Dynamics 365 AI Readiness Audit: 12 Checks Before Copilot Deployment
Every Microsoft partner conversation about Copilot eventually hits the same wall. The demo looks flawless — AI-generated email summaries, opportunity insights, next-best-action recommendations rolling out in seconds. Then the customer goes live, and within two weeks the enthusiasm evaporates. Copilot is surfacing the wrong account owner. It is summarising stale pipeline stages from 2022. It cannot find meeting notes because they were typed into free-text fields with no structure. The technology is not broken. The environment is.
This is the uncomfortable truth that most Copilot deployment guides gloss over: the quality of your AI outputs is a direct reflection of the quality of your CRM data and configuration. Garbage in, garbage out has never been more consequential than when an AI assistant is confidently presenting that garbage to your sales director in a boardroom summary.
At CRMONCE, we have audited dozens of Dynamics 365 environments ahead of Copilot rollouts. What follows is the structured 12-point readiness framework we use — covering data, process, security, and configuration — along with a Red/Amber/Green scoring system and a realistic remediation roadmap so you know not just what to fix, but in what order and by when.
Why CRM Data Quality Is the Single Biggest Predictor of Copilot ROI
Copilot features in Dynamics 365 — whether that is Copilot for Sales, the embedded meeting summaries, or the opportunity pipeline assistant — rely on your existing CRM records as their primary context. Unlike a general-purpose language model, these features are grounded in your data. That is the feature, not a limitation. But it means every structural flaw in your CRM is amplified rather than hidden.
Consider three common data problems we encounter repeatedly:
- Duplicate account records: When the same company exists as "Contoso Ltd," "Contoso Limited," and "Contoso UK" in your system, Copilot cannot consolidate the relationship history. Meeting summaries reference only one branch of the account, activity timelines appear incomplete, and AI-recommended contacts may be attached to the wrong parent.
- Missing or incorrect owner records: Opportunities owned by deactivated users, or accounts assigned to a generic team queue, produce AI summaries with no meaningful owner context. Copilot for Sales cannot generate personalised outreach suggestions when it does not know who owns the relationship.
- Stale opportunity stages: Deals sitting in "Proposal Sent" for 18 months, never closed or progressed, pollute pipeline forecasting outputs. Copilot's deal summary will present these as active pipeline risks — because to the system, they are — creating noise that erodes trust in every subsequent AI output.
These are not edge cases. In a recent audit of a 500-seat Dynamics 365 environment, we found that 34% of account records had at least one missing mandatory relationship field, and over 1,200 opportunities had not been updated in more than six months. Deploying Copilot into that environment would not accelerate the sales team — it would accelerate their confusion.
The 12-Point Dynamics 365 Copilot Readiness Checklist
Use the RAG rating to score each item: Red = blocker, fix before deployment; Amber = significant risk, fix within 30 days; Green = acceptable, monitor ongoing.
Data Readiness Checks
Check 1 — Duplicate Record Rate (Accounts & Contacts)
Run a duplicate detection job in Dynamics 365 and export the results. If your duplicate rate exceeds 5% for accounts or 8% for contacts, rate this Red. Between 2–5%, rate Amber. Copilot's account summarisation and relationship mapping is directly impaired by duplicates — this is a hard blocker at scale.
Check 2 — Mandatory Field Completeness
Audit your core entities (Account, Contact, Opportunity, Lead) for completeness on fields that Copilot uses as context: Industry, Account Type, Primary Contact, Estimated Revenue, Close Date, and Current Stage. Use a Power BI report or a simple Advanced Find export. Below 80% completeness on any single key field is Red.
Check 3 — Record Owner Validity
Query all active records where the assigned owner is a disabled user or a shared mailbox without a mapped user identity. Any result above zero is Amber minimum. If more than 3% of pipeline records are affected, escalate to Red. Copilot uses owner identity to personalise outputs — orphaned records produce generic, useless responses.
Check 4 — Opportunity Stage Freshness
Flag any opportunity where the Stage Modified Date has not changed in over 90 days for early-stage deals (Qualify, Develop) or 180 days for later-stage deals. These are stale pipeline records. Export, review, and either progress, close, or archive before Copilot deployment. A rate above 15% of your open pipeline is Red.
Check 5 — Activity and Email Tracking Coverage
Copilot for Sales summarises relationship activity. If your team is not logging calls, emails, and meetings consistently in Dynamics 365, there is nothing for Copilot to summarise. Check what percentage of active accounts have had at least one logged activity in the past 60 days. Below 50% is Red. This is often a process issue as much as a data issue — see Check 9.
Check 6 — Notes and Attachment Structure
Semantic search features in Copilot (used in meeting summaries and account research) rely on structured or semi-structured text. Free-text notes that contain abbreviations, internal jargon, or are extremely brief (<20 words) have low retrieval value. Review a sample of 50 notes across your top accounts. If the majority are unstructured fragments, rate Amber and create a notes template or structured activity type to guide future input.
Process Readiness Checks
Check 7 — Audit Trail Configuration
Dynamics 365 audit logging must be enabled at both the organisation level and the entity level for Copilot features that surface change history and data lineage. Navigate to Settings > Auditing and confirm that audit is active for Account, Contact, Opportunity, and Lead at minimum. Disabled audit on any core entity is Red — not just for Copilot, but for compliance.
// Power Automate or Plugin check — confirm audit is enabled
// In Dataverse, retrieve organisation audit settings:
GET [org]/api/data/v9.2/organizations?$select=isauditenabled,auditretentionperiodv2
// Entity-level audit — check via metadata:
GET [org]/api/data/v9.2/EntityDefinitions(LogicalName='opportunity')?$select=IsAuditEnabled
Check 8 — Business Process Flow Consistency
If multiple Business Process Flows (BPFs) exist for the same entity — for example, three different sales processes for Opportunity — Copilot cannot reliably interpret stage context. Audit your active BPFs and rationalise to one primary flow per entity. Multiple conflicting BPFs on core entities are Amber; if stages are inconsistent across records in the same pipeline view, escalate to Red.
Check 9 — CRM Adoption and Data Entry Discipline
Copilot is an amplifier — it makes good habits more productive and bad habits more visible. Review your team's CRM usage metrics: percentage of opportunities with a logged call this month, percentage of contacts with a valid email address, percentage of meetings tracked via the Dynamics 365 Outlook add-in. If adoption is below 60% on core activity logging, Copilot will surface sparse, unreliable summaries. Rate below 60% as Red.
Security and Configuration Checks
Check 10 — Security Role Hygiene
Copilot respects Dynamics 365 security roles — it will only surface records a user has permission to read. However, if security roles are poorly configured (users with System Administrator access who should have Salesperson roles, or custom roles with excessive read permissions), Copilot may surface sensitive records to users who should not see them. Audit your top 10 most-used security roles. Any role with write access to system settings that is assigned to non-admin users is Red.
Check 11 — Copilot Feature Entitlement and Licence Mapping
Confirm that users who will access Copilot features hold the correct licences — Dynamics 365 Sales Premium or Microsoft 365 Copilot, depending on the feature set. Mismatched licence assignments cause silent feature degradation where Copilot appears to work but operates in a reduced capacity without surfacing an error. Map every target user to their licence assignment in the Microsoft 365 Admin Centre before go-live. Any gaps are Red.
Check 12 — Data Residency and Privacy Configuration
For organisations in regulated industries or operating under GDPR, confirm that your Dynamics 365 environment is configured for your required data residency region and that Copilot's connected experiences comply with your data processing agreements. In the Power Platform Admin Centre, verify that "Move data across regions" is set appropriately and that you have reviewed the Microsoft Copilot data privacy documentation. Non-compliance with your data residency requirements is an immediate Red and a legal matter, not just a technical one.
Your 30/60/90-Day Remediation Roadmap
Once you have scored your environment, prioritise remediation in the following sequence. All Red items must be resolved before Copilot goes live. Amber items should be resolved within the first 30 days post-deployment, with monitoring in place from day one.
Days 1–30: Clear the Blockers
- Run and resolve duplicate detection jobs for Account and Contact entities. Use the built-in Duplicate Detection Rules and assign ownership of merge operations to a named data steward.
- Reactivate or reassign orphaned record owners. Export the list, confirm with line managers, and bulk-reassign using a Power Automate flow or the Dynamics 365 bulk edit feature.
- Enable audit logging at organisation and entity level. Document the configuration in your internal runbook.
- Resolve security role misassignments identified in Check 10. Create a role review sign-off process with your IT security team.
- Confirm licence assignments for all target Copilot users.
Days 31–60: Structural Improvement
- Archive or close stale opportunities after a review cycle with sales managers. Set up a monthly pipeline hygiene meeting with a Power BI report surfacing deals not updated in 60+ days. (If you have read our CRM Drift post, this is the operational muscle that prevents drift from returning.)
- Introduce a structured notes template or custom activity type to improve the quality of relationship notes across key accounts.
- Rationalise Business Process Flows — retire unused flows and standardise the active pipeline BPF.
- Launch a short CRM adoption push: a 30-minute team session on logging activities via the Outlook add-in, with a leaderboard in Power BI tracking weekly activity volume per user.
Days 61–90: Monitor and Optimise
- Deploy a Power BI Copilot Readiness Dashboard tracking: duplicate rate trend, mandatory field completeness by entity, open opportunity stage freshness, and weekly activity logging rate. Review monthly.
- Collect qualitative feedback from Copilot users — specifically examples where AI output was inaccurate or unhelpful — and trace each back to a data or configuration root cause.
- Establish a quarterly data health review as a standing agenda item, preventing the environment from regressing into the conditions that created the original readiness gaps.
Turning Readiness Into a Competitive Advantage
Most organisations treat Copilot deployment as a licence activation exercise. Switch it on, run the training session, and wait for productivity gains. The organisations that will actually realise the ROI are the ones that treat AI deployment as the forcing function it is — an audit trigger that surfaces every structural weakness in their CRM that they have been tolerating for years.
A clean, well-governed Dynamics 365 environment does not just produce better Copilot outputs. It produces better reporting, better forecasting, better onboarding for new sales staff, and better customer experiences. Copilot is the catalyst. The data and process discipline is the compound interest.
At CRMONCE, we offer a structured Dynamics 365 AI Readiness Assessment that delivers a scored version of this checklist specific to your environment, along with a prioritised remediation plan and implementation support. If your team is planning a Copilot rollout in the next quarter and you want to avoid the post-launch disappointment that comes from skipping this step, reach out to our team in Hyderabad — we will tell you exactly where you stand before you go live.
Source reference: CRM Software Blog, "7 CRM Data Problems That Hurt Your Business" — CRMONCE extends this foundation to include process, security, and configuration readiness with a scored remediation framework.