Cross Sell Target Identifier Agent Skill

Cross-Sell Target Identifier

When launching a new product or looking to expand product adoption, sales teams need to identify which existing customers are most likely to purchase. This skill analyzes the characteristics of successful customers for a given product, finds similar customers who don't own it yet, and provides prioritized recommendations with justification.

Instructions

Step 1: Identify the Target Product

When user asks "Which customers should I pitch Product X to?":

• Identify the Product:

SELECT productid, name, description, producttypecode, productstructure
FROM product
WHERE name LIKE '%[product name]%'
AND statecode = 0

• Confirm with user if multiple matches

Step 2: Analyze Successful Product X Customers

2.1 Find Customers Who Own Product X

Query won opportunities that included the target product:

SELECT op.opportunityid, op.customerid, op.accountid, op.actualvalue,
       op.actualclosedate, op.salesstage
FROM opportunity op
JOIN opportunityproduct opp ON op.opportunityid = opp.opportunityid
WHERE opp.productid = '[target_product_id]'
AND op.statecode = 1

Note: A product may appear in multiple opportunityproduct rows per opportunity. Deduplicate opportunityids programmatically after fetching results.

2.2 Build Success Profile from Winning Accounts

For each winning account, gather firmographic data:

SELECT accountid, name, industrycode, numberofemployees, revenue,
       customertypecode, address1_stateorprovince, address1_country,
       ownershipcode, createdon
FROM account
WHERE accountid IN ([list of winning account ids])

2.3 Analyze Success Patterns

Firmographic Analysis:

Calculate distribution across successful customers:
- Industry breakdown (industrycode): Which industries buy most?
- Company size (numberofemployees): What's the typical range?
- Revenue range: What's the typical revenue bracket?
- Geography (address1_stateorprovince/country): Regional concentrations?
- Customer type (customertypecode): Are they customers, partners, etc.?

Existing Product Ownership:

For each successful customer, identify other products owned:
SELECT a.accountid, a.name, p.name as product_name
FROM account a
JOIN opportunity o ON a.accountid = o.accountid
JOIN opportunityproduct op ON o.opportunityid = op.opportunityid
JOIN product p ON op.productid = p.productid
WHERE o.statecode = 1
AND a.accountid IN ([winning account ids])

Buying Pattern Analysis:

Identify patterns in successful deals:
- Average deal size for Product X
- Common bundled products
- Typical sales cycle length
- Time since becoming customer before purchasing Product X

Activity Pattern Analysis:

Review activities preceding successful deals:
- Types of engagement (calls, meetings, emails)
- Number of touchpoints before close
- Content/resources shared

Step 3: Generate Ideal Customer Profile (ICP)

Based on the analysis, create an Ideal Customer Profile:

IDEAL CUSTOMER PROFILE FOR [PRODUCT X]
════════════════════════════════════════════════════

FIRMOGRAPHIC CHARACTERISTICS:
- Industry: [Top 3 industries, e.g., "Financial Services (35%), Healthcare (28%), Technology (22%)"]
- Company Size: [Employee range, e.g., "100-500 employees (sweet spot)"]
- Revenue: [Revenue range, e.g., "$10M-$100M annual revenue"]
- Geography: [Regional patterns, e.g., "Primarily US, expanding to UK/EU"]

BEHAVIORAL INDICATORS:
- Already Owns: [Products commonly owned first, e.g., "80% have Product Y"]
- Customer Tenure: [Time as customer, e.g., "Typically 6-18 months as customer"]
- Recent Activity: [Engagement patterns, e.g., "High engagement with support/success"]
- Expansion Signals: [Growth indicators, e.g., "Recent hiring, new funding"]

BUYING PATTERNS:
- Average Deal Size: [$X]
- Typical Bundle: [Product X + Y + Z]
- Sales Cycle: [X days average]
- Common Champion: [Job title patterns]

SUCCESS INDICATORS FROM NOTES:
- Pain Points: [Common challenges mentioned]
- Use Cases: [How they use the product]
- Trigger Events: [What prompted purchase]

Step 4: Query Customer Base for Matches

4.1 Find Non-Owners of Product X

Note: Dataverse SQL does not support subqueries. Run two separate queries and exclude owners programmatically.

First, get all accounts that already own Product X (from Step 2.1 results — collect their accountids into a list).

Then query all active accounts:

SELECT a.accountid, a.name, a.industrycode, a.numberofemployees, a.revenue,
       a.customertypecode, a.address1_stateorprovince, a.createdon
FROM account a
WHERE a.statecode = 0

Filter out accounts whose accountid appears in the owner list programmatically after fetching.

4.2 Score Each Potential Target

For each non-owner account, calculate fit score:

Firmographic Fit (40%):

| Factor | Points | Scoring Logic |

|--------|--------|---------------|

| Industry Match | 0-15 | Exact match to top ICP industry = 15, Adjacent = 10, Other = 0 |

| Size Match | 0-15 | Within ICP range = 15, Close = 10, Outside = 5, Way off = 0 |

| Revenue Match | 0-10 | Within ICP range = 10, Close = 5, Outside = 0 |

Behavioral Fit (35%):

| Factor | Points | Scoring Logic |

|--------|--------|---------------|

| Owns Prerequisite Products | 0-15 | Has common prerequisite = 15, Related product = 10 |

| Customer Tenure | 0-10 | In ICP tenure range = 10, Close = 5 |

| Recent Engagement | 0-10 | High recent activity = 10, Moderate = 5, Low = 0 |

Buying Signals (25%):

| Factor | Points | Scoring Logic |

|--------|--------|---------------|

| Recent Purchases | 0-10 | Bought something in last 6 months = 10 |

| Expansion Behavior | 0-10 | Added users, upgraded = 10, Stable = 5 |

| Strategic Initiative Signals | 0-5 | Mentioned in notes/activities = 5 |

Step 5: Analyze Buying Signals

Important: Dataverse SQL Limitations

Dataverse SQL does NOT support: subqueries, DATEADD(), GETUTCDATE(), HAVING, DISTINCT, UNION, CASE statements, AVG on sentiment.

Calculate date filters programmatically before querying (e.g., '2025-09-01' for 6 months ago).

5.1 Recent Purchase Activity

SELECT a.accountid, a.name, o.opportunityid, o.name, o.actualclosedate, o.actualvalue
FROM account a
JOIN opportunity o ON a.accountid = o.accountid
WHERE o.statecode = 1
AND o.actualclosedate > '2025-09-01'

5.2 Expansion Indicators

Look for signals in activities and notes:

SELECT annotationid, objectid, subject, notetext, createdon
FROM annotation
WHERE objecttypecode = 'account'
AND createdon > '2025-09-01'

Keywords to Detect:

• Expansion: "growing", "scaling", "expanding", "new offices", "hiring"
• Strategic: "initiative", "project", "transformation", "migration"
• Pain: "struggling", "challenge", "problem", "need"
• Competition: "evaluating", "considering", "looking at"

5.3 Recent Cases (Support Indicators)

Query cases per account, then aggregate programmatically:

SELECT incidentid, customerid, createdon, prioritycode, msdyn_casesentiment
FROM incident
WHERE createdon >= '[6_months_ago]'

Group and count by customerid programmatically after fetching results.

High case volume could indicate:

• Active usage (good for expansion)
• Frustration (may need resolution first)
• Product limitations (potential for upsell to better solution)

Step 6: Rank and Present Target Accounts

Output Format:

CROSS-SELL TARGETS FOR [PRODUCT X]
════════════════════════════════════════════════════
Analysis Date: [Date]
Methodology: Compared against [N] successful Product X customers

SUMMARY:
- Total Eligible Accounts: [N]
- High Fit (Score 80+): [N] accounts
- Medium Fit (Score 60-79): [N] accounts
- Low Fit (Score <60): [N] accounts

════════════════════════════════════════════════════
TOP 10 CROSS-SELL TARGETS
════════════════════════════════════════════════════

1. CONTOSO CORPORATION
   Fit Score: 92/100
   ────────────────────────────────────────────────
   
   WHY THEY'RE A FIT:
   ✓ Industry: Financial Services (top ICP industry)
   ✓ Size: 350 employees (in sweet spot 100-500)
   ✓ Already Owns: Product Y, Product Z (common prerequisite)
   ✓ Customer Since: 14 months (optimal tenure range)
   ✓ Recent Activity: 8 touchpoints in last 30 days
   
   BUYING SIGNALS DETECTED:
   • Mentioned "scaling operations" in recent meeting notes
   • Purchased add-on licenses last month (expansion behavior)
   • Attended Product X webinar 2 weeks ago
   
   RECOMMENDED APPROACH:
   • Lead with [specific value prop based on industry]
   • Reference success story from [similar customer]
   • Contact: [Primary contact name and role]
   
   ESTIMATED DEAL SIZE: $45,000 (based on similar deals)

2. FABRIKAM INDUSTRIES
   Fit Score: 87/100
   ────────────────────────────────────────────────
   
   WHY THEY'RE A FIT:
   ✓ Industry: Manufacturing (adjacent to ICP)
   ✓ Size: 800 employees (slightly above sweet spot)
   ✓ Already Owns: Product Y
   ✓ High engagement with Customer Success
   
   BUYING SIGNALS DETECTED:
   • New CTO joined 3 months ago (leadership change)
   • Mentioned "digital transformation" in discovery call
   
   POTENTIAL CONCERNS:
   ⚠ Above typical company size - may need enterprise approach
   ⚠ No activity with Sales in last 60 days
   
   RECOMMENDED APPROACH:
   • Re-engage through Customer Success warm intro
   • Position as part of transformation initiative
   • Consider executive sponsor engagement

[Continue for top 10...]

Step 7: Create Action Items

Generate Follow-up Tasks:

For each top target, offer to create:

Use create_record with tablename: task
{
  "subject": "Cross-sell outreach: [Product X] to [Account Name]",
  "description": "Target identified as high fit for [Product X].\n\nFit Score: [X]/100\n\nKey talking points:\n- [Point 1]\n- [Point 2]\n\nContact: [Recommended contact]",
  "regardingobjectid": "[accountid]",
  "scheduledend": "[appropriate date]",
  "prioritycode": [based on fit score]
}

Update Account with Cross-Sell Flag:

Consider adding notes to account:

Use create_record with tablename: annotation
{
  "subject": "Cross-sell opportunity identified: [Product X]",
  "notetext": "[Summary of why they're a fit and recommended approach]",
  "objectid": "[accountid]",
  "objecttypecode": "account"
}

Dataverse Tables Used

| Table | Purpose |

|-------|---------|

| product | Identify target product |

| opportunity | Find won deals with target product |

| opportunityproduct | Link opportunities to products |

| account | Customer firmographic data |

| contact | Stakeholder information |

| activitypointer | Engagement history |

| annotation | Notes containing buying signals |

| incident | Support case patterns |

| task | Create follow-up tasks |

Key Fields Reference

product:

• productid (GUID) - Unique identifier
• name (NVARCHAR) - Product name
• productnumber (NVARCHAR) - SKU/product number
• producttypecode (CHOICE) - Sales Inventory(1), Misc Charges(2), Services(3), Flat Fees(4)
• productstructure (CHOICE) - Product(1), Family(2), Bundle(3)
• statecode (STATE) - Active(0), Retired(1), Draft(2), Under Revision(3)

opportunity:

• accountid (LOOKUP → account) - Related account
• statecode (STATE) - Open(0), Won(1), Lost(2)
• statuscode (STATUS) - In Progress(1), On Hold(2) [Open]; Won(3) [Won]; Canceled(4), Out-Sold(5) [Lost]
• actualvalue (MONEY) - Won deal value
• actualclosedate (DATE) - When deal closed
• originatingleadid (LOOKUP → lead) - Source lead

opportunityproduct:

• opportunityid (LOOKUP → opportunity) - Parent opportunity
• productid (LOOKUP → product) - Product in the deal
• quantity (DECIMAL) - Units sold
• priceperunit (MONEY) - Unit price
• extendedamount (MONEY) - Line total (calculated)
• manualdiscountamount (MONEY) - Line discount

account:

• industrycode (CHOICE) - Accounting(1), Agriculture(2), Broadcasting(3), Brokers(4), Building Supply(5), Business Services(6), Consulting(7), Consumer Services(8), etc.
• numberofemployees (INT) - Company size
• revenue (MONEY) - Annual revenue
• customertypecode (CHOICE) - Customer classification
• openrevenue (MONEY) - Total open pipeline value *(rollup field; availability depends on org configuration — query opportunity table directly if not present)*
• opendeals (INT) - Number of open opportunities *(rollup field; availability depends on org configuration)*

Cross-Sell Best Practices

• Start with success: Always analyze existing successful customers first
• Multi-factor matching: Don't rely on single criteria for fit
• Watch for timing: Recent purchases indicate budget availability
• Leverage relationships: Warm introductions beat cold outreach
• Segment recommendations: Different approaches for different segments
• Track results: Monitor conversion rates to refine the model

Examples

Example 1: Find Cross-Sell Targets for New Product

User says: "Who should I pitch our new Analytics Pro product to?"

Actions:

• Search product table for "Analytics Pro"
• Find accounts that have purchased Analytics Pro
• Build ideal customer profile from successful accounts
• Query non-owners matching the profile
• Rank by fit score and provide recommendations

Result:

IDEAL CUSTOMER PROFILE FOR ANALYTICS PRO:
- Industry: Financial Services (45%), Healthcare (30%)
- Size: 200-1000 employees
- Already owns: Platform Basic (80% correlation)

TOP 10 CROSS-SELL TARGETS:
1. Northwind Bank (92% fit) - Financial Services, 450 employees, owns Platform Basic
2. Alpine Health (87% fit) - Healthcare, 800 employees, recent support engagement

Example 2: Upsell Existing Customers

User says: "Which customers should upgrade to Enterprise tier?"

Actions:

• Identify customers on lower tiers
• Analyze Enterprise customers for common traits
• Find Standard tier customers matching Enterprise profile
• Factor in engagement and growth signals

Result:

UPGRADE CANDIDATES (Standard → Enterprise):
1. Contoso Ltd - Growing usage, added 50 users last quarter
2. Fabrikam Inc - Multiple support cases about feature limits
3. Tailspin Toys - Recent funding, headcount doubling

Example 3: Product Bundle Opportunity

User says: "Who bought Product A but not Product B?"

Actions:

• Find accounts with Product A in won opportunities
• Exclude accounts with Product B in won opportunities
• Rank by recency and relationship strength

Result:

PRODUCT A CUSTOMERS WITHOUT PRODUCT B:
- 23 accounts identified
- Average fit score: 78%
- Top 5 recommendations with outreach scripts provided

Troubleshooting

Error: Product not found

Cause: Product name doesn't match exactly or product is inactive

Solution:

• Search with partial name match (LIKE '%name%')
• Check statecode to include active products only
• List available products for user to select

Error: No customers found for target product

Cause: New product with no sales history yet

Solution:

• Use similar product's customer base as proxy
• Define ideal customer profile manually
• Start with industry/size matching only

Error: Too few differentiation signals

Cause: Winning customers too diverse for clear pattern

Solution:

• Increase sample size (longer date range)
• Focus on top-performing accounts only
• Add firmographic filters (industry, size)