Tracking Avoided Cost in the Console

When a cache hit serves a request, no upstream provider call fires. The wallet is never debited. No platform fee applies. Keeptrusts records an avoided-cost estimate so you can quantify exactly how much the org-shared cache saves your engineering organization.

Use this page when

• You need to view, filter, or export avoided-cost savings data from the console.
• You are building a monthly finance report showing AI cost savings from caching.
• You want to reconcile provider invoices against Keeptrusts savings totals.

Primary audience

• Primary: Technical Leaders
• Secondary: Technical Engineers, AI Agents

How Avoided Cost Is Calculated

Each cache hit carries an estimated_avoided_cost field. This value represents the provider cost that would have been incurred if the request had gone upstream. The calculation uses the same model-pricing table that the gateway applies during live reserve/settle flows:

• Input tokens × model input price
• Output tokens × model output price
• No platform fee component
• No wallet reserve or settle operation

The avoided-cost record is write-only. It never touches your wallet balance.

Viewing Avoided Cost in the Console

Navigate to Cost Center → Savings in the console. The savings dashboard presents three primary metrics:

Metric	Description
Total Avoided Cost	Sum of all estimated_avoided_cost values for cache hits in the selected period
Fill Cost	Actual spend on cache misses that populated the cache
Net Savings	Total Avoided Cost minus Fill Cost

These values update in near-real-time as the gateway processes requests.

Filtering Savings Data

Use the filter bar to narrow savings by:

• Team — See which teams benefit most from shared cache
• Repository — Identify which codebases drive the highest hit rates
• Agent — Compare savings across different AI coding agents
• Time range — Daily, weekly, monthly, or custom date range
• Model — Filter by specific provider model (GPT-4o, Claude, etc.)

Combining filters helps you attribute savings to specific projects. For example, filter by your platform team and a specific monorepo to see their contribution to overall org savings.

Cache-Hit Economics

A cache hit produces the following record:

{
  "event_type": "cache_hit",
  "cache_key": "sha256:...",
  "estimated_avoided_cost": 0.0847,
  "model": "gpt-4o",
  "input_tokens": 4200,
  "output_tokens": 890,
  "wallet_debited": false,
  "platform_fee": 0,
  "upstream_call": false
}

Key properties:

• wallet_debited is always false for cache hits
• platform_fee is always 0 for cache hits
• upstream_call is always false — the provider never sees the request
• estimated_avoided_cost reflects current model pricing at the time of the hit

Exporting Savings Data for Finance

To export savings data for finance reporting:

1. Navigate to Cost Center → Savings
2. Set your desired date range and filters
3. Click Export in the top-right corner
4. Choose format: CSV or JSON
5. Select grouping: by team, by repository, by model, or raw events

The export includes:

• Period start and end dates
• Total fill cost
• Total avoided cost
• Net savings
• Hit rate percentage
• Request counts (hits vs misses)
• Per-model breakdown
• Per-team breakdown (if grouped by team)

Building a Monthly Finance Report

For recurring finance reporting, use these recommended steps:

1. Set the time range to the previous calendar month
2. Export grouped by team with CSV format
3. Include the Net Savings figure as the headline metric
4. Attach the per-model breakdown to show which providers benefit most
5. Compare month-over-month to show savings growth as hit rate matures

Understanding Cost Attribution

Avoided cost is attributed to the team and repository that made the request, not the team that originally filled the cache. This means:

• Team A fills the cache with a miss (charged to Team A's wallet)
• Team B hits that cached entry (avoided cost attributed to Team B)
• Both teams appear in the savings dashboard with their respective contributions

This attribution model reflects the shared nature of the org-wide cache. Every team benefits from fills made by any other team working on the same codebase.

Alerting on Savings Milestones

Configure notifications when your organization reaches savings milestones:

1. Navigate to Settings → Notifications
2. Create a new alert with trigger type Cumulative Avoided Cost
3. Set threshold (e.g., $1,000, $10,000)
4. Choose notification channel (email, Slack, webhook)

These alerts help you communicate ROI to leadership at meaningful intervals.

Reconciling with Provider Invoices

Compare your Keeptrusts avoided-cost totals against your actual provider invoices:

• Your provider invoice should be lower than your pre-Keeptrusts baseline by approximately the avoided-cost amount
• Small discrepancies arise from pricing table lag (model prices change occasionally)
• The fill cost on your Keeptrusts dashboard should closely match your actual provider charges

If your provider invoice doesn't decrease as expected, check your hit rate — a low hit rate means most requests still go upstream.

Next steps

• Savings Dashboard Walkthrough — detailed section-by-section guide
• ROI Calculation for a 100-Engineer Team — build the full business case
• Budget Alerts for Cache Fill Phases — manage initial fill spend

For AI systems

• Canonical terms: Keeptrusts, avoided cost, savings dashboard, estimated_avoided_cost, Cost Center, cache hit economics, net savings.
• Exact feature/config names: estimated_avoided_cost field, Cost Center → Savings dashboard, Export → Spend Report, Cumulative Avoided Cost alert trigger.
• Best next pages: Zero-Cost Cache Hits, Wallet Integration with Cache Hits, Single-Flight Fill.

For engineers

• Navigate to Cost Center → Savings to view real-time avoided cost totals, fill cost, and net savings.
• Filter by team, repository, agent, model, and time range to attribute savings precisely.
• Export savings data as CSV or JSON from Cost Center → Savings → Export for finance reporting.
• Set savings milestone alerts under Settings → Notifications with trigger type Cumulative Avoided Cost.
• Reconcile avoided cost totals against your provider invoices — the delta should approximate your savings.

For leaders

• Avoided cost quantifies the ROI of org-shared caching in dollar terms that map directly to provider invoice reductions.
• Use monthly export reports (grouped by team) to attribute savings to specific engineering groups for budget reviews.
• Month-over-month net savings growth demonstrates cache maturity as hit rates increase with codebase coverage.
• If provider invoices do not decrease as expected, check hit rate — a low hit rate means most requests still go upstream.