Best Practices Guide: AI CRM Concierge

What You’ll Do

• Set up automations that define when and how CRM updates are triggered
• Choose between manual review or automatic syncing
• Decide whether to use Magic Mode (AI fills in all possible fields) or Strict Mode
• Define CRM field context to guide what the AI extracts
• Control who can access or modify each automation

Do It Today

• Pilot a CRM automation for a key meeting type like “Demo” or “Handover”
• Enable manual approval mode first to review suggestions
• Add short prompts or hints in your CRM field descriptions (e.g., "Include KPIs mentioned on call")
• Restrict sync access to relevant users or teams for testing

What a Great CRM Automation Looks Like

Copy-Paste AI Context Examples:

1. Scoring Framework

Evaluate each of the seven MEDDIC components individually:

- Metrics
- Economic Buyer
- Decision Criteria
- Decision Process
- Identify Pain
- Champion
- Competitor

Each component must be assessed according to the following qualitative scale:

- 1 (Poor): Insufficient or unclear information; weak understanding or execution of this MEDDIC element.
- 2 (Fair): Partial understanding or moderate evidence of this MEDDIC element; room for improvement.
- 3 (Excellent): Strong evidence and clear articulation of this MEDDIC element; well-developed and validated.

2. Score Calculation

- The maximum possible raw score is 21 points (7 components × 3 points each).
- A score of 21 points = 100% completion.
- For each empty or missing field, apply a deduction of 14.28% of the total possible score.
- The final MEDDIC score should be normalized and rounded to two decimal places.

Calculation Formula:

Final Score=Raw Score×(1−0.1428×Number of Missing Fields)\text{Final Score} = \text{Raw Score} \times (1 - 0.1428 \times \text{Number of Missing Fields})Final Score=Raw Score×(1−0.1428×Number of Missing Fields)

3. Output Example

If 6 out of 7 fields are completed and the total score from those fields is 18:

- Raw Score: 18
- Deduction: 14.28% (for 1 missing field)
- Final MEDDIC Score:
 18×(1−0.1428)=15.4318 × (1 − 0.1428) = 15.4318×(1−0.1428)=15.43
- Completion Percentage:
 (15.43÷21)×100=73.47(15.43 ÷ 21) × 100 = 73.47%(15.43÷21)×100=73.47

4. Output Requirements

- Display both the final MEDDIC score (two decimal places) and the completion percentage.
- Ensure all numeric outputs are rounded to a maximum of two decimal places.
- Always reflect deductions clearly when fields are missing.

You are responsible for updating structured records with new values under predefined fields mentioned in the "Include fields" section.

Follow these rules strictly when adding a new value:

1. For each field:
   - If there is an **existing value**, do not overwrite it.
   - **Only prepend the new value** at the top of the field.
   - Insert a **date stamp (YYYY-MM-DD)** on its own line directly above the new value.
   - Add **exactly one blank line** after the new value.
   - **Do not read, reuse, or copy** any content below the first blank line. It is preserved as-is for historical tracking and must not be duplicated or altered.

2. If the field is empty, simply add the new value with a date stamp (no blank line needed at the end).

3. Always preserve the historical trail of values in **reverse chronological order** (newest first).

---

### Special Instructions for the field `"Meeting Summary"`:
- This field contains a **summary of the most recent strategic account call**.
- Write the summary in a **clear, structured format** with **spaces between points**.
- Do **not exceed 600 characters** for the summary block.
- Place a **blank line** below your block to separate it from the historical content.
- **Do not reuse or re-copy** any previous summaries already in this field.

---

### Example Format:

[2025-09-28]  
• Customer requested an accelerated pilot deployment for two business units.  
• Internal legal review pending; decision expected by 2025-10-03.  
• Next status update call scheduled for 2025-10-06.  

<blank line>
(past content follows – do not touch)

1. Only make suggestions to update fields if they were actually discussed in the meeting.

2. For the text fields, please ensure that you overwrite the field instead of appending new information.

3. Ensure the "Follow-up Details" field does not exceed 255 characters. Always overwrite this field.

4. Context for "Customer Goals" field:
Capture concrete goals for onboarding or renewal — e.g. go-live dates, KPIs, rollout milestones.

5. Context for "Follow-up Date" field:
Enter the date of the next customer meeting, follow-up call, or agreed next step. Only update this field if a specific date was mentioned (e.g., "Next check-in on Oct 12").

1. Only update fields that were explicitly discussed during the call.

2. For all editable fields, fully overwrite the content — do not append.

3. The "Renewal Risk Notes" field should be limited to 2 short sentences.

4. Context for "Decision-Making Status":
Indicate if the account can renew or expand directly. If unclear, use "TBD" or "Unknown."

5. Context for "Renewal Risk Reason":
List concrete blockers or concerns raised during the meeting — e.g., budget freeze, leadership change.

6. Context for "Follow-up Date":
Update only if a renewal discussion or pricing review was scheduled with a specific date.

1. Suggest updates only for confirmed meeting topics.

2. Overwrite all text fields. Do not append.

3. "Engagement Notes" should be under 400 characters.

4. Context for "Partner Type":
Use CRM picklist values such as "Reseller", "Referral", or "Implementation Partner".

5. Context for "Engagement Notes":
Summarize any actions, enablement, or co-selling plans shared during the meeting.

6. Context for "Follow-up Date":
Only update if a joint sync or next enablement step was scheduled and mentioned by date.

1. Only suggest updates to fields if they were clearly discussed in the meeting.

2. Overwrite all text fields instead of appending.

3. The "Follow-up Details" field must not exceed 255 characters. Always overwrite.

4. Context for "Purchase Decision Behavior":
Use the correct CRM option to reflect the customer's purchasing authority. Refer to:
- "Deciding"
- "Deciding, no invoice"
- "Not deciding, no invoice"
- "Not deciding, only invoice"
- "Unknown"

5. Context for "Follow-up Date":
Update only if the meeting included a specific, date-based follow-up (e.g. “Next check-in on Sept 15”).

Common Use Cases for AI CRM Concierge

1. Automated Handover from Sales to CSx

Use CRM Concierge to streamline the transfer of key deal information from Sales to Customer Success. This ensures that critical context, such as blockers, stakeholders, or next steps, is consistently captured and synced into handover-specific CRM fields.

✅ Best practice: Set up an automation triggered by the “Customer Handover” meeting type, and include fields like “Implementation blockers”

❌ Pitfall: Avoid syncing to generic notes fields. CSx teams often miss important context if the data isn’t structured properly.

2. Automated Deal Scoring (e.g., MEDDIC)

Let the AI dynamically update deal scores based on what’s discussed in each call. By configuring a deal scoring field (like a MEDDIC score) and adding a helpful field prompt, you can ensure consistent scoring across reps and calls.

✅ Best practice: Add prompts like “Use call insights to calculate MEDDIC score (max 10)” to the CRM field description.

❌ Pitfall: Without this context, the AI may leave the field blank or fill it inconsistently.

3. Structured & Consistent Information Capture

Use CRM Concierge to consistently capture key data points, like KPIs, stakeholder roles, or blockers, directly into your CRM. This eliminates the need for scattered notes and ensures that your CRM becomes a true source of truth.

✅ Best practice: Configure automations that include fields like Stakeholders, Next Steps, and Timeline, each with short context prompts.

❌ Pitfall: Avoid relying on a single open-text “Summary” field. This makes reporting and coaching difficult later.

4. Continuous Data Enrichment

As meetings happen, the AI can enrich your CRM with additional details, such as job titles, product interests, or priorities, without overwriting existing data. This creates a living record that evolves over time.

✅ Best practice: Use Magic Mode to capture as much as possible, while excluding fields that must remain unchanged.

❌ Pitfall: If you don’t exclude sensitive fields (like pricing or internal tags), they may be unintentionally overwritten.

5. Custom Rules by Team or Meeting Type

Every team has different CRM needs. With CRM Concierge, you can create separate automations for Sales, CSx, or other departments, each with custom sync rules and AI Context.

✅ Best practice: Build department-specific automations. For example, Sales teams might sync deal stage and budget, while CSx focuses on go-live blockers and success metrics.

❌ Pitfall: Avoid using the same automation for all teams. This reduces relevance and increases the risk of syncing noise.

6. Smart Automation Routing with AI Meeting Context

When meeting names are unreliable, because reps don't follow naming conventions, or because invites come from tools like Outreach or Salesforce that name them automatically, you can use AI Meeting Context to trigger the right CRM automation based on what the meeting actually was, not what it was called.

✅ From your calendar description: If your sales tool (e.g., Outreach, Salesloft) writes the meeting type into the calendar invite description, you can add a condition like: "If the calendar invite description contains 'Discovery', apply the Discovery CRM fields." This makes your automation reliable even when reps use inconsistent titles.

✅ From your CRM event fields: If your CRM (e.g., Salesforce) auto-generates structured events with a Type field — like in a multi-step onboarding flow (Kickoff, Handover 1, Handover 2) — the AI can read that field to trigger the matching CRM update, without depending on what the rep named the meeting.

✅ By deal stage: The AI can check the opportunity stage in the CRM and apply different field sets — e.g., qualification fields for early-stage calls and pricing/timeline fields for late-stage ones.

✅ By customer segment: If your CRM tracks SMB vs. Enterprise accounts, the AI can route to segment-specific automations automatically — lighter field sets for SMB, more detailed for Enterprise.

✅ New vs. returning prospect: The AI can detect whether there's prior activity on the account in the CRM and apply a re-engagement template that focuses on what's changed, rather than a first-meeting format.

❌ Pitfall: Don't describe conditions that are too vague (e.g. "if it's an important call"). The more specific your natural language description, the more reliably the AI will route correctly.

How Our Team Uses the AI CRM Concierge

Play Vimeo video

Automations & Sync Modes

• Manual Approval: CRM suggestions are generated and sent via email. Users review and confirm the changes before they sync to the CRM.
• Automatic Sync: AI-suggested CRM updates are pushed directly into the CRM without review, freeing users to focus on meetings instead of updating records manually.
• Magic Mode: The AI fills as many relevant fields as possible automatically, providing richer data with minimal input.
• Strict Mode: Only the selected fields are analyzed and updated, ensuring tighter control over which data is changed.
• AI Context: Guide the AI’s logic by adding field descriptions or prompts in your CRM. This helps the AI extract and populate the right information.
• Access Control: Restrict who can view or edit each automation by team, department, or user to maintain data quality and security.
• AI Meeting Context: Instead of relying on meeting names or fixed filters, the AI reads the full meeting context, including the transcript and available CRM data, to understand what type of meeting just happened. Use this to route different CRM fields to different automations automatically. For example, describe in plain language when a rule should apply: "If this is a handover meeting" or "If the calendar description mentions Discovery."

What You’re Aiming For

• Clean, structured CRM entries without manual friction
• Visibility into how reps handle updates
• Data consistency to support coaching and forecasting

Best Practice: Weekly Sync Review

• Check what is being synced by checking your rep's meetings
• Spot-check updated deals for accuracy and consistency
• Use CRM Concierge data to drive coaching conversations (“What were the KPIs here?”)

Use AI CRM Concierge to Answer Strategic Questions

• Are we capturing blockers and success metrics consistently?
• Are reps overriding AI suggestions frequently?
• Which team has the most accurate CRM hygiene?

✅ Manager Actions (At a Glance)

• Spot-check key deals each week
• Guide reps in reviewing AI suggestions
• Use synced data in 1:1s or pipeline reviews

What You’ll Use It For

• Skip post-call manual CRM updates
• Keep deals updated with less time spent on admin
• Receive an AI-generated CRM update after every meeting

💡 Pro Tip: Don’t Skip the Review

Even if your team has auto-sync enabled, glance at the changes from time-to-time. Your CRM is only as accurate as what goes in.

✅ User Actions (At a Glance)

• Confirm or reject the AI suggestion
• Pick the correct deal or create one if needed
• Edit any field that looks off
• Leave feedback so Admins can improve automations