as of 2026-08-29
Self-reported readiness 78 of 100
Pre-consent organization self-report; not measured from product evidence · as of 2026-08-29
Target
Counted AI readiness and outcome adoption · Readiness 100 of 100
Cumulative 85.0 · Frontier
Counted AI readiness and outcome adoption · Readiness 100 of 100
Cumulative 85.0 · Frontier
Effort / cost estimate
Programme scope and delivery hours are not yet estimated
Not estimated — approve scope before committing spend
Programme scope and delivery hours are not yet estimated
Not estimated — approve scope before committing spend
Dependency
Approved adoption owner, outcome definitions, and completed-run measurement
Approved adoption owner, outcome definitions, and completed-run measurement
Landing
2026-10-21
Projected from the organization’s own completed-run trend; recalculated when a counted run lands
2026-10-21
Projected from the organization’s own completed-run trend; recalculated when a counted run lands
Approval / version
Strategy approved by the organization on 2026-08-29; this move has no separate approval record
strategy-v1 · owner Adoption lead
Strategy approved by the organization on 2026-08-29; this move has no separate approval record
strategy-v1 · owner Adoption lead
Fig · The path to your own approved target of 85. The remaining points are decomposed, not apportioned: each step is worth exactly what closing that pillar is worth on the weighted score, and the four steps sum to the distance. Landing dates are projected from your own fitted line across your last counted runs — re-drawn as each run lands, never asserted.
How we counted this
why this move · how it was costed
So what: a strategy you can act on is one decision, pre-built, with the arithmetic shown. Approve it and the next issue shows the result — or says honestly that it didn’t move.
Defined: waiting list — people who asked for a seat (or whose demand is counted), ranked by readiness. provenance — “proven in N orgs” appears only when ≥2 real outcomes back the move; no evidence → no claim.
See every move, costed
what it is worth · why it is first
1 running · 0 hit target · 4 awaiting one tap
Activate the cold Copilot cohortAssigned-seat activation: 68% → 75%
Activate 1280 cold Copilot seatsone tap
Rescue 1280 at-risk users before they lapseone tap
Fund Cowork credits for the 1920-person ready cohortone tap
Introduce "Microsoft Viva: Viva Glint - Copilot admin configuration assistance" — the intro is already writtenone tap
Question 5 · How do we know it’s working?
As of Jul 30, 2026
Last counted Run #4 2026-07-30
Because we never nudge 12% of your cold seats — and the rest are 14 points ahead of them.
Fig · The never-nudged control group. Both arms were drawn on day 0 by a fixed rule and neither has been re-drawn since. The gap between them is the only part of this report that answers “would it have happened anyway?” — and billing uses the confidence interval's lower bound, never the gap itself.
How we counted this
the working, the words, the sources
So what: this is the number a CFO can defend — a built-in control group, measured for 90 days. If the line ever flattens, this page will say so: the score is allowed to go down.
Defined: control group / holdout — a deterministic 12% of cold seats, never nudged, set on day 0. proven — the gap at 95% statistical confidence; below that bar this report says “directional”, never “proven”.
See the two arms, in full
the control group · the confidence interval
Counted Run #3 → #4: score 70 → 74 across valid completed evidence
Every claim on this page survives a boardroom: counted signals, a control group, and a number that is allowed to go down.
This is the Day-60 proof every approved move points to — the promise you approve each move on: pay only for proven activation, measured here against the holdout. Not a refund to chase — a number that has to earn itself.
Projected activation PROJECTION · NOT MEASURED
~80% next month — this is an estimate, not a measurement: simple linear trend from the last 4 weekly readings. The solid line is your real weekly readings; the dashed line is where the trend points. It re-draws as each week lands.
Question 6 · What’s coming that we should care about?
Watch now: Viva Glint - Copilot admin configuration assistance
nextRolling out
Unified Agent and App Installation Management Across Microsoft 365 and Teams Admin Center
Microsoft public roadmap
Review source
laterNo sourced item
No future item has a verified rollout state yet.
How we counted this
the working, the words, the sources
So what: Microsoft ships constantly; you shouldn’t have to track it. Aristo surfaces one thing a month from the public roadmap, only when it reaches the licences you already hold.
Defined: skill — a saved, reusable instruction file Copilot follows (a recipe it keeps); included with Copilot licences. roadmap — Microsoft’s public feed of what’s shipping and rolling out.
Where you stand today, on the same scale this report opened with.
See this for your own organization.
One read-only consent, about thirty minutes to your first read, and every number on this page becomes yours — counted from your own tenant, then kept current every month.
§7 · Methodology
Where every number on these pages comes from — and where the words come from too.
Plain words, no marketing. Nothing here is estimated unless it says so; a pillar that isn’t available is listed as missing, never silently scored zero.
A Frontier Firm (Microsoft, 2025 Work Trend Index) is an organization rebuilt around AI — intelligence on tap, human-agent teams — not just using AI but delegating real work to it. Aristo owns the measurement; Microsoft owns the concept, and both are printed here so a first-time reader never has to take either on trust.
Where the term comes from
2025: The Year the Frontier Firm Is BornMicrosoft WorkLab · the canonical report, 23 April 2025The 2025 Annual Work Trend IndexThe official Microsoft blog announcementWork Trend Index (the hub)Microsoft’s research home for the index
What is a Frontier Firm?
Microsoft’s definition · the three pillars · the four bands
A Frontier Firm (Microsoft, 2025 Work Trend Index) is an organization rebuilt around AI — intelligence on tap, human-agent teams — not just using AI but delegating real work to it.
Aristo does not re-define the term — it measures it, for your tenant, from your own counted signals. Here is the arithmetic, in full.
Fig · The three pillars and their fixed weights. A pillar that is not available is dropped and the remaining weights re-normalized — never silently scored zero.
Adoption · 50 pts
How far AI has actually spread — your readiness score: real activation, real engagement, real outcomes.
Activation · 25 pts
How much of the assigned Copilot estate is genuinely working — one minus the share of assigned seats sitting cold.
Security · 25 pts
Your Microsoft Secure Score — counted only once you have consented to share it, never assumed.
Fig · The four bands, at their true widths on the 0–100 scale. Versioned with every run, so a month-over-month comparison always states the scale it was scored on.
Emerging
0–29 of 100
The licences are in place; everyday AI use is just beginning.
Developing
30–54 of 100
Adoption is climbing, but genuine activity is still the minority.
Advancing
55–84 of 100
Counted adoption taking hold, proven lift under way.
Frontier
85–100 of 100
AI is how the everyday work runs.
Counted, never self-reported. Every pillar above is read from your tenant’s own signals. Where a number can only be estimated, this report says so on the number itself.
Every number on these seven pages, sourced
the score · the bands · the holdout · money · peers
The Frontier Score (0–100) — three pillars, fixed weights, computed monthly from counted signals, never survey opinion: Adoption 50% (your Readiness score), Activation 25% (share of assigned Copilot seats genuinely active), Security posture 25% (your Microsoft Secure Score, only when you’ve consented to share it). If a pillar isn’t available it is dropped and the remaining weights re-normalized; a self-reported pre-consent seed flags the whole score estimated.
The bands — Emerging 0–29 · Developing 30–54 · Advancing 55–84 · Frontier 85–100. Band definitions are versioned with each run (bandsV 2 = this scale), so a month-over-month comparison always states the scale it was scored on. Bands never move retroactively.
The path to 100 — the points you have not yet earned, decomposed across the pillars at their own re-normalized weights: a pillar’s remaining points are (1 − its score) × its share of the total weight × 100, and those remainders sum to exactly 100 minus your score. The steps are therefore a DECOMPOSITION, not a ranking — nothing is apportioned by judgement. Landing dates come from a least-squares line through your own last runs and appear only when that line exists and is climbing; below three runs there are no dates, and the report says so.
The holdout — how “proven” is proven — on day 0 a deterministic 12% of your cold seats are a never-nudged control (stable, not hand-picked, not resettable). After the window, lift = nudged rate − control rate with a 95% confidence interval. Billing uses the CI lower bound, never the point estimate; below 95% confidence the report says directional, never proven. Negative lift is reported as negative. Valid only with ≥14 days and ≥8 people per arm.
Money figures — Activation opportunity = assigned-but-cold Copilot seats × your seat cost. Total recoverable exposure = that activation opportunity + priced bought-but-unassigned seats; the two are never presented as the same amount. Shelf seats = bought − assigned per SKU. Credit spend is a labeled envelope estimate until you connect Cost Management (read-only) — then it becomes measured and is labeled so.
Peer comparisons — fleet percentiles shown only as bands (top 10% / top 25% / above median…), never another org’s number, and omitted entirely below the k-anonymity floor of contributing tenants.
The stability promise — formula changes are versioned and dated; a report never silently re-scores the past. The score is allowed to go down.
Naming — “Frontier Firm” is Microsoft’s concept (WorkLab); Aristo owns the measurement, measured for your tenant.
Is your org a Frontier Firm? See yours — aristo.phoenixhalo.com/example · make it yours in 30 minutes →