MSP AI Services Scope and Pricing Template: Readiness, Governance, and Managed AI
An AI services quote needs three separate offers, each with its own scope, exclusions, and pricing inputs. Sell them as one line item called "AI" and the client will read it as unlimited.
The three offers are an AI readiness assessment, an AI governance setup, and a managed AI service. The first two are fixed-fee projects with a finish line. The third is recurring work with a support boundary. Keep the license resale on its own line for all three.
This template is for the vCIO or account lead who has to turn "can you help us with AI?" into something delivery can staff and the owner can price. It gives you a scope table per offer, a copyable scope record, a cost worksheet, the change triggers that should reopen the quote, and a client summary.
If you are still deciding what to charge, the MSP AI services pricing guide covers packages and price logic. This page is the paperwork that keeps those prices true after the client says yes.
The three offers at a glance
| Offer | Client trigger | What they get | Pricing basis |
|---|---|---|---|
| AI readiness assessment | "Are we safe to turn on Copilot?" or a department already using AI tools nobody approved | Findings on permissions, data exposure, licensing, and user groups, plus a prioritized remediation list | Fixed fee from estimated assessment hours |
| AI governance setup | Leadership wants rules before rollout, or an auditor or insurer asked about AI use | Approved tool list, acceptable use policy draft, access and data decisions, review cadence | Fixed fee from estimated workshop, drafting, and configuration hours |
| Managed AI service | AI tools are live and someone has to own licenses, usage, onboarding, and support | Recurring license and usage administration, reporting, and a defined support allowance | Monthly fee from recurring hours, plus license and usage pass-through |
Sell them in that order when you can. The assessment produces the findings that size the governance work, and the governance decisions define what the managed service has to enforce.
Offer 1: AI readiness assessment
Most of the readiness work is permissions work. Microsoft's Copilot documentation says it only surfaces organizational data that individual users have at least view permissions to. That sounds reassuring until you remember how many SharePoint sites in a ten-year-old tenant are shared with "Everyone except external users." Turn on Copilot and that oversharing gets a search box.
| Scope field | Fill in before pricing | Evidence or acceptance |
|---|---|---|
| Tenant and users | [Tenant], [licensed user count], [departments in scope], [pilot group] | User list reconciled against the client's current staff list |
| Data locations | [SharePoint sites], [Teams], [OneDrive scope], [shared mailboxes] | Locations reviewed and listed with owner and sharing level |
| Permissions review | [Sampling method], [broad-access groups to check], [external sharing settings] | Each overshared location logged with owner, exposure, and proposed fix |
| Sensitive data | [Data types the client names], [existing labels or classification, if any] | Client confirms which data types matter before findings are ranked |
| AI tools already in use | [Known tools], [how discovered: survey, sign-in logs, expense reports] | Inventory of tools with owner and business use |
| Licensing | [Current licenses], [AI add-ons under consideration] | License fit noted per user group |
| Deliverable | [Findings report], [remediation list], [review meeting] | Client sponsor accepts the report in a review meeting |
Out of scope unless quoted separately: fixing the permissions you find, deploying sensitivity labels, migrating data, building agents, and training users. The assessment ends with a decision list. Remediation is the next quote.
Assumptions to write down: the client provides admin access for the review window, a sponsor who can approve findings, and named data owners for the locations you flag. When nobody owns a site, record that as a finding and move on.
Offer 2: AI governance setup
Governance is where the client decides who may use which AI tools, with which data, and who checks. The NIST AI Risk Management Framework is a voluntary framework organized around four functions: govern, map, measure, and manage. It is a useful structure for the conversation with an SMB client, and it keeps you from inventing a framework in a workshop.
| Scope field | Fill in before pricing | Evidence or acceptance |
|---|---|---|
| Decision makers | [Executive sponsor], [policy approver], [data owners] | Names recorded before the first workshop |
| Approved tools | [Tools to approve], [tools to block or restrict], [request process for new tools] | Approved list signed off by the sponsor |
| Acceptable use policy | [Draft scope], [departments], [review by client counsel or HR] | Client approves the final policy text |
| Data rules | [Data types allowed in AI tools], [data types prohibited], [labeling decisions] | Rules mapped to the readiness findings |
| Access and configuration | [Groups licensed], [tenant settings to change], [logging to enable] | Change list approved and completed in the agreed window |
| Review cadence | [Quarterly or semiannual review], [who attends], [what triggers an early review] | First review date set before handoff |
Out of scope unless quoted separately: legal advice, regulatory certification, writing department-specific procedures, and enforcing the policy on unmanaged personal devices. Say plainly that the client owns the policy. You draft it and configure the controls you manage.
Offer 3: Managed AI service
Once AI tools are live, someone has to add and remove licenses, watch usage, answer "why can't I see Copilot in Outlook?", and report to the buyer. That work is recurring, and it grows with headcount and tool count.
| Scope field | Fill in before pricing | Evidence or acceptance |
|---|---|---|
| Covered tools | [Named AI tools and license types] | Tool list matches the governance approved list |
| Covered users | [Licensed user count], [billing unit], [minimum] | Count reconciled monthly against license reports |
| License administration | [Adds, removes, reassignments], [request route] | Changes logged with requester and approver |
| Usage and cost reporting | [Report cadence], [recipients], [usage or credit metrics included] | Report delivered and reviewed at the QBR |
| Support allowance | [Included hours or tickets per month], [what counts as AI support] | Overage tracked and reported before it is billed |
| Policy upkeep | [Review cadence], [tool request handling] | Approved list updated after each review |
Out of scope unless quoted separately: custom agent or workflow builds, prompt libraries for each department, data cleanup projects, and any promise about productivity gains. If a client wants an automation built, treat it as a project with its own acceptance test. The documentation and runbook scope worksheet shows how to separate authoring from maintenance.
Usage-based charges need their own rules before the first invoice. The AI billing guide for MSPs covers included usage, overage thresholds, and invoice language.
Copyable AI services scope record
Fill this in per client, per offer. If a field stays blank, that blank is the next discovery question.
Client / offer / scope revision:
Offer: readiness assessment / governance setup / managed AI service
Executive sponsor and approver:
Tenant / licensed users / departments in scope:
AI tools in scope (approved list version):
Data locations in scope:
Data owners named (yes / no / partial):
Admin access window and method:
Deliverables:
Acceptance method and approver:
Assumptions:
Exclusions:
Included hours or allowance:
License and usage pass-through items (product, unit, unit cost, qty):
Change triggers that reopen this scope:
Recurring review date:
Unresolved items / owner / due date:
Pricing worksheet
Price each offer from your own labor cost and margin target. License resale and usage charges get their own lines, so the client can see what is your work and what is the vendor's price.
| Input | Readiness assessment | Governance setup | Managed AI service |
|---|---|---|---|
| Labor hours | [Discovery, review, report, meeting] | [Workshops, drafting, configuration, handoff] | [Monthly admin, reporting, support allowance] |
| Loaded labor cost per hour | [Your number] | [Your number] | [Your number] |
| Tooling cost | [Any assessment tooling] | [Usually none] | [Reporting or monitoring tools, allocated] |
| License resale | Separate line | Separate line | Separate line, with your markup policy |
| Usage or credit pass-through | Not applicable | Not applicable | Separate line, billed per the agreed threshold |
| Target gross margin | [Your target] | [Your target] | [Your target] |
Worked example: one readiness assessment and one managed service
Illustrative scenario only. These are invented planning inputs, not market prices, customer results, or Scopable pricing. Assume a 50-user client, a loaded labor cost of $75 per hour, and a 40% direct gross margin target.
The readiness assessment takes an estimated 16 hours across discovery, permissions review, the report, and the review meeting.
Direct cost = 16 hours x $75 = $1,200
Required price = direct cost / (1 - target gross margin)
Required price = $1,200 / (1 - 0.40) = $2,000
The managed AI service needs an estimated 3 hours a month for license changes, reporting, and routine questions.
Monthly direct cost = 3 hours x $75 = $225
Required monthly fee = $225 / (1 - 0.40) = $375
Planned gross profit = $375 - $225 = $150
Now say the sales manager asks for "a few prompts for the team" and someone spends 4 extra hours building them inside the monthly fee. Direct cost becomes $525 against $375 of revenue. That month loses $150 before anyone notices, and next month the marketing team asks for the same thing.
Prompt writing is a fine thing to sell. Put it on the quote first. The MSP margin protection guide covers the wider quoting habits that keep this from happening.
Change triggers that reopen the quote
Write these into the scope so a change in the client's environment starts a pricing conversation before anyone does free work:
- Licensed users grow past the agreed band or minimum.
- The client adds a new AI tool to the approved list.
- A department asks for a custom agent, automation, or prompt library.
- A readiness finding turns into remediation work, such as a permissions cleanup or labeling project.
- A vendor changes license terms, usage pricing, or credit rules.
- The client acquires another company or tenant.
When one fires, use a change order with the new scope, price, and approver. The MSP change order template has the format.
Client-facing summary
Adapt this to your agreement with commercial and legal review. It describes the service. Your agreement still governs.
Service: [Offer name] for [client].
What we will do: [deliverables in one or two sentences].
What you will provide: [sponsor, admin access, data owners, review attendance].
Included: [hours, allowance, or deliverable list].
Not included: [exclusions in plain words].
Pricing:
- [Offer] fee: $[amount] [one-time / per month]
- [AI license or product]: [qty] x $[unit price], billed separately
- [Usage or credits, if any]: billed per [threshold rule]
Changes: New tools, new departments, custom builds, or license changes
are quoted before work starts.
Review: [cadence], starting [date].
Carry the findings into scope
The readiness assessment is where the rest of the AI work gets sized. Findings about overshared sites, unknown tools, and missing data owners become the governance scope and the first remediation quote. That only works when the findings and assumptions stay attached to the client record where the next quote starts.
Scopable's assessment and scoping workflow turns client facts, findings, risks, and assumptions into a reviewable scope before anyone prices the work, then carries the approved scope into the quote. Scopable does not deploy Copilot, change tenant permissions, or monitor AI usage. Your team and tools do that work.
To turn your next AI readiness assessment into scoped, priced work, start a free Scopable account.
Frequently Asked Questions
How should an MSP price an AI readiness assessment?
Price it as a fixed-fee project from estimated hours: discovery, a permissions and sharing review, an inventory of AI tools already in use, the findings report, and the review meeting. Multiply the hours by your loaded labor cost, divide by one minus your target gross margin, and quote licenses on a separate line. Remediation of the findings is a separate quote.
What belongs in an MSP managed AI service scope?
Name the covered AI tools and licensed users, then define license administration, usage and cost reporting, a monthly support allowance, and policy upkeep. Bill license resale and usage or credit charges as separate lines, and list the change triggers that reopen the price, such as new tools, user growth, or a request for a custom build.
What should an MSP exclude from an AI services quote?
Exclude permissions cleanup and labeling projects found during the assessment, custom agent or workflow builds, department prompt libraries, legal advice on AI policy, and any promised productivity gain. Each can be sold, but each needs its own scope, acceptance test, and price.
