Direct answer: per-resolution pricing means you pay a per-unit fee every time an AI agent successfully closes a customer conversation, on top of whatever base plan or seat fees you already owe. Intercom Fin charges $0.99 per resolution, Zendesk charges $1.50 to $2.00 per automated resolution, Gorgias charges $0.85 to $1.00 per resolved interaction, and Salesforce Agentforce charges $2 per conversation or a Flex Credits equivalent of roughly $0.10 per action. None of these are flat fees. Every one of them moves with volume, and every one of them defines "resolution" slightly differently, which is exactly where an unexpected invoice comes from.
If you landed here because a quote or an invoice came in higher than you expected, you are not imagining it. Per-resolution and outcome-based billing is now the default pricing model for most enterprise-grade AI support agents, and it is structurally different from a monthly SaaS subscription: it rewards the vendor when the AI works well, because a better resolution rate means more billable units, not fewer.
This guide is vendor-neutral. It walks through what actually counts as a "resolution" on each platform, lines up the billing mechanics side by side, and builds three fully-shown math tables at 500, 2,000, and 10,000 conversations a month so you can audit the arithmetic yourself. For a deep dive on any single vendor, see our breakdowns of Intercom Fin pricing and Gorgias pricing in full detail.
Quick verdict
- What counts as a "resolution": a real definition that varies by vendor, usually a conversation that closes without a human agent stepping in, or without the customer asking for more help
- Cheapest per-unit rate: Gorgias, $0.85–$1.00/resolution, but stacked on top of its own per-ticket Helpdesk fee
- Most transparent single number: Intercom Fin, a flat $0.99/outcome, though seats are billed separately and rise with plan tier
- Widest published range: Zendesk, $1.50/resolution committed vs $2.00 pay-as-you-go, a 33% swing depending on how you buy
- Least like the others: Salesforce Agentforce, priced per conversation or per action (Flex Credits) rather than strictly per resolution
- The real risk: the better any of these agents perform, the more resolutions it logs, and the higher the bill climbs, the opposite of what "automation savings" usually means
- The flat-fee alternative: platforms like Heeya's RAG chatbot charge a fixed monthly tier instead, so a strong resolution month costs the same as a quiet one
Table of Contents
- What actually counts as a "resolution"?
- Side by side: how each vendor bills
- The math at 500, 2,000, and 10,000 conversations a month
- The deflection paradox: success raises the bill
- Hidden costs: seats, add-ons, minimums, overage
- When per-resolution is actually the right model
- Flat-fee alternatives and their real trade-offs
- Model your own break-even in 10 minutes
- FAQ
What Actually Counts as a "Resolution"?
A "resolution" is the billable unit in per-resolution pricing, and it is defined by the vendor, not by any industry standard. This single definition determines whether your bill tracks your actual support volume or drifts above it, and the four vendors below do not define it the same way.
Intercom Fin: a resolution, a handoff, or a disqualification
Intercom bills a "billable outcome" at $0.99 each, capped at once per conversation regardless of how many questions Fin answers inside it. An outcome fires when the customer confirms Fin's answer solved the issue, when the conversation ends without the customer asking for more help, or when Fin completes a defined workflow or hands off to a human. According to Intercom's pricing page (verified September 2026), there is also a minimum monthly commitment of roughly 50 outcomes for standalone Fin use.
Zendesk: a 72-hour quiet period defines success
Zendesk's outcome-based pricing counts an automated resolution when its AI agent, generative reply, or a linked help-center article closes a ticket and the customer does not reopen it within a defined quiet window, commonly cited at 72 hours. If the customer comes back with the same issue inside that window, the resolution generally does not bill. Zendesk's own explanation of the model is published at zendesk.com's outcome-based pricing overview (referenced September 2026).
Gorgias: a resolution and a ticket are billed separately
Gorgias defines a billable AI resolution as any conversation its AI Agent closes independently, without a human agent touching it. The catch is structural rather than definitional: Gorgias also charges for the underlying ticket itself under its per-ticket Helpdesk plan, so a single conversation the AI fully resolves can still count against your ticket allowance and trigger a separate AI resolution charge. See the full mechanics in our Gorgias pricing explained breakdown.
Agentforce: billed per conversation or per action, not strictly per "resolution"
Salesforce does not use the word "resolution" as its billing unit. Agentforce is priced per conversation ($2 each, regardless of outcome) or through Flex Credits, where a completed action, not a completed resolution, consumes credits. A conversation that needs many actions to close costs more under Flex Credits than a one-step exchange, even if both count as a single successful outcome.
For teams evaluating whether that per-conversation model is worth it, see our roundup of Salesforce Agentforce alternatives for SMBs. Microsoft's competing enterprise platform, Copilot Studio, uses a different consumption-credit model altogether; our Copilot Studio vs RAG chatbot comparison breaks that down.
| Vendor | Billable event | Bills once per conversation? |
|---|---|---|
| Intercom Fin | Confirmed answer, silent close, or completed workflow | Yes |
| Zendesk | Ticket stays closed past a 72-hour quiet period | Yes, per resolved ticket |
| Gorgias | AI closes the conversation without a human agent | No, resolution fee is separate from the ticket fee |
| Agentforce | Conversation opened, or each action taken (Flex Credits) | Depends on model chosen |
Side by Side: How Each Vendor Bills
Beyond the definition of a resolution, the base billing structure differs just as much. Here is what each vendor actually charges, verified against its own pricing page.
- Intercom Fin: $0.99 per outcome, plus per-seat plans of $29/seat/month (Essential), $85/seat/month (Advanced), or $132/seat/month (Expert), all billed on top of Fin's usage fee. Source: intercom.com/pricing (verified September 2026).
- Zendesk: $1.50 per automated resolution on committed volume, or $2.00 pay-as-you-go, layered on Suite/Support seat plans starting near $19/agent/month and rising to $89–$115/agent/month on Growth and Professional tiers. Source: zendesk.com/pricing and Zendesk's outcome-based pricing explainer (verified September 2026).
- Gorgias: bundled Helpdesk + AI Agent plans at $40/month (Starter, 50 tickets), $90/month (Basic, 300 tickets), $550/month (Pro, 2,000 tickets), and $1,430/month (Advanced, 5,000 tickets), each with ticket overage of $0.36–$0.40 per extra ticket, plus a per-resolution AI Agent rate of $1.00/resolution on Starter down to roughly $0.85/resolution on Advanced. Source: gorgias.com/pricing and Gorgias's AI Agent pricing explainer (verified September 2026).
- Agentforce: $2 per conversation, or Flex Credits at $500 per 100,000 credits, where a standard action consumes 20 credits (about $0.10/action) and a voice action consumes 30 credits (about $0.15/action). Source: Salesforce's official Agentforce pricing article (verified September 2026).
One pattern holds across all four: the headline per-unit rate is never the whole story. Seats, ticket allowances, and overage all stack underneath it, which is exactly what the next two sections unpack.
The Math at 500, 2,000, and 10,000 Conversations a Month
Every figure below is built from the verified per-unit prices above, using stated, adjustable assumptions: a 60% AI resolution rate (a conservative mid-point below Intercom's own reported median of 67%), a support team that scales roughly with volume (3 seats at 500 conversations, 6 at 2,000, 12 at 10,000), and Gorgias billed on the plan tier that matches each conversation volume as a ticket count. Adjust these to your own numbers using the worksheet in section 8.
500 conversations/month (300 resolved at 60%)
| Vendor | Calculation | Monthly total |
|---|---|---|
| Intercom Fin | (300 × $0.99) + (3 seats × $29) = $297 + $87 | $384 |
| Zendesk | (300 × $1.50) + (3 seats × $19) = $450 + $57 | $507 |
| Gorgias | Basic $90 + (200 extra tickets × $0.40) + (300 × $0.90) = $90 + $80 + $270 | $440 |
| Agentforce | 500 × $2.00 | $1,000 |
| Heeya (flat fee) | Standard plan, €19/mo flat, up to 500 messages included | €19 (~$22) |
2,000 conversations/month (1,200 resolved at 60%)
| Vendor | Calculation | Monthly total |
|---|---|---|
| Intercom Fin | (1,200 × $0.99) + (6 seats × $29) = $1,188 + $174 | $1,362 |
| Zendesk | (1,200 × $1.50) + (6 seats × $19) = $1,800 + $114 | $1,914 |
| Gorgias | Pro $550 (2,000 tickets included, no overage) + (1,200 × $0.90) | $1,630 |
| Agentforce | 2,000 × $2.00 | $4,000 |
| Heeya (flat fee) | Premium plan, €99/mo flat, up to 2,500 messages included | €99 (~$113) |
10,000 conversations/month (6,000 resolved at 60%)
| Vendor | Calculation | Monthly total |
|---|---|---|
| Intercom Fin | (6,000 × $0.99) + (12 seats × $29) = $5,940 + $348 | $6,288 |
| Zendesk | (6,000 × $1.50) + (12 seats × $19) = $9,000 + $228 | $9,228 |
| Gorgias | Advanced $1,430 + (5,000 extra tickets × $0.36) + (6,000 × $0.85) = $1,430 + $1,800 + $5,100 | $8,330 |
| Agentforce | 10,000 × $2.00 | $20,000 |
| Heeya (flat fee) | Exceeds Premium's 2,500-message cap | Enterprise, custom quote |
Gorgias figures treat every AI resolution as billable at the full per-resolution rate for clarity; each tier also bundles a small number of free automated interactions before that rate kicks in, which would modestly lower the real total. See the exact bundle sizes in our Gorgias pricing explained breakdown. Heeya's dollar figures use a €1 ≈ $1.14 conversion (September 2026 average). Above the published message caps, Heeya requires an Enterprise quote; no flat number is published at that volume, so none is estimated here. Seat counts for Intercom and Zendesk are a modeling assumption, not a vendor requirement, adjust to your own headcount.
The Deflection Paradox: The Better It Works, the More You Pay
Per-resolution pricing creates a structural tension: the entire pitch for AI support is that it deflects volume away from expensive human agents, but every conversation it deflects successfully is also a new billable unit. A support leader who tunes the knowledge base, fixes gaps, and pushes resolution rate from 50% to 80% is not just improving customer experience; they are directly and proportionally raising the invoice.
Intercom itself reports a median Fin resolution rate of 67%, with top-performing accounts above 90%, according to figures cited in our Intercom Fin pricing breakdown. Run that improvement through the math above: at 2,000 conversations a month, moving from a 60% to an 80% resolution rate on Intercom Fin adds roughly 400 more billable outcomes, or about $396 more per month, purely because the AI got better at its job.
This is not a flaw unique to any one vendor; it is the mechanical consequence of billing per successful outcome instead of per seat or per tier. It also means budgeting for these tools requires forecasting your resolution rate, not just your conversation volume, a variable that is much harder to predict and that improves precisely when you invest in the product.
Hidden Costs: Seats, Add-Ons, Minimum Commitments, Overage
The per-resolution rate is rarely the full bill. These are the line items that most frequently surprise a finance team reviewing the invoice for the first time.
- Per-seat fees stack independently of AI usage. Intercom's seat price triples from Essential ($29/seat/month) to Expert ($132/seat/month), unrelated to how many resolutions Fin logs.
- Minimum monthly commitments. Intercom applies a floor of roughly 50 billable outcomes a month for standalone Fin, so a quiet month still generates a minimum charge.
- Add-ons priced separately. Intercom's AI Copilot, Zendesk's advanced automation add-ons, and Gorgias's Voice and SMS channels all sit outside the core per-resolution or per-ticket fee.
- Two overage meters at once. Gorgias charges ticket overage ($0.36–$0.40/ticket) independently of its AI resolution fee, so a busy month can trigger both simultaneously.
- Annual lock-in for the advertised rate. The lower seat prices cited above (Zendesk's entry tier, Intercom's Essential tier) apply to annual billing; month-to-month runs higher.
- Committed vs pay-as-you-go gaps. Zendesk's $1.50 committed rate versus $2.00 pay-as-you-go is a 33% swing based purely on whether you pre-commit to volume.
When Per-Resolution Is Actually the Right Model
It would be dishonest to present per-resolution billing as universally bad. For the right buyer, it is the more rational choice, and being clear about who that buyer is matters more than selling against the model.
Per-resolution pricing suits high-ticket, high-margin support organizations where the value of a resolved conversation vastly exceeds its billing cost. A SaaS company selling a $50,000/year contract, or a business where a single unresolved support conversation risks a large churned account, can justify $1–$2 per resolution without blinking, because the downside of a bad support experience dwarfs the AI's usage fee.
It also suits volume that is genuinely unpredictable and spiky in the opposite direction from what small teams fear: an enterprise that wants to guarantee it never runs out of AI capacity during a surge, and would rather pay proportionally for exactly what gets used than negotiate a large committed tier upfront. And it suits teams that already run a mature help-center or Articles knowledge base and want deep CRM context baked into every answer, which is where Intercom Fin and Zendesk's AI agents are built to excel.
Where it breaks down is at SMB and mid-market volume with a fixed marketing or support budget, where a strong month is punished with a bigger bill instead of rewarded with lower unit costs, exactly the profile modeled in the tables above.
Flat-Fee Alternatives and Their Real Trade-Offs
The honest alternative to per-resolution billing is a flat monthly tier: you pay the same amount whether the AI resolves 50 conversations or 500 within that tier's published cap. This removes the deflection paradox entirely, a strong month and a quiet month cost the same, but it comes with real trade-offs, not a free lunch.
- Volume caps. A flat-tier platform like Heeya publishes a message ceiling per plan (500/month on Standard, 2,500/month on Premium), and once you cross it, you either upgrade to the next tier or move to a custom Enterprise quote, the same discrete-step model that the Gorgias article above describes for its ticket allowances.
- Fewer integrations than enterprise suites. Flat-fee, RAG-native chatbots are typically narrower in scope than a full helpdesk platform: no shared inbox with ticket routing across a large team, and a smaller integration catalog than Intercom or Zendesk offer after a decade of platform investment.
- Not a like-for-like replacement. Gorgias and Zendesk are full multichannel helpdesks with human-agent workflows; a flat-fee AI chatbot is usually the self-serve, pre-ticket layer, not a wholesale swap for a support inbox with a human team behind it.
Some SMBs weigh the flat-fee AI chatbot route against a simpler option entirely: hiring a part-time human to handle routine messages. Our AI chatbot vs virtual assistant cost comparison runs that math over a full year. Others are comparing it against a phone-based service; see AI receptionist pricing for small business for that flat-fee-vs-per-minute breakdown.
The trade a flat-fee tool asks you to make is real: less platform depth, in exchange for a bill that never moves within your tier. For SMB and mid-market teams whose volume fits comfortably under a published cap, that trade is usually worth it. See how much an AI chatbot actually costs in 2026 for the full landscape of pricing models, not just per-resolution ones.
How to Model Your Own Break-Even in 10 Minutes
You do not need a spreadsheet consultant to figure out which pricing model wins for your business. Five inputs, in order.
Step 1: pull your busiest month, not your average
Per-resolution bills spike exactly when volume does, so an average-month estimate always understates your true annual cost.
Step 2: estimate your realistic AI resolution rate
Start conservative, 50–60%, unless your knowledge base is already mature, in which case 65–75% is defensible against Intercom's own published median of 67%.
Step 3: multiply resolutions by each vendor's per-unit rate
Use the verified rates in section 2: $0.99 (Fin), $1.50–$2.00 (Zendesk), $0.85–$1.00 (Gorgias), or $2.00/conversation (Agentforce).
Step 4: add seats, minimums, and known add-ons
Multiply your realistic human headcount by the seat price, then add any minimum commitment and add-ons you already know you need (Copilot, Voice, SMS).
Step 5: compare against a flat-tier quote at the same volume
Check whether your busiest-month volume fits inside a published flat tier, and at what price. Our AI chatbot ROI calculator offers a second worked-example methodology to cross-check your numbers, and Heeya's pricing page shows the current flat tiers to compare against.
Frequently Asked Questions About Per-Resolution Pricing
What is per-resolution pricing?
Per-resolution pricing means you pay a set fee every time an AI agent successfully closes a customer conversation, instead of a flat subscription. Intercom Fin charges $0.99/resolution, Zendesk charges $1.50–$2.00, and Gorgias charges $0.85–$1.00, each on top of separate seat or plan fees. The bill rises directly with how many conversations the AI resolves.
What is a "resolution" in Intercom Fin?
A Fin resolution is a billable outcome charged at $0.99, capped once per conversation. It fires when the customer confirms Fin's answer solved the issue, when the conversation ends silently with no follow-up, or when Fin completes a defined workflow or handoff. A minimum of roughly 50 outcomes/month applies for standalone Fin use, per Intercom's pricing page (verified September 2026).
How does Zendesk's outcome-based pricing work?
Zendesk charges $1.50 per automated resolution on committed volume, or $2.00 pay-as-you-go, on top of Suite or Support seat plans starting near $19/agent/month. A resolution counts when a ticket closes and the customer does not reopen it within a quiet period, commonly cited at 72 hours.
How much does Agentforce cost per conversation?
Agentforce offers two models: a flat $2 per conversation, or Flex Credits at $500 per 100,000 credits, where a standard action consumes 20 credits (about $0.10 each). Flex Credits are cheaper only if a conversation needs fewer than about 20 actions to resolve; above that, the flat $2/conversation rate wins.
Why does my AI support bill go up when the AI gets better?
Because per-resolution pricing charges per successful outcome, not per conversation attempted. Improving your knowledge base raises the resolution rate, which directly increases billable units each month. That's the deflection paradox: the AI doing its job better is exactly what increases the invoice.
Is per-resolution pricing ever a good deal?
Yes, for high-ticket, high-margin support organizations where a resolved conversation's value far exceeds its per-unit cost, and for enterprises with genuinely unpredictable volume that would rather pay proportionally than pre-commit to a large tier. It tends to fit poorly for SMB and mid-market teams on a fixed monthly budget.
What is the alternative to per-resolution AI pricing?
Flat-fee, tiered pricing: you pay a fixed monthly amount for a published cap on conversations or messages, and the bill does not move within that cap regardless of resolution rate. The trade-off is a lower volume ceiling and typically a narrower integration catalog than enterprise helpdesk suites like Intercom or Zendesk.
How do I calculate my break-even between per-resolution and flat-fee pricing?
Take your busiest month's volume, apply a conservative resolution rate (50–60% unless your knowledge base is mature), multiply resolutions by the vendor's per-unit rate, add seat and add-on costs, then compare the total against a flat-tier plan's published price at the same volume. Section 8 above walks through each step.
Tired of watching your invoice move with your resolution rate?
Heeya charges a flat monthly fee, from a genuine free tier, with no per-resolution meter running underneath it. See exactly what's included on the Heeya pricing page.
Further Reading
- Heeya vs Intercom Fin: Pricing at 500–5,000 Chats
- Gorgias Pricing Explained: Per-Ticket Costs in 2026
- AI Agent Pricing in 2026: How Much Does It Cost?
- AI Chatbot Pricing 2026: Costs by Type + 6 Tools Compared
- AI Chatbot ROI Calculator 2026: Formula + 3 Worked Examples
- Top Intercom Alternatives for SMBs in 2026
