Chatbot Deflection Rate Calculator

Work out chatbot deflection rate instantly with clear inputs, formula shown and shareable results.

About Chatbot Deflection Rate Calculator

Containment is the share of conversations closed without escalation to a human, and cost avoided is that count times the fully loaded cost of an agent contact. The metric is easy to game: abandoned conversations and users who give up and phone instead both count as contained unless you check resolution separately, so containment should always be reported alongside a satisfaction or repeat-contact measure.

Formula

Deflection

deflection rate = resolved without agent / total conversations; cost avoided = deflected x cost per agent contact

Worked example

With the default values below, Chatbot Deflection Rate Calculator returns Deflection / containment rate of 65.00%. Change any input above and every figure here is recalculated in your browser.

Inputs used

Conversations per monthminimum 1: 120,000
Conversations resolved without an agentminimum 0: 78,000
Cost per agent contactminimum 0: 6.5
Containment targetaccepts 1–100 %: 70 %

Results

Deflection / containment rate: 65.00%
Agent contacts avoided: 78,000
Monthly cost avoided: $507,000.00
Gap to the containment target: 5.00%

Step by step

1Values used
2Deflection
3Deflection / containment rate= 65.00
4Agent contacts avoided= 78,000
5Monthly cost avoided= 507,000.00
6Gap to the containment target= 5.00

Assumptions and limits

The result is derived only from the 4 inputs listed above. Anything not among them is not part of the calculation.
In particular, this calculator does not take inflation, tax or compounding frequency into account.
Figures are rounded for display to 2 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Frequently Asked Questions

How do I avoid overstating deflection?

Exclude abandoned sessions, and track repeat contacts within 24 hours. A conversation that ends because the user gave up is a failure counted as a success.

What containment is realistic?

Well-scoped FAQ and account-status use cases reach 60 to 80 percent. Broad open-ended support is much lower, and pushing the number by refusing to escalate destroys satisfaction.