Skip to main content

Cost Management & Forecasting

Track API spending, forecast future costs, and optimize your budget with comprehensive cost management tools.

Overview

Cost Management helps you:

  • Track costs per query, provider, and user
  • Forecast spending for budget planning
  • Set budgets and receive alerts
  • Analyze trends to identify cost patterns
  • Optimize spending with data-driven insights

Accessing Cost Management

Navigate to Cost Management:

Dashboard → Admin → Costs
URL: /dashboard/admin/costs

Note: Available to Admin and Editor roles only

Cost Dashboard

Overview Metrics

Top Cards Display:

┌─────────────────────┐  ┌─────────────────────┐  ┌─────────────────────┐
│ Total Spend │ │ This Month │ │ Avg per Query │
│ $24.67 │ │ $8.32 │ │ $0.0047 │
│ (All time) │ │ (Oct 2025) │ │ (5,245 queries) │
└─────────────────────┘ └─────────────────────┘ └─────────────────────┘

┌─────────────────────────────────────────────┐
│ Budget Status │
│ $8.32 / $50.00 (16.6%) │
│ ████░░░░░░░░░░░░░░░░░░░░░░░░░ │
└─────────────────────────────────────────────┘

Metrics Explained:

  • Total Spend: All-time API costs across all providers
  • This Month: Current calendar month spend
  • Avg per Query: Total spend ÷ total queries
  • Budget Status: Current vs monthly limit with visual meter

View Daily/Weekly/Monthly Trends:

Daily Trend Example:

Daily Cost Trend - Last 30 Days

Cost (USD)

1.50│ ●
│ ● ●
1.00│ ● ●
│ ● ●
0.50│ ● ●
│ ● ●
0.00└─────────────────────────────────→ Date
Oct 1 5 10 15 20 25 30

Trend Insights:

  • Spending patterns by day of week
  • Monthly growth rate
  • Seasonal variations
  • Anomaly detection (unusual spikes)

Provider Breakdown Tab

Cost by Provider:

Provider Cost Analysis

Claude (Anthropic)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 2,145
Cost: $6.43 (26.1% of total)
Avg: $0.003 per query
Trend: Stable ━

GPT-4 (OpenAI)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 1,892
Cost: $18.92 (76.7% of total)
Avg: $0.010 per query
Trend: Increasing ↗

Gemini (Google AI)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 856
Cost: $0.00 (Free tier)
Avg: $0.000 per query
Trend: Stable ━

Perplexity
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 352
Cost: $0.35 (1.4% of total)
Avg: $0.001 per query
Trend: Stable ━

Provider Insights:

  • Which providers cost most
  • Cost per query comparison
  • Volume vs cost tradeoffs
  • Trend direction (↗ ↘ ━)

User Cost Tab

Cost by Team Member:

User Cost Breakdown

john@example.com (Admin)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 3,245
Cost: $14.67
% of Total: 59.4%

sarah@example.com (Editor)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 1,456
Cost: $7.23
% of Total: 29.3%

mike@example.com (Analyst)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries: 544
Cost: $2.77
% of Total: 11.2%

Use Cases:

  • Track team usage patterns
  • Identify heavy users
  • Allocate costs internally
  • Plan per-user budgets

Cost Forecasting

Forecast Tab

Predictive Cost Analysis:

30-Day Cost Forecast

Method: Exponential Smoothing
Confidence: 85%

Current Spend: $8.32 (Oct 1-9)
Projected (Oct 31): $27.50 ± $4.20

Forecast:
$35 │ ●─── (Upper Bound)
│ ●───●
$30 │ ●───● ●─── (Forecast)
│ ●───●
$25 │ ●───● ●─── (Lower Bound)
│ ●───●
$20 │ ●───●
└─────────────────────────────────→ Date
Oct 9 15 20 25 30

Forecast Methods:

Linear Regression:

  • Simple straight-line projection
  • Good for steady growth
  • Fast calculation

Exponential Smoothing:

  • Accounts for recent trends
  • Better for changing patterns
  • More accurate for short-term

Moving Average:

  • Smooths out fluctuations
  • Good for stable patterns
  • Less reactive to anomalies

Forecast Configuration

Adjust Forecast Parameters:

Lookback Period:

  • 7 days: Recent trends (volatile)
  • 14 days: Balanced (recommended)
  • 30 days: Long-term patterns (stable)
  • 90 days: Seasonal patterns

Forecast Horizon:

  • 7 days: Short-term planning
  • 30 days: Monthly budget (recommended)
  • 90 days: Quarterly planning

Confidence Intervals:

  • 68% (±1 σ): Narrow range
  • 85% (±1.5 σ): Balanced (recommended)
  • 95% (±2 σ): Wide range

Example Configuration:

Forecast Settings:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Method: Exponential Smoothing
Lookback: 14 days
Horizon: 30 days
Confidence: 85%

[Update Forecast]

Forecast Accuracy

Accuracy Metrics:

Forecast Performance

Mean Absolute Error (MAE): $1.23
Root Mean Square Error (RMSE): $1.89
Accuracy: 94.2%

Historical Forecasts vs Actual:

Forecast Actual Difference
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Sep 2025: $22.50 $23.10 +2.7%
Aug 2025: $18.30 $17.85 -2.5%
Jul 2025: $15.20 $16.10 +5.9%

Average Accuracy: 96.3% ✅

Improving Accuracy:

  • Use longer lookback for stable usage
  • Use exponential smoothing for changing patterns
  • Review forecast weekly
  • Adjust method if accuracy <90%

What-If Scenarios

Scenario Planning:

Scenario Analysis

Base Case (Current Usage)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries/day: 175
Projected (30d): $27.50

Scenario 1: Double Query Volume
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries/day: 350 (+100%)
Projected (30d): $55.00 (+100%)

Scenario 2: Use Only Free Tier (Gemini)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries/day: 175
Projected (30d): $0.00 (100% savings)

Scenario 3: Replace GPT-4 with Claude
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Queries/day: 175
Projected (30d): $12.40 (-55% cost)

Use Cases:

  • Budget planning for growth
  • Cost optimization analysis
  • Provider switching decisions
  • Campaign budget forecasting

Provider Cost Comparison

Cost Per Query

Comparative Analysis:

Cost per Query by Provider

Provider Cost/Query Est. 1000 Queries Quality Score
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
GPT-4 $0.0100 $10.00 ⭐⭐⭐⭐⭐ (9.2/10)
Claude $0.0030 $3.00 ⭐⭐⭐⭐⭐ (9.0/10)
Perplexity $0.0010 $1.00 ⭐⭐⭐⭐ (7.8/10)
Gemini $0.0000 $0.00 ⭐⭐⭐ (6.5/10)

Quality Score Based On:

  • Citation count per response
  • Citation relevance
  • Response depth
  • Response accuracy

Cost Efficiency Ranking

Best Value Providers:

Rank 1: Perplexity

  • $0.001 per query
  • 5.0 avg citations
  • Citations per dollar: 5,000
  • Best for: High-volume testing

Rank 2: Claude

  • $0.003 per query
  • 6.3 avg citations
  • Citations per dollar: 2,100
  • Best for: Balanced quality/cost

Rank 3: GPT-4

  • $0.010 per query
  • 6.0 avg citations
  • Citations per dollar: 600
  • Best for: High-value analysis

Rank 4: Gemini

  • $0.000 per query (free tier)
  • 4.4 avg citations
  • Citations per dollar: ∞ (free)
  • Best for: Budget-conscious volume

Provider Switching Savings

Calculate Potential Savings:

Provider Switch Analysis

Current Mix:
- Claude: 40% of queries ($6.43)
- GPT-4: 35% of queries ($18.92)
- Gemini: 15% of queries ($0.00)
- Perplexity: 10% of queries ($0.35)

Total Current: $25.70/month

Optimized Mix (Same Quality):
- Claude: 60% of queries (+20%)
- Perplexity: 30% of queries (+20%)
- Gemini: 10% of queries (-5%)
- GPT-4: 0% of queries (-35%)

Total Optimized: $12.85/month

Savings: $12.85/month (50% reduction) ✅

Cost Optimization Recommendations

Automatic Insights

AI-Generated Recommendations:

Cost Optimization Opportunities

1. Replace GPT-4 with Claude for Standard Queries
Impact: -$12.50/month (48% savings)
Quality Impact: Minimal (0.2 citation difference)
Confidence: High (95%)
[Apply Recommendation]

2. Use Gemini for High-Volume Testing
Impact: -$3.20/month (12% savings)
Quality Impact: Moderate (lower citation rate)
Confidence: Medium (70%)
[Apply Recommendation]

3. Batch Similar Queries
Impact: -$1.80/month (7% savings)
Reduces duplicate queries
Confidence: High (90%)
[Apply Recommendation]

Total Potential Savings: $17.50/month (68%) ✅

Manual Optimization Strategies

Strategy 1: Provider Tiering

High-Value Queries (GPT-4 or Claude):

  • Competitive analysis
  • Client deliverables
  • Strategic decisions
  • Cost: $0.003-0.010/query

Standard Queries (Claude or Perplexity):

  • Content research
  • Regular monitoring
  • Internal analysis
  • Cost: $0.001-0.003/query

High-Volume Testing (Gemini or Perplexity):

  • Query discovery
  • Experimentation
  • Trend testing
  • Cost: $0.000-0.001/query

Strategy 2: Batch Processing

Before Optimization:

Execute 100 similar queries individually
Cost: $0.30 (Claude × 100)

After Optimization:

Batch 100 similar queries
- Detect duplicates: 15 queries
- Execute unique: 85 queries
Cost: $0.26 (15% savings)

Strategy 3: Caching Results

Coming Soon: Automatic result caching

Future Feature:

  • Cache query results for 7 days
  • Re-use for identical queries
  • Savings: 20-40% for repeat queries

Budget Management Integration

View Budget Status

Budget Meter in Cost Dashboard:

Monthly Budget Status

$8.32 / $50.00
████░░░░░░░░░░░░░░░░░░ 16.6%

Days Remaining: 22
Avg Daily Spend: $0.92
Projected Month-End: $28.50

Status: ✅ On Track (57% budget remaining)

Status Indicators:

  • On Track: <50% budget used, >50% month remaining
  • ⚠️ Warning: >50% budget used, >50% month remaining
  • 🔴 Over Budget: >100% budget used

Budget Alerts in Cost View

Alert Integration:

⚠️ Budget Alert

You've reached 80% of your monthly budget.
Current: $40.00 / $50.00

Projected Month-End: $52.50 (Over budget by $2.50)

Actions:
→ Review high-cost queries
→ Switch to lower-cost providers
→ Pause non-essential queries
→ Increase budget limit

[View Budget Settings]

See: Budget Management Guide


Cost Tracking Details

Query-Level Costs

View Individual Query Costs:

Recent Queries - Cost View

Query Provider Cost Date
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
"best seo tools" GPT-4 $0.012 Oct 9
"content optimization tips" Claude $0.003 Oct 9
"project management software" Perplexity $0.001 Oct 8
"free cms platforms" Gemini $0.000 Oct 8

Sort by Cost:

  • Identify most expensive queries
  • Find cost optimization opportunities
  • Budget allocation planning

Export Cost Data

CSV Export:

query_id,query_text,provider,cost_usd,executed_at,user_id
uuid-1,"best seo tools",gpt4,0.012,2025-10-09T10:23:00Z,user-1
uuid-2,"content tips",claude,0.003,2025-10-09T09:15:00Z,user-1
uuid-3,"pm software",perplexity,0.001,2025-10-08T14:30:00Z,user-2

Use Cases:

  • Accounting reconciliation
  • Internal cost allocation
  • Tax reporting
  • Custom analysis in Excel

Cost Analytics Best Practices

Regular Monitoring

Daily (for high-volume users):

  • Check daily spend
  • Monitor budget status
  • Review any anomalies

Weekly (recommended):

  • Review cost trends
  • Check forecast accuracy
  • Analyze provider mix
  • Optimize based on insights

Monthly:

  • Deep-dive analysis
  • Budget vs actual review
  • Provider performance comparison
  • Strategy adjustment

Cost Control Strategies

Set Clear Budgets:

  • Monthly organization budget
  • Per-user limits (coming soon)
  • Per-provider caps (coming soon)

Use Alerts:

  • 50% budget threshold
  • 80% budget threshold
  • 100% budget (pause queries)

Optimize Provider Mix:

  • Use GPT-4 sparingly (highest cost)
  • Favor Claude for most queries
  • Use Gemini for volume testing
  • Leverage Perplexity for cost efficiency

Eliminate Waste:

  • Remove duplicate queries
  • Pause low-value queries
  • Batch similar queries
  • Cache results (coming soon)

Cost Attribution

By Project:

  • Tag queries with project codes
  • Track project-level costs
  • Allocate budgets per project
  • ROI calculation per project

By Department:

  • User-based cost allocation
  • Department budget tracking
  • Chargeback reporting

By Campaign:

  • Campaign-specific query sets
  • Cost per campaign tracking
  • Campaign ROI analysis

Advanced Cost Features

Programmatic Access:

GET /api/v1/costs/trends?period=30d

Response:

{
"total_cost": 24.67,
"period_cost": 8.32,
"daily_average": 0.92,
"forecast_30d": 27.50,
"trends": [
{"date": "2025-10-01", "cost": 0.45},
{"date": "2025-10-02", "cost": 1.23}
]
}

Custom Cost Reports (Phase 18)

Automated Reports:

  • Weekly cost summary emails
  • Monthly budget reports
  • Cost forecasting reports
  • Provider comparison reports

See: Scheduled Reports


Cost Optimization Automation (Coming Soon)

Auto-Optimization Features:

  • Automatic provider selection (lowest cost)
  • Query result caching
  • Duplicate query detection
  • Budget-based query queuing

Target Savings: 30-50% cost reduction


Troubleshooting

Cost Discrepancies

Problem: Dashboard cost ≠ Provider invoices

Common Causes:

  1. Timing differences

    • Dashboard: Execution timestamp
    • Provider: Billing timestamp
    • Solution: Compare monthly totals
  2. Rounding

    • Small rounding differences per query
    • Normal: ±5% variance
    • Solution: Accept minor differences
  3. Estimation vs Actual

    • Some costs are estimated
    • Actual costs from provider API
    • Solution: Wait for final billing

Forecast Inaccuracy

Problem: Forecast far from actual

Solutions:

  1. Increase lookback period (more data)
  2. Change forecast method (try exponential)
  3. Wait for more data (need 14+ days min)
  4. Check for anomalies in historical data

Missing Cost Data

Problem: Queries showing $0.00 cost

Causes:

  • Gemini queries (free tier)
  • Failed queries (not charged)
  • Cost calculation error

Solutions:

  1. Verify query executed successfully
  2. Check provider response logs
  3. Re-execute if needed
  4. Contact support if persistent


FAQs

Q: How accurate is cost tracking? A: Within ±5% of actual provider costs. Variances due to rounding and timing.

Q: Can I set per-user budgets? A: Coming soon. Currently organization-level only.

Q: Which provider is most cost-effective? A: Perplexity for volume ($0.001/query), Claude for balanced quality/cost ($0.003/query).

Q: How often are forecasts updated? A: Real-time. Forecast updates as new queries execute.

Q: Can I export cost data for accounting? A: Yes. CSV export available with all transaction details.

Q: Do failed queries cost money? A: No. Only successful responses are charged.

Q: How can I reduce costs by 50%? A: Switch from GPT-4 to Claude, use Gemini for testing, eliminate duplicates.


Last Updated: October 9, 2025 Feature Status: Fully Implemented