Token-to-ROI calculator

Enter one queue's numbers to see what an AI agent costs per resolved contact against a person, where the money goes, and how fast it pays back. Everything runs in your browser and nothing is sent anywhere.

Agent vs human

Cost per resolved contact

Agent cost includes tokens, platform, human escalations, review and repeat contacts.

Person

Agent (on suitable contacts)

Saving per month

Saving per year

Payback on setup

Year-one ROI

Your queue

Use actuals from the last three months where you can.

Currency

Contacts per month

Cost per seat per monthFully loaded: pay, on-costs, attrition, training, QA, tech, management.

Contacts handled per seat per month

Share of volume suitable for an agent (%)Repetitive, rules-based, low-risk contact types.

Agent token and platform cost

Defaults are illustrative. Replace them with pilot data or your vendor's usage bill.

Input tokens per attemptPrompt, context and tool results across every step.

Output tokens per attempt

Price per million input tokens

Price per million output tokens

Retry and loop multiplier1 means no retries. Looping agents can run 2 to 5 times.

Platform cost per attemptOrchestration, retrieval, monitoring, telephony.

Handoff and quality

This is where agent ROI is usually won or lost.

Escalation rate to a person (%)

Escalated contact cost vs normal (x)Escalations are harder and arrive with a frustrated customer.

Repeat contact rate after agent "resolution" (%)

Agent-resolved contacts reviewed by a person (%)

Review cost as % of a person-handled contact

Investment

Monthly platform, licence and support

One-off setup, integration and tuning

What if escalation or looping changes?

Agent cost per contact at different escalation rates (rows) and retry multipliers (columns), other inputs unchanged. Green is at least 20% cheaper than a person, amber is within 20%, red is dearer.

What to gather for the full health check

Three months of volumes by contact type, with peaks by hour or day.
Handle time, first-contact resolution, CSAT and rework by contact type.
Fully loaded cost per seat, or the outsourcer's rate card.
Any current chatbot or AI spend, including usage or token bills.
20 to 30 anonymised transcripts for each of the top contact types.
The SOPs or knowledge base behind those contact types.
Compliance limits such as GDPR, FCA and data residency.
A named sponsor and one hour for an interview.

No system access or customer personal data is needed. To run this on your own queue, book a Health Check.

Want this run on your real numbers?

A Token-to-ROI Health Check takes two weeks and gives you a clear verdict: scale it, fix it or stop it.

Book a Health Check