Resolution-based pricing

Resolution-based pricing

Resolution-based pricing

TL;DR

TL;DR

Resolution-based pricing is a billing model in which an AI support vendor charges only for issues its agent solves end to end with no human handover, so escalations and failed attempts cost nothing.

Resolution-based pricing is a billing model in which an AI support vendor charges only for issues its agent solves end to end with no human handover, so escalations and failed attempts cost nothing.

What is resolution-based pricing?

Resolution-based pricing is a billing model in which an AI support vendor charges only for issues its agent solves end to end with no human handover. Escalations, abandoned sessions and failed attempts are not billed. The billable unit is a completed outcome, not a seat, a login or a conversation.

Key takeaways

  • Pay for outcomes: you are billed only when the AI agent solves the issue with no human handover.

  • The definition is the price: what the contract calls a resolution decides what the bill means.

  • Five contract parts: definition, detection, attribution window, exclusion list, and reconciliation.

  • Run your own meter: count, sample, and rate billed resolutions instead of trusting one export.

  • Watch the incentive: paying for closures rewards closing, so pair the meter with reopen windows and quality review.

The model spread as AI agents began closing tickets on their own. A seat licence prices capacity, and capacity stopped being the constraint once software could handle many conversations at once.

For example, Fini AI pricing is $0.49 per resolved ticket. For worked monthly costs at three volumes, see per-conversation vs per-resolution pricing cost tables, and for the seat comparison, per-resolution vs per-seat AI pricing.

How resolution-based pricing works

A workable resolution-based contract has five parts. Read in order, they explain most invoice disputes.

Flow diagram: how resolution-based pricing works, definition to detection to attribution window to exclusions to reconciliation

How resolution-based pricing works: the five contract parts, read in order.

Part

What it does

1. Definition

States what a resolution is, usually a conversation the AI agent carried to closure with no human transfer and no reopen inside an agreed window

2. Detection

Proves it happened, from ticket status changes, transfer events, and reopen flags in the help desk. Aggregated monthly, these events supply the numerator of the resolution rate, though that rate also needs a total-inquiry denominator the billing meter never sees

3. Attribution window

Decides the edge cases: a ticket reopened inside it is credited back, one reopened after it stands

4. Exclusion list

Removes spam, duplicates, internal test traffic, and abandoned sessions before billing

5. Reconciliation

Matches the vendor’s billed list against your own help desk events every cycle

Because a billable event requires that the issue stay closed, the counting logic sits closer to first contact resolution than to ticket volume. Average resolution time remains a quality signal that never touches the invoice.

What counts as a resolution, and what does not

Outcome

Billed?

Why

Contained conversation

Yes

The agent answered or completed the task, with no human transfer, and the issue did not return inside the attribution window

Actioned request

Usually

The agent changed a system of record, such as issuing a refund or updating an address, even with a short reply

Escalated handoff

Almost never

Transferred to a human, by customer request or a confidence threshold

Reopened ticket

Credited back

The issue returned inside the attribution window, when the contract is written honestly

Non-issue

No

Spam, duplicates, internal test traffic, and abandoned sessions, removed by a standing exclusion rule

Silence alone is not evidence of a resolution, which is why stronger contracts pair the containment test with a survey or a sampled review.

Resolution-based pricing vs per-seat vs per-conversation pricing

Buyers compare these models on unit price, which hides the fact that each one bills a different event. Resolution-based pricing charges only when the issue is closed and stays closed, which moves delivery risk onto the vendor and makes the contract’s definition of “closed” the most valuable sentence in the document.


What it counts

What it misses

Typical benchmark

Resolution-based pricing

Issues closed with no human transfer and no reopen

Partial help, assisted cases, a graceful decline

No published norm; the definition is negotiated per contract

Per-seat pricing

Licensed seats provisioned for the period, used or not

Whether any seat resolved anything

Moves with headcount plans

Per-conversation pricing

Every session opened, resolved or abandoned

Outcome quality and repeat contacts

Moves with contact volume, failures included

Deflection-based pricing

Contacts that did not reach a human, including customers who gave up

Whether the customer got a usable answer

Reported figures vary with each vendor’s definition

Hybrid base charge plus resolutions

A fixed platform charge and a variable outcome meter

The boundary between the two when scope shifts

Two meters on one invoice; audit both

If you can instrument closures and reopens in your own help desk, and you want the vendor carrying delivery risk, the resolution meter is the right one. If volume is flat and contact reasons are stable, a fixed fee is easier to forecast and cheaper to administer.

Why resolution-based pricing matters for customer experience

Pricing shapes behavior on both sides of a contract.

  • It moves the cost of vague answers. Paid per conversation, a vendor earns from a second attempt. Paid per resolution, the vendor carries that cost.

  • It ties spend to real savings. A seat count says nothing about how many customers left satisfied. A human contact carries queue time, handling time, and after-call work, so a contact resolved without one saves money, while a contact merely avoided may only be deferred. That is why the gap between deflection and resolution rates gets argued over so hard.

  • It creates a closing incentive. An agent that ends ambiguous conversations early books revenue while quietly damaging satisfaction.

How is resolution-based pricing measured?

No standards body sets a price per resolution a buyer is expected to hit, and no neutral registry defines what a resolution is. Any figure offered as an industry average needs its definition and its denominator attached before it means anything.

What you can measure is the meter itself, in three passes:

  1. Count. Pull closure, transfer, and reopen events from your own help desk and reconcile them against the vendor’s billed list every cycle.

  2. Sample. Draw a random set of billed conversations each month and have a reviewer judge whether the customer’s problem was solved.

  3. Rate consistently. Use a written protocol, a fixed scale, and more than one independent reviewer. Raw agreement overstates reliability when reviewers can guess alike, which is why chance-corrected measures of interrater agreement exist.

The output is a billable-accuracy figure of your own: the share of billed resolutions your reviewers agree were resolutions, reported with the agreement level behind it. That figure belongs in the dispute clause.

How AI agents change resolution-based pricing

What changes is the evidence. When a human closes a ticket, the proof is a status field set by hand. When an AI agent closes one, the transcript, the tool calls, the record it updated, and the reopen history all exist as machine-readable events, and an outcome that can be observed independently can be billed for.

  • Vendors expand coverage. Every intent the vendor still escalates is unearned revenue, so it has reason to take those on.

  • Confidence thresholds matter more. A case damaged by an over-eager agent is a reopen nobody gets paid for.

  • Buyers negotiate per contact reason. Those who understand the incentive set coverage targets reason by reason.

  • Actions add outcomes. The economics of per-resolution versus per-seat pricing shift again once the agent acts inside other systems, since each action is another observable outcome.

What to look for in a resolution-based pricing agreement

Clause

What to require

Coverage

Which contact reasons the vendor commits to handling, and what happens to the ones it declines

Integration surface

Closure, transfer, and reopen data flow to your systems as well, so both parties see the same events

Governance

Who may change the resolution definition mid-term, how long the dispute window runs, and a line-item export of every billed conversation. The clause buyers skip

Security

SOC 2 Type II, ISO 27001, HIPAA with a signed BAA, and GDPR obligations, which apply to the transcripts the meter is built on

Forecasting

A plan for budgets that move with demand, since the meter is variable

Resolution-based pricing and support quality programs

A billing meter built on outcomes only works if someone inspects the outcomes. That is the job of customer service quality assurance, whose scorecards can be pointed at billed AI conversations exactly as at human ones, turning a sampling program into a billing control.

The second connection is contractual. A service level agreement committing to response and resolution times sits naturally beside a resolution meter, because the same closure events that trigger a service credit also determine what the vendor may invoice that month.

What does resolution-based pricing mean in plain terms?

Think of it as paying a plumber when the leak stops, not for the hours the van sat outside. The company selling the software carries the risk that it works, and in exchange it gets a say in what counts as a stopped leak.

Under a seat-based deal, you buy ten licenses in January and justify all ten in December. Under this model, a quiet month produces a small bill, and a month where the software struggles produces a small bill too, so spending tracks results.

The tradeoff is that everything hangs on one word. If the seller decides what “resolved” means, and that definition quietly includes conversations where the customer shrugged and walked away, you are buying outcomes on paper and activity in practice. Agreeing the definition in writing, with a reopen window and an audit right, is where the negotiation should concentrate.

Common resolution-based pricing mistakes

Mistake

Why it hurts

What to do

Signing before defining

The definition is the price, so deferring to the vendor’s dashboard means no agreed price. Drift arrives through a release that reclassifies escalations as containments

Agree the definition in writing first

Comparing headline unit rates

Identical rates mean different things when one counts abandoned chats and the other excludes them

Normalize the definitions, then compare

Trusting a single meter

When only the vendor can count, reconciliation becomes a negotiation over their export

Pull your own closure and reopen events

Ignoring the quality incentive

A model that pays for closures rewards closing, so the billed number can improve while experience degrades

Pair the meter with reopen windows and sampled quality review

Frequently Asked Questions

How much does resolution-based pricing cost per resolution?

Resolution-based pricing has no standard market rate; it moves with contact complexity, integration depth and volume, and depends on the definition of a resolution attached to it. Fini AI pricing is $0.49 per resolved ticket.

What is the difference between resolution-based pricing and per-seat pricing?

Resolution-based pricing bills for outcomes the software produces, while per-seat pricing bills for provisioned capacity regardless of use. The first makes spending move with demand and shifts delivery risk to the vendor. The second is stable and easy to forecast, which suits teams with flat volume and predictable staffing plans.

Is deflection the same as a resolution for billing purposes?

Deflection and resolution measure different things. A deflected contact never reached a human, which includes customers who abandoned the chat unsatisfied. A billable resolution requires that the issue was actually closed and did not return inside the attribution window. Contracts that blur the two tend to overbill, so define both terms explicitly.

How do vendors verify that a resolution actually happened?

Verification runs on event data from the help desk: ticket closure, absence of a transfer to a human, and no reopen within an agreed window. Stronger contracts add a survey signal or a sampled quality review. Buyers should pull the same events independently and reconcile them against the vendor’s billed list each cycle.

Does resolution-based pricing save money compared with hiring more agents?

Resolution-based pricing changes the cost curve rather than guaranteeing savings. A resolved contact is charged once, while a human contact carries salary, queue time, handling time, wrap-up, and management overhead. Savings appear when automated coverage is high across repeated contact reasons and disappear when most conversations still escalate to people.

What happens if a customer reopens a ticket that was already billed?

A reopened ticket falls under the attribution window written into the contract. If the customer returns inside that window, the resolution should be credited back on the next invoice. Reopens after the window usually stand as billed. The window length is negotiable, and short windows favor the vendor.

Learn More

Learn More

Knowledge base

K

Average handling time (AHT)

A

Telephony

T

Customer acquisition cost (CAC)

C

AI tokens

A

Human in the loop (HITL)

H

AI grounding vs retrieval-augmented generation (RAG)

A

Short message service (SMS)

S

Call center

C

Data annotation

D

Ticket routing

T

Customer service quality assurance (QA)

C

Live chat

L

Speech Synthesis Markup Language (SSML)

S

Batch inference

B

Barge-in

B

SLA compliance rate

S

Queue management

Q

Prompt versioning

P

Emotion detection

E

Retrieval-augmented generation (RAG)

R

Natural language understanding (NLU)

N

Text classification

T

Call routing

C

Customer churn rate

C

Speech-to-speech

S

Intent recognition

I

Voice of the employee (VoE)

V

Confidence score

C

AI personalization

A

Voice cloning

V

Asynchronous messaging

A

Hallucination

H

ReAct agent pattern

R

Long-term memory

L

Forecast accuracy

F

Customer feedback loop

C

Structured output

S

Outbound voice AI

O

AI guardrails

A

Direct preference optimization (DPO)

D

Prompt chaining

P

SIP transfer

S

Fallback intent

F

Conversation summarization

C

Auto-tagging

A

Cost per contact

C

VoIP jitter

V

Model card

M

Ticket prioritization

T

Sentiment analysis

S

Agent utilization rate

A

Speech-to-intent

S

Prompt engineering

P

Knowledge atlas

K

SOC 2 AI support

S

Prosody

P

Chatbot containment rate

C

Speech synthesis

S

ISO 42001

I

Intent-based search

I

After-call work (ACW)

A

Chatbot

C

AI agent

A

Prior authorization automation

P

AI customer service

A

Ticket deflection

T

AIUC-1

A

Workforce management (WFM)

W

Skill-based routing

S

Interactive voice response (IVR)

I

Contact center as a service (CCaaS)

C

Warm transfer

W

Customer segmentation

C

Reinforcement learning

R

Voice activity detection (VAD)

V

Tiered support

T

Echo cancellation

E

Net Promoter Score (NPS)

N

Endpointing

E

>