Cost per call

In one sentence

Cost per call is the total cost of handling one customer interaction, including agent labor, technology and overhead, and it is the headline figure in every automation business case even though an honest one has to include the interactions the automation handles badly.

USD 7 to 12 vs USD 0.40 Widely cited industry range, 2025Last reviewed 4 August 2026

Not to be confused with Cost per resolution.

Definition

Cost per call is what it costs you to deal with one customer getting in touch, counting labor, technology and overhead.

It is the number in the board slide. The comparison against automation looks dramatic, and it is real, but the honest version includes the calls the system gets wrong and hands to a person anyway, which you then pay for twice.

Cost per call is straightforward to define and frequently calculated in ways that overstate savings. The difference between the flattering number and the credible one is almost entirely about what gets left out.

What a complete calculation includes

  • Agent labor, including salary, benefits and employer costs.
  • Non-productive time. Agents are not on calls for their entire shift, so occupancy has to be factored in.
  • Training, recruitment and attrition costs, which are substantial in contact centers.
  • Technology, telephony and infrastructure.
  • Supervision, quality assurance and workforce management.
  • Facilities and overhead.

What is frequently omitted, inflating apparent savings

  • The cost of interactions the automation handles badly, which return as repeat contacts.
  • The cost of building and maintaining the automation, including knowledge base upkeep.
  • Escalated interactions, which cost the automation attempt plus the full human handling.

The correct comparison

  • Not automated cost against human cost per interaction.
  • But total cost of resolving a given volume of customer need, across both paths, including failures.
  • That is cost per resolution rather than cost per call, and it is the honest measure.

The widely quoted range, and how to use it

  • A human-handled call at roughly USD 7 to USD 12 against roughly USD 0.40 for an agent-handled call is the figure that appears in nearly every business case.
  • It is an industry range rather than an audited figure, and it varies enormously by geography. Indian contact center labor costs differ from US costs by a large multiple.
  • Any real business case should substitute local labor cost rather than using the published range.

Common misconception

That the cost per call comparison settles the automation business case. It establishes the direction. Cost per resolution, including failures and escalations, establishes the magnitude, and the two frequently disagree.

Why it matters commercially

This is the number in the board slide. Presenting it honestly, with the escalation and maintenance costs included, produces a smaller number and a business case that survives contact with reality. A projection built on the published range and nothing else tends to be met with disappointment.

In voice specifically

A voice interaction runs speech recognition and speech synthesis that a text chat never touches, so an agent-handled call is not the near-free line the USD 0.40 figure can imply. The comparison is still favorable against a human call, but the per-call figure for voice carries a stack that text does not.

Where AsqVox fits

A website conversation frequently has no human alternative at all. The visitor with an unanswered question leaves without contacting anyone, so the comparison is not cost saving but revenue that would otherwise not exist. AsqVox logs each conversation with its transcript and sentiment, which is the raw material for checking whether a contained call was genuinely resolved or merely ended.

Visual

The comparison everyone makes, and the one that is true

The comparison everyone makes, and the one that is trueUSD 7, human call floorUSD 12, human call ceilingUSD 0.40, the flattering agent-only figuresequential210 msEscalations paying both coststhe automation attempt plus full human handling

Cost per call sets the direction. Cost per resolution sets the number.

The bar is drawn in US cents and is illustrative above the agent attempt: the source publishes the human band (USD 7 to 12), the agent figure (~USD 0.40) and the all-in per-minute range, not per-component escalation figures, so those widths name what the line is rather than a measured amount. The point is the ordering. The flattering figure (marker, ~USD 0.40) sits below the honest automated resolve stack, which still sits well below the human band (thresholds, USD 7 to 12). Smaller saving, credible number. The published range is US-weighted; Indian labor cost differs by a large multiple, so substitute local cost.

Statistics

Every figure carries its source and year. Vendor numbers are labelled as vendor numbers, and where no reliable figure exists this page says so rather than borrowing one.

A human-handled call costs roughly USD 7 to USD 12 against roughly USD 0.40 for an agent-handled call.

USD 7 to 12 vs USD 0.40industry range

Widely cited industry range, 2025 - An industry range rather than an audited figure, varying substantially by geography and complexity. Indian contact center labor differs from US costs by a large multiple, so a real business case should substitute local labor cost.

All-in production voice agent cost lands around USD 0.11 to USD 0.33 per minute, once recognition, the model, synthesis and telephony are stacked.

USD 0.11 to 0.33 per minindustry range

Industry-reported production cost range, 2026 - The delivered cost, not the platform layer. Where a deployment lands is driven mostly by conversation duration and prompt size.

Platform base rates start around USD 0.05 per minute for Vapi, USD 0.07 for Retell and USD 0.09 for Bland.

USD 0.05 to 0.09 per minvendor claim

Vendor published pricing (Vapi, Retell, Bland), 2026 - The platform layer only, covering the platform margin rather than the underlying model and telephony costs. It sits below the all-in figure.

In India, Bolna prices from around Rs 7 per minute falling to around Rs 3 per minute at volume.

Rs 7 to Rs 3 per minvendor claim

Bolna published pricing, 2026 - A volume-tiered range, and the clearest illustration of why the US-weighted published range should be replaced with local cost. Vendor pricing changes often, so date it when quoting.

Conversational AI in contact centers was forecast to cut agent labor costs by USD 80 billion by 2026, with one in ten agent interactions automated by 2026.

USD 80bnanalyst forecast

Gartner press release, attributed to VP analyst Daniel O'Connell, 2022 - Dated 31 August 2022, and the date belongs in any citation. The one-in-ten figure is the half worth quoting: a ceiling that low on near-term automation sits well under the containment rates vendors advertise.

There is no reliable current cross-industry cost per call benchmark that accounts consistently for overhead, occupancy and attrition.

-no reliable figure

Which is why a borrowed benchmark is usually apples to oranges. The defensible figure is your own, built from local labor cost and your own occupancy.

There is no published data on knowledge base maintenance cost as a component of automated interaction cost, despite it being a real and recurring line.

-no reliable figure

It is one of the lines most often omitted from an automation business case, which is part of why projected savings come in high.

Examples

In practice

A business case projects savings using published cost figures and achieves substantially less. Reconciliation finds escalated interactions paying both the automation cost and the full human cost, and repeat contacts from interactions the agent handled poorly. Recalculating as cost per resolution across both paths produces a smaller projected saving that subsequent performance actually meets.

The everyday version

Cost per call is what it costs you to deal with one customer getting in touch. The comparison with automation looks dramatic, and it is real, but the honest version includes the calls the system gets wrong and hands to a person anyway, which you then pay for twice.

Usage

Who says it

  • Finance and operations, in business cases, where it is the headline saving.
  • Vendors, prominently, usually using the published range rather than a local one.

Where it turns up

  • In an RFP, next to pricing, volume commitments, escalation rates and expected containment.
  • On the board slide, where the flattering version tends to appear unless someone insists on the honest one.

Common misuse

  • Using published ranges rather than local labor cost, which is US-weighted and can be off by a large multiple.
  • Omitting escalation, repeat contact and maintenance costs, which is the most common source of overstated savings.
  • Comparing per-call cost rather than per-resolution cost, which flatters the automation whenever it fails and escalates.

Questions people ask

What does cost per call include?

A complete calculation includes agent labor with benefits, non-productive time driven by occupancy, training and attrition, technology and telephony, supervision and quality assurance, and facilities and overhead. The figures that flatter savings usually leave out escalations that pay both costs, repeat contacts from bad automation, and the cost of building and maintaining the automation.

What is the cost per call for a voice agent versus a human?

The widely cited range is roughly USD 7 to USD 12 for a human-handled call against roughly USD 0.40 for an agent-handled one. Treat it as an industry range rather than an audited figure. It varies enormously by geography, so a real business case should substitute local labor cost. In India, Bolna prices from around Rs 7 per minute falling to around Rs 3 at volume.

What is the difference between cost per call and cost per resolution?

Cost per call is the cost of one interaction. Cost per resolution is the total cost of solving a problem across every attempt and channel it took, including failures and escalations. Cost per call sets the direction of an automation business case; cost per resolution sets the magnitude, and the two frequently disagree.

Why do projected savings from automation come in high?

Usually because the projection omits the escalations that pay both the automation cost and the full human cost, the repeat contacts from interactions the agent handled badly, and the ongoing cost of maintaining the automation. Recalculating as cost per resolution across both paths produces a smaller number that later performance actually meets.

Share this definition

Last reviewed 4 August 2026. Written and reviewed by Dhruv Dholakia, founder of AsqVox.