Every load-bearing idea in the Copilot Practice Playbook, compressed to what you need to act on — with a link back to the full explanation every time you need the story, the source, or the granular math.
The playbook is 90 minutes. This is 15.
Full version: § The Bottom Line: The AI Frontier Is Open → and § Introducing AgentOps →
Three doors. Two are closed, one is open — and the open one is the whole playbook:
Independent studies converge on 20–60 min saved/user/day where it's deployed properly. The product works.
At $18/mo promo pricing, break-even is under 8 minutes—a week. Price isn't the obstacle.
Same product, same trial — the variable that moves outcomes is how the rollout is run. That variable lives in this room.
It's a use-case problem, and the fix sits squarely on the partner channel's shoulders.
Customers who haven't bought yet. They already have free Copilot Chat — the sale is “AI that reads your data, governed properly.” That's your conversation, not Microsoft's.
Customers who bought but aren't using it. Weekly active use runs just 20–30% of purchased seats — without intervention, the usage report writes the “don't renew” conversation for them.
Both markets are served by the same five services: Readiness › Activation › Integration › Measurement › Managed AI.
For every $1 of Copilot licensing sold, partners running this playbook generate $8.45 in services revenue on top — a floor, not a ceiling, and only for partners who wrap every license in readiness, adoption, agent builds, and managed services. IDC #US52483124, 638 partners surveyed. The license-only partner captures zero of it.
| Term | What it means |
|---|---|
| AgentOps | The work. The ongoing practice of keeping a customer's agents accurate, grounded, governed, and valuable. |
| AgentCare | The offer. The wrapped, priced, deliverable monthly service the customer buys. |
| Managed AgentOps | The packaging. Three tiers — Essentials / Operations / Transformation — detailed in §7. |
Not a Microsoft SKU, not a certification track, not analyst-defined. “The word is negotiable. The practice underneath it isn't.” The reason it has to be recurring revenue: an unmanaged agent degrades — hallucinations creep in, grounding goes stale, permissions drift — and a build-and-bill SOW leaves that gap wide open.
Full version: § Four Assets, One Practice → and § The CPB Workbook & Engagement Ops Guide →
The reader's journey: Build Belief → Model the Practice → Run the Engagement → Showing the Work. Four pieces, each doing one job.
You're reading the field guide to it now. The data, models, and motions behind a real Copilot practice. Read first — builds the belief and the business case.
Open the full Playbook →9-tab Excel. The bottom-up P&L behind every number in the playbook, plus live engagement trackers on Tabs 8–9. Open when you're ready to enter your own numbers.
Download .xlsx →The why behind every gate and phase on Workbook Tabs 8–9, with real customer stories mapped to each checkpoint. Read once before your first Activation Sprint.
Open Ops Guide →What “Frontier Firm” and “Frontier Partner” mean, the Frontier Partner specialization criteria, and where TD SYNNEX fits as a Frontier Distributor. Open when thinking 12–24 months out.
Open Frontier →How they fit: Read the playbook to build conviction. Open the Workbook to model your practice. Follow the Ops Guide when you run a live engagement. Use the Frontier Playbook to aim at the badge.
Full version: § The Market Reality → and § The Partner Failure Playbook →
Adoption struggles are not a technology problem — it's how partners position and deliver it. Seven patterns recur, every one correctable without new headcount or certifications:
| Mistake | The Fix |
|---|---|
| 1. Leading with the plumbing | Value first, architecture second. Open with a live scenario (meeting summary, contract review in 30 sec), get “I want that,” then introduce backend work as protection, not prerequisite. |
| 2. Treating Copilot like a license sale | Build a practice, not a transaction. Package Discovery → Pilot → Adoption Sprint → Renewal/Expand as billable phases. |
| 3. Piloting with the wrong people | Seed champions, not administrators. A 30-min conversation to find 5–10 “Day 1 Champions” — power users in high-friction roles (see §4). |
| 4. Dropping licenses with no change plan | Adoption is a deliverable. A billed 30-60-90 day plan: champion kickoff, role-specific prompt guides, 30-day check-in, 60-day ROI conversation. |
| 5. Selling “AI for AI's sake” | Scenarios before speeds and feeds. Walk in with 2–3 role-specific use cases from Microsoft's Scenario Library, not a TED Talk. |
| 6. Can't defend the price | Lead with the business case, not the bill. Use Microsoft's Business Case Builder / Value Envisioning Tool; articulate the service layer Microsoft-direct can't provide — never compress margin. |
| 7. Passing licensing confusion to customers | Own the clarity. Build a one-page SKU map for your customer (see §6's SKU ladder) and stay fluent on the promo landscape. |
Kloudify's Feb 2026 audit named 8 industry-wide failure modes, corroborated by a Microsoft Principal CSA and the Microsoft Learn rollout playbook. All 8 map directly onto the 7 mistakes above plus the Revenue Runway's Managed AgentOps tier (governance/compliance). See the full cross-walk table in §5 of the playbook →
Full version: § The Champion Profiles →
Seven roles deliver the fastest time-to-value. Identify one before your first training session — and use the demo tip live.
| Role | Why they win fast | 30-second demo |
|---|---|---|
| Account Manager / Outside Sales | ★ Highest-impact. 85% feel more confident as advisors; 45% less time composing email. | Generate a follow-up email + action items live from a recent Teams transcript. |
| Project Manager / Ops Lead | Meeting prep: 30 min → 10 min. | Meeting summary + action items by person + status email from a recorded standup, under 60 sec. |
| Office Manager / Exec Assistant | Cross-functional impact; turns a brain-dump into a polished message. | Turn a messy note into a professional email. |
| HR Manager / People Ops | 25% of time on policy/JD drafting; often a 1–2 person shop. | Draft a job description, then an onboarding checklist, from one 3-sentence brief. |
| Finance Manager / Controller | 35% increase in Excel formula use; fast trend analysis. | Ask Copilot to find top 3 trends and flag outliers in an anonymized sheet. |
| Marketing Coordinator | 67% use Copilot in Word for content; 26% less editing time. | Turn one product description into 3 formats in 30 seconds. |
| Meeting-Heavy Middle Manager | 60–70% of the week in meetings — #1 wish-list item across 30+ orgs. | Ask about a missed meeting's decisions and follow-ups — “the moment the deal closes.” |
Full version: § The Show and Tell Sales Motion →
Replace the slide deck. Show first, prove second, protect third. Tagline: “Give Us an Hour Today — We'll Give You an Hour Back Every Day.”
If your own team isn't using Copilot internally, that's the first problem to solve. Three ready demos: The Breakfast Briefing (voice triage of email/Teams before the office), The Information Hunter (kill the search tax with 3 no-setup prompts), and The Meeting Prep Agent (the one you built for yourself — the “laptop-flip” moment). Revenue unlocked: credibility no competitor can match.
Personal agents ground in SharePoint/OneDrive/Teams with no admin ticket — until the customer asks “can it pull from our CRM?” That question, typically surfaced within 60–90 days, is the Copilot Studio opportunity. Revenue unlocked: $1.5K–$3K enablement workshops + Studio pipeline.
“If anyone can build an agent grounded in our company data, we need to know our data is locked down first.” Not a separate conversation — the natural conclusion of the demo you just ran. Revenue unlocked: $10K–$15K data governance & security project. That project funds the agent builds; the agent builds fund the managed retainer.
Full version: § The Agent Maturity Model → and § The Copilot SKU Ladder →
It's a destination, not a prerequisite — you only need to be one step ahead of your local market. Anchor customer conversations around progression, not products.
| Stage | Licensing | Where you bill |
|---|---|---|
| 1. Copilot Chat | Included with M365 | One-time readiness assessment & prompt training |
| 2. Copilot Studio Basic | Copilot license required | Adoption workshops + first agent build (project + early retainer) |
| 3. Studio + Connectors | Power Platform add-on | Integration build + agent governance (recurring retainer begins) |
| 4. Studio + Custom Code | Azure Functions | Custom dev project + expanded managed AI services |
| 5. Azure AI Agent Service | Azure AI Foundry | Full AI estate management — highest-value managed tier |
| 6. Custom SaaS Platform | Azure OpenAI / Partner IP | Recurring platform licensing — own IP, own revenue stream |
| Stage | SKU | Price | Partner Revenue |
|---|---|---|---|
| 1 | Copilot Chat | Free | Conversation starter only |
| 2 | M365 Copilot Business | $21/user/mo | Margin + $5K–$10K Data Readiness |
| 2.5 | Personal Agent Creation | Included | $1.5K–$3K Enablement Workshops |
| 3 | M365 Business Premium | $22/user/mo | Margin + $10K–$15K Security Deployment |
| 3.5 | Copilot Cowork (Agentic) | Included | Accelerates Premium/Purview + AgentOps pipeline |
| 4 | Copilot Studio | $200/mo (25K msgs) | $15K–$50K+ Custom Agent Development |
| 5–6 | E3+Copilot / E5 | $66 / $57/user/mo | Licensing uplift + $50K+ Enterprise Migration |
| 7 | M365 E7 (Frontier) | $99/user/mo | Managed AgentOps retainer opportunity |
| Add-on | Agent 365 | $15/user/mo (or in E7) | Managed governance retainer; pairs with AgentOps tier |
300-seat hard cap on Copilot Business and all Business SKUs. Agent 365 (GA May 1, 2026) is the governance control plane — not included in E3 or E5. Always name the next stage before you leave the room.
Full version: § The Revenue Runway →
Of the $8.45 generated per $1 of licensing: ~$3 is one-time project work; ~$5.45 is ongoing Managed AgentOps. Two frameworks, kept deliberately distinct — a customer can be at Stage 2 of their journey while you're still selling at Tier 1.
| Stage | Timeline | Pro Services Revenue | Managed Tier |
|---|---|---|---|
| 1. Deployment & Adoption | Months 1–3 | $8K–$23K | Tier 1: $10–15/user/mo |
| 2. Agent Creation & Automation | Months 3–9 | $3.5K–$8K+/agent | Tier 2: $20–25/user/mo |
| 3. Power Platform Integration | Months 6–18 | $15K–$50K+ | Tier 2 (expanded) |
| 4. Azure AI Custom Solutions | Month 12+ | $30K–$75K+ | Tier 3: $45–55/user/mo |
Modeled — not a documented case study, validated against Forrester/IDC benchmarks (see § Showing Our Work →). 3-Year engagement value from a single SMB customer at the Stage 3 level: $75,000–$110,000 — more if the customer reaches Stage 4.
| Tier | Price | Who it's for | Margin |
|---|---|---|---|
| Tier 1 — AI Essentials | $10–$15/user/mo | 10–50 users, just licensed, no agents yet | ~48% |
| Tier 2 — AI Operations | $20–$25/user/mo | 20–100 users, ≥1 active agent | ~35% |
| Tier 3 — AI Transformation | $45–$55/user/mo | 50–200 users, fully committed to AI | ~32% |
Start at Tier 1 for 6–12 months. Tier 2 enters once you've built a stable base and certified your first AI Specialist. It's fine to start as a broker via TD SYNNEX ServiceSolv (see §10).
| Component | Rate | Role |
|---|---|---|
| A. Per-User Base | $10–15/user/mo | Floor in every tier — governance, adoption monitoring, QBR, helpdesk |
| B. AgentCare | $350–400/agent/mo, ~55% GM | Core MRR from Tier 2 on. Never absorbs the build fee ($3,500–$8,000+/agent, billed separately) |
| C. Credit Wrap | 25–35% markup, ~85% GM | Azure/Copilot Studio consumption markup at CSP cost — near-pure-margin add-on |
| D. Agent 365 Overlay | $10/user/mo surfaced (list $15) | Optional at Tier 2, default at Tier 3 — governance control plane |
Full version: § Outcome-Based Project Work →
Fee tied to a measurable change in a business metric, not delivery of a system. A finite sprint (60–120 days): one KPI agreed in advance, a documented baseline, a fixed base fee + 10–25% success kicker, five gates, and a pre-agreed conversion path into MRR.
The customer can name one measurable number; the baseline is accessible before you quote; the sponsor personally cares; the MRR conversion path is signed before the sprint starts.
No baseline and no time to build one; the KPI depends on variables you don't control; the sponsor is director-or-below; it's really ongoing managed work that should be retainer-priced.
| Gate | When | What clears it |
|---|---|---|
| 1. Scoping | Week 2 | KPI agreed, baseline captured, counter-metric named, scope locked with a Plan B clause |
| 2. Build | Week 6 | Agent live in tenant, weekly regression eval set (50–150 prompts), pilot cohort scheduled |
| 3. Midpoint | Day 45 | KPI trending; trigger Plan B openly if >25% behind |
| 4. Measurement | Day 75 | Cohort read, counter-metric read, KPI certified or not |
| 5. Close-Out | Day 90 | Kicker invoiced, MRR conversion signed, runbook handed over |
| Sprint | Tier | Pricing | Converts into |
|---|---|---|---|
| Activation Sprint | Tier 1 | $15–25K + 15% kicker | Tier 1 retainer on day 91 |
| Workflow Sprint | Tier 2 | $7,500 + 20% kicker/agent | $350/agent/mo AgentCare |
| Quarterly Business Impact Commitment | Tier 3 | $15–25K/qtr + 20% kicker | Layers alongside AgentCare, not instead of it |
Failure mode to avoid: treating “the agent is live” as the outcome. The outcome is the KPI moving.
Full version: § The Competitive Battle Card →
Don't argue model quality — Gemini, ChatGPT, and Claude are capable platforms and the gap is narrowing. Argue data gravity, governance inheritance, and channel economics.
| Platform | Partner resale margin | SMB channel readiness |
|---|---|---|
| Microsoft Copilot (CSP) | Recurring, stacked FY26 incentives: 3.75% core + 7% strategic + 7.5% growth accelerator | Built for the channel |
| Google Gemini Enterprise | ~40% Y1 net-new, 50–70% managed services, but renewal margin compressed ~20%→~12% | Hunter-weighted, not farmer-weighted |
| Anthropic Claude | None for SMB — $100M Claude Partner Network anchored to Accenture/Cognizant/Infosys/Deloitte | No volume SMB program |
| OpenAI ChatGPT Enterprise | None published for SMB — narrow reseller list, mostly direct | No SMB channel |
Same underlying Claude agentic model — Microsoft built Copilot Cowork in partnership with Anthropic. Claude Cowork runs locally, no tenant context, no governance: a transactional sale. Copilot Cowork runs in the M365 tenant, powered by Work IQ, inherits Purview/DLP/audit: a practice-builder. The $100M Anthropic just invested is going to Accenture and Infosys — the CSP model was built for you.
Almost no SMB runs an all-Microsoft AI stack. Your Microsoft stack (Defender for Cloud Apps, Purview, Entra ID) is the governance layer for every AI tool in the building — ChatGPT and Claude included.
| Rung | Timeline | Price |
|---|---|---|
| 1. Shadow AI Discovery Sprint | 2–3 wks | $3.5K–$7.5K |
| 2. AI Acceptable-Use Policy & Enablement | 1 wk | $1.5K–$3.5K |
| 3. ChatGPT or Claude Enterprise Cutover | 2–4 wks | $5K–$12K |
| 4. Custom GPT / Claude Project Pack | 3–5 wks | $6K–$18K |
| 5. DLP-for-AI Deploy | 2–4 wks | $4K–$10K |
| 6. AgentCare Managed Retainer | Ongoing | Per Ops Guide |
Rungs 1–5 stack to $20K–$51K in fixed-fee work before the retainer even starts, sized for a 10–100 seat SMB.
Full version: § The CPB Workbook & Engagement Ops Guide →
Plan the practice: open the Workbook (Tabs 1–3, 5). Onboard a Delivery Lead: send the Ops Guide first, then hand over Tabs 8–9. Run a live engagement: keep the Workbook open as tracker; pop to the matching Ops Guide section when a gate decision stalls. “The Workbook is the tool. The Ops Guide is the manual. You want both.”
Download Workbook (.xlsx) → · Open Ops Guide (HTML) →
Full version: The Frontier Partner Playbook →
| Tier | What it unlocks |
|---|---|
| 1. MAICPP Membership | CSP/NCE transaction rights, Partner Center access — “keeps the lights on,” no differentiation |
| 2. Solutions Partner | Meaningful IUR licensing, CSP Deployment Accelerator eligibility, Copilot SMB funding — the first economically rational tier |
| 3. Advanced Specialization | Larger funding tranches, priority engagements — where services credibility replaces price pressure |
| 4. Frontier Partner Specialization | Frontier-aligned co-sell eligibility, Microsoft Agent pre-purchase plan (P3) credits, Microsoft 365 E7 licensing, and Concierge support. Live September 30, 2026; the older Frontier Partner badge retires at the end of June 2027 |
| 5. Frontier Distributor | Distributors only — 83-point third-party audit (TD SYNNEX holds this, since March 2026) |
Most SMB-focused partners should not chase the Frontier Partner specialization immediately. Path: Solutions Partner → the Microsoft 365 Copilot specialization first (fastest ROI). Pursue Frontier only once you're ready to run Copilot Studio at scale and make AgentOps a real revenue line. “The goal is not to reach the top — it's to be one tier ahead of every competitor in your market.”
Full version: § Showing Our Work → and § Real Partners, Real Results →
3× campaigns per month. Campaign briefs: 3 hrs → 30–60 min. Legal team uses Copilot as a “junior associate.”
Built an entire internal LMS — slides, scripts, process docs — without contracting an outside vendor.
Scientific content creation dropped from weeks to days; staff time shifted from formatting to analysis and QA.
Flatter org — senior-level strategic input on every brief without adding headcount.
Source: Microsoft Customer Stories, customers.microsoft.com. “A food company, a law firm, a health agency, and a creative shop. Four different industries. One story.” None waited for a perfect environment or a six-month readiness assessment.
Synthesized from §1 (“Why This Conversation Is Worth Having”), the Tier 1 prerequisites, and the MCI program. Full detail: § Why This Conversation Is Worth Having →
Full sourcing: § Showing Our Work →
The partners who move first, build credibility through their own deployments, and wrap AI into a structured services practice will own the next decade of SMB IT. This guide gets you the shape in fifteen minutes. The full playbook → gets you the rest.