Unit Economics for Cloud Workloads

Knowing your total cloud bill is useless without knowing what generates that cost. Unit economics answers: “How much does it cost to serve one user?” “What is the cost per transaction?” “Which features are profitable?” This is the bridge between FinOps cost management and business decision-making.

---

What Are Unit Economics

Total cloud bill: $500,000/month
  → Meaningless without context

Unit economics:
  Cost per user:        $0.42/month   (1.2M active users)
  Cost per transaction: $0.003        (167M transactions/month)
  Cost per API call:    $0.0001       (5B calls/month)
  Cost per GB stored:   $0.023/month  (2.5PB data)
  
Why it matters:
  Revenue per user:     $12.99/month
  Cost per user:        $0.42/month
  Gross margin:         96.8%  ← Healthy
  
  But per feature:
  Core product:         $0.15/user   (cost efficient)
  Video feature:        $2.30/user   (expensive!)
  ML recommendations:   $0.85/user   (moderate)
  
  → Video feature costs 15x more per user than core product
  → Is it generating 15x more value?

---

Cost Attribution Model

class UnitCostCalculator:
    def __init__(self, billing_data, usage_data):
        self.billing = billing_data
        self.usage = usage_data
    
    def cost_per_user(self, period="month"):
        """Total infrastructure cost divided by active users."""
        total_cost = self.billing.total_cost(period)
        active_users = self.usage.active_users(period)
        return total_cost / active_users
    
    def cost_per_transaction(self, service_name):
        """Cost of a specific service divided by its transactions."""
        service_cost = self.billing.service_cost(service_name)
        transactions = self.usage.transaction_count(service_name)
        return service_cost / transactions
    
    def marginal_cost(self, additional_users):
        """Cost of adding N more users (incremental, not average)."""
        # Fixed costs don't increase with users
        fixed = self.billing.fixed_costs()  # DB licenses, reserved instances
        variable = self.billing.variable_costs()  # Compute, bandwidth
        
        current_users = self.usage.active_users()
        variable_per_user = variable / current_users
        
        return variable_per_user * additional_users
    
    def feature_cost_breakdown(self):
        """Cost per feature based on resource tagging."""
        features = {}
        for resource in self.billing.tagged_resources():
            feature = resource.tags.get("feature", "untagged")
            features[feature] = features.get(feature, 0) + resource.cost
        
        return {
            feature: {
                "total_cost": cost,
                "cost_per_user": cost / self.usage.feature_users(feature),
                "pct_of_total": cost / self.billing.total_cost(),
            }
            for feature, cost in features.items()
        }

---

Benchmarking

Industry benchmarks (approximate):

SaaS B2B:
  COGS (infrastructure) as % of revenue: 15-25%
  Cost per user/month: $0.50 - $5.00
  Gross margin target: 75-85%

E-commerce:
  Cost per transaction: $0.001 - $0.01
  Cost per page view: $0.00001 - $0.0001
  
Media/Streaming:
  Cost per stream-hour: $0.01 - $0.05
  Storage per user/month: $0.10 - $0.50

Fintech:
  Cost per payment: $0.005 - $0.05
  Compliance overhead: 10-20% of infrastructure cost

---

Executive Dashboard

unit_economics_dashboard:
  headline_metrics:
    - name: "Cost per Active User"
      value: "$0.42/month"
      trend: "-5% MoM"  # Improving
      target: "< $0.50"
    
    - name: "Infrastructure Gross Margin"
      value: "96.8%"
      trend: "+0.3% MoM"
      target: "> 95%"
    
    - name: "Cost per Transaction"
      value: "$0.003"
      trend: "Flat"
      target: "< $0.005"
  
  feature_breakdown:
    - feature: "Core Product"
      cost_per_user: "$0.15"
      revenue_per_user: "$8.99"
      margin: "98.3%"
    
    - feature: "Video"
      cost_per_user: "$2.30"
      revenue_per_user: "$3.00"  # Add-on
      margin: "23.3%"  # RED FLAG

---

Anti-Patterns

Anti-Pattern	Consequence	Fix
Track only total spend	Cannot optimize, no business context	Unit economics per user/feature/transaction
No resource tagging	Cannot attribute costs to features	Mandatory tagging in IaC templates
Average cost only	Hides expensive outliers	Track P50, P90, P99 cost per user
No marginal cost analysis	Over-provision for growth	Separate fixed vs variable costs
Finance and engineering silos	No shared cost language	Shared unit economics dashboard

Unit economics is the Rosetta Stone between engineering and finance. Engineers see requests, latency, and compute. Finance sees revenue, margin, and COGS. Unit economics translates between the two.