Estimate the unit economics for support conversations using workload-specific operating assumptions. Adjust the inputs to match measured production behavior, then use the primary estimate and supporting metrics to compare deployment choices, identify constraints, and document a repeatable planning baseline. The calculator runs locally in your browser and does not send operational data to a server.
The main result is the direct variable cost for one successful outcome. Attempt cost separates raw execution expense, token cost isolates model usage, and failure cost shows the uplift caused by incomplete attempts.
Exclude fixed engineering and staffing costs only if you intentionally want a variable-cost view. Enter current provider prices yourself because pricing and cache rules can change.
For 900 input tokens at $2.50/M, 300 output tokens at $10/M, $0.015 tool cost, and 92% completion:
Add the expected escalation cost to tool and platform cost when you want a blended operational estimate rather than model-only cost.
Use a blended input price based on the cached and uncached share, or calculate the two portions separately before entering the combined amount.
Attempts that fail still incur cost, so their expense must be distributed across the smaller number of successful outcomes.
Yes. Use zero when external APIs, search, CRM, ticketing, or platform fees are not charged per task or are excluded.
Not automatically. Allocate fixed subscriptions across expected successful tasks and add that amount if you need fully loaded unit cost.
| Variable | Meaning |
|---|---|
| Input/output tokens | Average tokens consumed per attempt |
| Prices per 1M tokens | Current model rates entered by user |
| Tool cost | Variable non-token expense per attempt |
| Completion rate | Share of attempts that become successful tasks |