Required deflection rate
42.8%Step 1 Provider and Model
Switch model assumptions using prices from the selected snapshot.Step 2 Support Scenario
Set ticket volume, human baseline, and escalation assumptions before tuning the AI workload.Mid-volume SaaS support queue with moderate retrieval depth and tighter per-ticket savings headroom.
- Need help sizing assumptions? Read the support cost guide and cache-hit guidance.
Step 3 AI Workload Assumptions
Set the support-turn assumptions first, then open advanced retrieval and infra fields if needed.Show advanced inputs
Open these only when retrieval, reranking, vector, or infra detail changes the support decision.
Save or Share Scenario URL
Copy this link to reopen or share these exact assumptions later. Pricing uses the published snapshot shown on the page.Save it for later or paste it into ChatGPT or Claude to discuss this scenario.
Decision Signal
Above thresholdModeled support deflection needs to reach 42.8% to beat the $2.4 human baseline.
Your current deflection assumption is +11.2% above the modeled threshold.
Required deflection
42.8%Assumed deflection
54.0%Gap vs threshold
+11.2%Blended cost / ticket
$1.9843Most Sensitive Threshold Inputs
Most sensitive threshold inputs when each is moved up by 10%.Totals
Summary metrics for the current support scenario.Assumed deflection rate
54.0%Gap vs threshold
+11.2%AI cost / deflected ticket
$0.2806All-escalated cost / ticket
$3.9842Blended cost / ticket
$1.9843Monthly savings vs human
+$914.59| Required deflection rate | 42.8% |
| Assumed deflection rate | 54.0% |
| Gap vs threshold | +11.2% |
| AI cost / deflected ticket | $0.2806 |
| All-escalated cost / ticket | $3.9842 |
| Blended cost / ticket | $1.9843 |
| Monthly savings vs human | +$914.59 |
Blended Cost Breakdown (USD/ticket)
Each cost component is computed independently, then summed.Largest cost block: Human escalation.
CacheSavings from repeated support requests served by cache. Negative means lower total cost.
$-0.031Embeddings IngestionAmortized per-ticket share of the fixed monthly support corpus embedding refresh.
$0GenerationModel input/output token spend for support turns.
$0.0365InfraNon-model infra overhead per support request.
$0.0057RerankingReranker cost based on docs rescored per support request.
$0.1661RetrievalExtra model input spend from retrieved support context.
$0.0128Vector DbVector database query cost across support requests.
$0.0001Human escalationHuman handling cost applied only to the escalated share of tickets.
$1.794| CacheSavings from repeated support requests served by cache. Negative means lower total cost. | $-0.031 |
| Embeddings IngestionAmortized per-ticket share of the fixed monthly support corpus embedding refresh. | $0 |
| GenerationModel input/output token spend for support turns. | $0.0365 |
| InfraNon-model infra overhead per support request. | $0.0057 |
| RerankingReranker cost based on docs rescored per support request. | $0.1661 |
| RetrievalExtra model input spend from retrieved support context. | $0.0128 |
| Vector DbVector database query cost across support requests. | $0.0001 |
| Human escalationHuman handling cost applied only to the escalated share of tickets. | $1.794 |
Sensitivity RankingChange in required deflection rate, in percentage points, when one variable is increased by 10%.
| Variable | Delta required deflection (pts) |
|---|---|
| Human cost / ticketCurrent fully loaded human handling cost per resolved ticket. | -6.5% |
| Human cost / escalated ticketHuman handling cost on escalated tickets only. | +5.5% |
| Rerank DocsDocs reranked per support request. | +0.3% |
| AI requests / ticketAverage AI turns needed to fully resolve a ticket. | +0.3% |
| AI requests before handoffAI turns spent before a ticket escalates to a human. | +0.1% |
| Cache Hit RateFraction of support requests served from cache. | -0.1% |
| Output TokensGenerated tokens per support response. | +0.0% |
| Input TokensBase prompt tokens per support request. | +0.0% |
| Retrieved ChunksRetrieved chunk count per support request. | +0.0% |
| Tokens Per ChunkAverage chunk size in retrieved support context. | +0.0% |
| Infra Cost Per Request | +0.0% |
| Vector Queries Per RequestVector lookup count per support request. | +0.0% |
| Vector Cost Per Query | +0.0% |
| Embedding Ingestion Tokens Monthly | +0.0% |
| Monthly ticketsTicket volume used for monthly totals and fixed-term amortization. | -0.0% |
Assumptions and Units
Explicit assumptions keep the support model reproducible and auditable.- CurrencyUSD
- Unit of analysisResolved support ticket
- Pricing snapshot2026-03-19
- Selected model rowOpenAI/GPT-5.2
- Monthly tickets2,200 tickets for monthly totals and fixed-term amortization
- Embedding refreshAmortized across monthly tickets as a fixed support corpus refresh term
- Deflection definitionShare of tickets fully resolved without human handoff
- Escalation definition$3.9 human handling cost on escalated tickets only
- AI before handoff2.1 AI requests on escalated tickets before human takeover
Recommended Next Step
Use these links to move from modeled support economics into the next pricing or implementation check.If the threshold looks realistic, translate it into a blended support cost at your expected escalation mix.
Sources and Snapshot
Pricing comes from a daily snapshot generated by batch workflows.Active Pricing Row
Selected model
OpenAI / GPT-5.2
- Input tokens$1.75 / 1M
- Output tokens$14 / 1M
Shared retrieval defaults
- Embedding input$0.02 / 1M
- Rerank docs$1 / 1K
- Snapshot date: 2026-03-19
- Source links and update notes: Pricing Snapshot Reference
Continue Analysis
Move to the next tool or guide without losing your current scenario.Switch tools
Read guides