Per-seat AI pricing hides your real bill behind seat minimums. Per-resolution pricing hides it behind definitions of "resolution." AI response pricing is cleaner — but only if you understand what counts. Here's how each model actually works, and which one aligns cost with value.
If you've been shopping AI platforms in the last year, you've probably seen quotes that don't look anything like each other. That's because vendors are using three fundamentally different billing units, and the cheap-looking number at the top of the page can balloon fast depending on which one you pick.
The three you'll run into:
None of these is secretly evil. But each one aligns cost with a different thing, and mismatches produce ugly surprises at renewal.
Per-seat is the familiar SaaS model — $X per user per month. The pitch is clean: you only pay for the team members who need access. The trap is two-fold:
Seat minimums. Many vendors require 3, 5, or 10 seats to unlock the tier that has the AI. A 5-seat minimum at $90/seat is $450/mo whether you have two people or five. If your goal is "get an AI on my website" — a customer-facing capability, not an internal one — per-seat is asking you to pay for people who don't need dashboard access to deliver the feature you want.
The AI isn't the seat feature. When you buy a seat on an old-school support tool, the seat is the thing. When you buy a seat on an AI-ified support tool, you're paying for the seat and usage fees on top. The seat math looks cheap compared to per-response pricing, right up until you realize the AI resolutions are billed separately.
Per-seat makes sense when your actual job is giving your team access to a shared CRM. It stops making sense when the job is "put a conversational AI on my website and phone."
Per-resolution pricing sounds value-aligned — you pay when the AI actually solves something. In practice, "resolution" is defined by the vendor, and the definition matters a lot.
Common definitions seen in the wild:
Pricing ranges from roughly $0.50 to $1.50+ per resolution. The spread matters. A site doing 3,000 qualified conversations a month at $0.99/resolution is $2,970 — before seats.
The subtle kicker: resolution-based pricing rewards the vendor when the AI doesn't escalate. That's usually fine, but it can nudge product decisions toward "keep the customer in the AI" rather than "hand off cleanly." Always ask: what happens on escalation? Is that conversation still a billable resolution? You'd be surprised how often the answer is "yes."
Per-AI-response is cleaner: the agent generates a reply, you pay for that reply. One AI reply = one billable unit. It's the model Kazozo uses.
The advantages:
Things to check before signing any per-response contract:
Imagine a home-services business doing 1,500 website conversations a month and 300 inbound phone inquiries. You want AI handling both. Here's how the three models shake out:
At this volume, per-AI-response is 2–6× cheaper than per-seat with resolution overlays, and it's the only one where the bill is obvious before renewal. The ROI calculator lets you plug in your actual volume.
Before you sign any AI platform contract, get five things on paper:
If the sales team can't answer any of these in a single email, that's your answer: the pricing model isn't clean enough for an SMB budget.
Being fair: per-seat isn't always the enemy. If your entire team lives inside a shared CRM — SDRs working the pipeline, reps running sequences, managers pulling reports — per-seat charges for the thing you're actually using. Intercom, HubSpot Service Hub, and Salesforce are all good CRMs; their per-seat models make sense if the CRM is the product.
What doesn't make sense is paying a per-seat premium to unlock a customer-facing AI feature. That's a mismatch between billing unit and the job to be done.
AI response pricing isn't automatically the cheapest — it's the most legible. You can read your bill and know exactly why it's what it is. Per-seat and per-resolution models can work out fine, but only when you've asked the five audit questions above and written the answers into the contract.
If you want to see per-response pricing running on your actual volume — for both chat and voice — book a 20-minute demo. We'll pull your real traffic numbers and show you what the monthly bill would look like on Starter, Growth, and Pro.