Per-resolution pricing changes your 12-month AI support cost more than the rate itself does. The billing unit varies by vendor, and that variance changes your 12-month cost more than the rate does. This post walks through how per-resolution pricing works, what "resolution" means in a contract, and where the hidden costs tend to appear.

TLDR:

• Per-resolution pricing charges only when a ticket closes with no human handover, not per message or seat.

• Most vendors define "resolution" loosely. Verify zero human handover is required before you sign.

• Watch for hidden costs: voice billed separately, seat fees layered on top, and overage penalty rates.

• Model your 12-month cost using four variables: ticket volume, resolution rate, overage behavior, and headcount offset.

• Fini prices per resolved ticket at $0.89, $0.69, or $0.49 depending on plan, with no per-seat fees and a Zero-Pay Guarantee: 90% resolution in 90 days, or you pay $0.

What Per-Resolution Pricing Actually Means

Per-resolution pricing charges you only when an AI agent closes a customer issue end to end, with no human stepping in. The billing unit is the outcome, not the activity that produced it.

That distinction matters. A ticket is a record. A conversation is an exchange. A message is a single turn. None of those tell you whether the customer's problem was actually solved. Per-resolution pricing skips all three and goes straight to the result.

Under a per-seat model, you pay for agents whether they handle ten tickets or ten thousand. Under per-message or per-conversation models, every exchange costs something regardless of whether the issue closed. Per-resolution ties what you owe to whether the customer's problem is gone. Understanding AI customer support pricing models helps frame why that distinction matters across billing structures.

How Per-Resolution Pricing Compares to Per-Seat and Per-Conversation Models

The three billing models create fundamentally different incentive structures. Per-seat pricing rewards vendor growth when you hire more agents. Per-conversation accrues cost regardless of whether the customer's issue closed. Per-resolution only charges when the ticket is done.

Per-seat fits teams where humans stay central, at $25 to $150 per seat per month for each human agent license. Per-conversation suits high-volume, low-handle-time flows where every exchange carries roughly equal weight, regardless of outcome. Per-resolution, typically $0.50 to $2.00 per resolution, fits best when autonomous AI carries most of the load, because the cost curve falls as the agent improves and the vendor only earns when the customer is actually helped.

What "Resolution" Actually Means in a Vendor Contract

"Resolution" does a lot of work in a contract, and vendors define it differently. Some count any conversation where the customer stopped replying. Others count it when ticket status changed to closed, regardless of who closed it or why.

Those definitions matter because they shift what you actually pay for. A customer who goes quiet after a frustrating non-answer is not a resolution. A ticket closed by a human agent after the AI failed is not a resolution either. The distinction between deflection rate vs. true resolution rate is covered in a full platform ranking.

Our definition is stricter: a resolution is a ticket closed end to end with no human handover. If Fini escalates to your team, that conversation is free. You pay only when the customer's problem is solved and no human touched it.

Check three things in any vendor contract before signing:

• Does "resolution" require zero human handover, or does a human-assisted close still count?

• Are escalations billed at any rate, or are they free?

• If a customer reopens the same ticket within 24 or 48 hours, does it bill again?

How to Model Your 12-Month Cost Under Each Pricing Model

Four variables drive your 12-month number under any AI support pricing model: monthly ticket volume, expected resolution rate, overage behavior, and headcount assumptions.

Start with volume. If you handle 8,000 tickets a month, a per-seat model at $100/seat with 20 agents costs $24,000/year regardless of how many tickets close. Per-resolution at $0.69 on 7,200 resolved tickets runs about $59,760/year, but replaces most of that headcount cost.

That headcount offset is where the models diverge. The AI support ROI model: cost per resolution walks through these numbers in detail. Per-seat costs scale with agents, not outcomes. Per-resolution costs scale with resolved volume, which falls per-unit as the agent improves.

A rough framework:

• Monthly ticket volume multiplied by resolution rate equals billable resolutions

• Billable resolutions multiplied by per-resolution rate equals monthly AI cost

• Subtract agent hours freed multiplied by fully-loaded agent cost to get net savings

At high volume, annual billing adds up fast. Fini's Enterprise plan drops to $0.49 per resolution, and multi-year commitments carry an additional discount. Per-seat pricing does the opposite: more agents, higher fixed cost, no cost reduction from better AI.

What Hidden Costs Appear Beyond the Base Per-Resolution Rate

The base rate on a pricing page is rarely what you end up paying. Several cost layers commonly appear after signing.

• Voice billed separately. Many vendors charge per-resolution for chat and email, then add a separate per-minute or per-call rate for voice. The top AI support tools pricing and TCO comparison covers how these line items stack across vendors. That second line item can double the effective cost for support teams where voice carries meaningful volume.

• Seat fees layered on top. Some per-resolution plans still charge per agent seat for access to the admin UI, reporting, or escalation routing. You pay for outcomes and for the people reviewing them.

• Compliance certifications gated by tier. SOC 2 or HIPAA compliance sometimes sits behind a higher plan. The best AI support vendors for predictable TCO comparison surfaces which vendors gate these behind plan upgrades. If your security review requires it, the "starter" rate becomes irrelevant.

• Overage penalty rates. Some vendors charge a higher per-resolution rate above your plan allowance, not the same flat rate.

Fini's structure avoids most of these. Voice is at a published rate card available on every plan. There are no per-seat fees at any tier. SOC 2 Type II, PCI DSS Level 1, ISO 27001, GDPR, HIPAA-compliant, and BAA-eligible coverage is included, not gated.

Overages bill at the same per-resolution rate as the plan, unused allowance rolls forward one month, and implementation is part of the plan.

The one real variable is voice volume. Fini's published voice rate card, consistent across all plans, sets the following tiers: under 10,000 calls per month runs $0.89 per answered call, above 10,000 drops to $0.59, and above 50,000 to $0.35.

That rate covers speech recognition, AI reasoning, voice synthesis, and telephony with no separate line items.

How AI Support Vendors Structure Their Plans in 2026

Most vendors in 2026 layer AI pricing on top of existing seat plans instead of replacing them. Intercom Fin charges $0.99 per outcome, where an "outcome" can include a case where the customer simply stopped replying, not necessarily a full resolution, on top of seat-based plan tiers. Agentforce uses flex credit pricing tied to its broader platform, adding cost before any ticket closes. Both approaches mean you pay for access and usage before a single ticket closes. A direct Intercom Fin vs Zendesk AI vs Fini cost-per-resolution comparison is available.

Does a Per-Resolution Model Work Without a Free Tier

Freemium tiers are common in AI support for a reason: they lower the barrier to testing. But free tiers also reflect something about the underlying product. A self-serve tool with no implementation requirements can offer a free plan because setup costs nothing. An autonomous agent that goes live on real production traffic, connected to your helpdesk and knowledge base, cannot.

The tradeoff is straightforward. A free tier gives you access. A paid pilot with agreed performance targets gives you proof. Those are different things.

Fini's Growth and Scale plans have no free tier and no self-serve signup. The Enterprise plan includes a 90-day free pilot on live traffic, with resolution, accuracy, and CSAT targets agreed in writing before day one. You pay nothing during those 90 days. If the numbers land, the plan starts. If they do not, you walk.

The Zero-Pay Guarantee backs it further: 90% resolution in 90 days, or you pay $0. The benchmark offer goes further still. Send us 1,000 real tickets and we return resolved answers with an accuracy score before any contract is signed.

A free tier answers "can I try it?" A paid pilot with a written performance commitment answers "will it work on my volume, my policies, and my tickets?" For a VP CX running 500,000 tickets a year, the second question is the one that matters. See the VP CX platform evaluation guide for a full evaluation framework.

How to Assess an AI Support Pricing Guarantee

Four questions worth asking any vendor before a guarantee becomes a deciding factor:

• What is the resolution threshold, and how is resolution defined? A guarantee set at 60% or 70% costs the vendor little to back because most scripted chatbots already hit those numbers on FAQ traffic. See how AI support platforms rank on accuracy guardrails to assess what those thresholds mean in practice. 90% on live production tickets is a different commitment.

• What ticket scope counts: all volume, or only the subset the AI touches?

• If the guarantee is missed, is the remedy a credit, a refund, or an exit right?

• Does the guarantee persist into the paid term, or only through the pilot? Some expire after one period with no performance backstop.

Our Zero-Pay Guarantee answers all four. The threshold is 90% resolution. The window is 90 days. If we miss, you pay $0, not a partial credit. The benchmark runs before any contract, on 1,000 real tickets from your queue, so you see the number before you commit.

How Fini AI Pricing Works

Three plans, one pricing logic: you pay per resolved ticket, and nothing else.

No per-seat fees at any tier. No implementation surcharge. Escalations to your team are free and arrive with full context attached. For a broader view, see AI support platforms: pricing and TCO. Overages bill at the same flat rate as your plan, unused allowance rolls forward one month, and annual billing saves roughly 17%.

Enterprise buyers get a 90-day free pilot on live production traffic. Resolution, accuracy, and CSAT targets are agreed in writing before day one. See the Zero-Pay Guarantee for full terms.

Final Thoughts on Per-Resolution AI Support Pricing

Outcome pricing only works if "outcome" is defined strictly. A customer who went quiet is not a resolution. A ticket closed by a human after the AI failed is not a resolution either. The pricing model you pick will quietly shape every vendor conversation you have from here, so it's worth getting the definitions right before day one. Book a quick intro call if you want to walk through how the math lands on your ticket volume.

FAQ

How does Fini AI pricing compare to Intercom Fin or Zendesk AI over 12 months?

Fini charges per resolved ticket only, with no seat fees layered on top. Intercom Fin adds $0.99 per outcome on top of existing seat-based plan tiers, and Zendesk AI bundles automated resolutions into add-on pricing tied to agent seats. At 10,000 resolutions per month on Fini's Scale plan, your AI cost is roughly $6,210/month with no separate access fees. The 12-month difference widens further if your team size grows, since Fini's cost curve moves with resolved volume, not headcount.

What counts as a resolution under Fini's per-resolution pricing model?

A resolution is a ticket closed end to end with no human handover. If Fini escalates to your team, that conversation bills at $0. Tickets where a human agent steps in, or where the customer reopens the same issue, do not count. Before signing any AI support contract, check whether the vendor counts human-assisted closes, silent drop-offs, or status changes as resolutions, since each definition changes what you actually owe.

Can I test Fini AI pricing against my actual ticket volume before signing a contract?

Yes. Send 1,000 real tickets from your queue and Fini returns resolved answers with an accuracy score before any contract exists. Enterprise buyers also get a 90-day free pilot on live production traffic. See the Zero-Pay Guarantee for full terms.

How does Fini AI handle knowledge base updates without a manual QA process?

Knowledge Atlas, Fini's self-maintaining knowledge system, runs a nightly learning pipeline that ingests escalated conversations, detects gaps, drafts new articles, and flags conflicts or outdated policies for human review before publishing. Teams running without it spend roughly 20 hours per week on documentation and typically plateau at 50-60% resolution. Teams on Knowledge Atlas average 2 hours per week on docs and reach 85-90% resolution. Both figures are based on Fini's implementation data across customers. Hourly knowledge updates are available on the Enterprise plan; nightly scans are included on Scale.

Does Fini AI support pricing include voice, or is it a separate line item?

Voice is on a published rate card available across all plans, not gated behind a higher tier or billed as a separate product. The rate covers speech recognition, AI reasoning, voice synthesis, and telephony with no additional line items. Under 10,000 answered calls per month runs $0.89 per call. Above 10,000 it drops to $0.59, and above 50,000 to $0.35. Multilingual voice at 130+ languages is included at no extra cost.