Per resolution pricing: what it really costs you

Per resolution pricing means you pay your support software a fee every time its bot closes a customer question on its own. No fee when a person has to step in. It sounds fair, and for some teams it is. But the bill moves every month, the definition of "resolved" is set by the vendor rather than by you, and the better the bot gets, the more you owe. Here is how this billing model works, what a resolution actually costs in August 2026, and how to work out your own number before you sign anything.

What per resolution pricing actually means

Most help desks used to charge one way: a seat price, per person, per month. Your bill only changed when you hired someone.

Then bots got good enough to close real tickets, and vendors needed a way to charge for work that no longer had a person attached to it. Per resolution pricing was the answer. You keep paying for seats, and on top of that you pay a small fee each time the bot answers a customer without a handoff.

Two things make this different from a seat price. The bill is usage based, so it rises and falls with how many customers write in. And it is charged in arrears at most vendors, so you find out what you owe after the month is over rather than before it starts.

For a business that runs on a planned monthly budget, that second point matters more than people expect. The SBA's finance guide puts steady, forecastable outgoings at the centre of small business cash flow, and a support bill that arrives after the fact works against that.

What a resolution costs in 2026

Published rates across the bigger help desks run from roughly $0.60 to $2.00 per resolved conversation. Two we checked directly on the vendor's own pages, verified August 2026:

• Help Scout bills $0.75 per resolution, pay as you go, on every paid plan.
• Fin, the bot Intercom sells, lists $0.99 per resolution.

Figures reported for other desks sit either side of that. Committed volume deals tend to price lower per unit and charge a higher rate once you go past your allowance. Vendors change these numbers often, so check the current rate on their own pricing page before you commit to anything.

The unit price is the easy part. What that unit means is where the money actually moves.

What counts as a resolution is not standard

Every vendor writes its own definition, and they are not the same.

Help Scout's published rule is one of the clearer ones. You are charged once per chat session when the bot gives a full answer and the customer does not ask for more help. Greetings and follow-up questions on their own are not billable, and neither is a session where the customer asks for a person.

Other definitions are looser. Some count a resolution when the customer simply stops replying after a timeout, which bills you for people who gave up and left. Some bill on a ticket being marked closed, whoever closed it. A few wait for the customer to confirm the fix, which is the strictest reading and the one that bills you least.

Three questions get you a straight answer:

• What exact event triggers a charge? Get it in writing, not in marketing copy.
• What happens if the customer writes back the next day about the same problem? Does that reopen the ticket or bill a second resolution?
• Can I see a list of the conversations I was billed for? If you cannot audit the line items, you are trusting the vendor's own scoring of its own homework.

The math on a small support queue

Vendor comparisons of per resolution pricing tend to be written for teams handling fifty thousand conversations a month. Almost nobody reading this has that problem. Here is the same arithmetic at small business volume, using $0.75 as the rate and a 60 percent bot resolution rate, which is a fair mid-range assumption.

• 150 tickets a month: 90 resolved by the bot, about $68 on top of your seats.
• 400 tickets a month: 240 resolved, about $180.
• 800 tickets a month: 480 resolved, about $360.

Now add the seats. Three people on a desk that charges $30 a seat is $90 a month before a single automated answer. At 400 tickets that team is paying around $270, and the larger half of that bill is the part they cannot predict.

Push the bot rate up to $0.99 and the same 400-ticket month costs $238 in resolutions alone. Same volume, same team, a third more money, purely from the unit price.

Where per resolution pricing gets expensive

The uncomfortable feature of this model is that it charges you for success. Every improvement you make raises your bill.

Write better help articles and the bot answers more questions, so you owe more. Get a busy month from a marketing push and you owe more. Fix a confusing checkout page so fewer customers write in at all, and you finally owe less, which is the one lever that works in your favour.

Vendors have a reasonable defence here. An automated answer that costs you $0.75 replaced a person's time that cost several dollars, so your total support cost still falls. That argument holds up at scale, where you have enough volume that the alternative really was hiring. It holds up much less for a three-person business where nobody was going to be hired either way, and the resolution fee is simply a new line on the invoice.

The other cost is the ceiling. Most desks let you set a monthly spending cap, and when you hit it the bot switches off until the next cycle. That protects your budget. It also means your busiest week is the week your automated answers stop working.

Per resolution pricing compared with a flat monthly fee

A flat fee does the opposite. You pay a set amount, the bot answers as many questions as it can, and the invoice is identical whether you had a quiet month or a brutal one.

Neither model is dishonest. They suit different shapes of business.

Per resolution pricing tends to win when your volume is low or very uneven, when you are still testing whether a bot works on your content at all, or when you want your spend tied to something you can point at and measure. If you get 60 tickets in a quiet month, paying $27 beats paying a fixed fee you barely used.

A flat fee tends to win when volume is steady or growing, when you would rather know the number in advance than optimise it, or when you are actively trying to push more questions to self service. It also removes the awkward incentive where you quietly hope the bot answers slightly fewer questions.

A worked example at 400 tickets a month

Take a two-person shop getting 400 tickets a month, on a desk at $25 a seat, with a bot closing 60 percent of them at $0.75 each.

• Metered: $50 in seats plus $180 in resolutions. Around $230, and only the first $50 is knowable in advance.
• Flat: $39 on our Growth plan, which covers up to three seats and every automated answer the bot gives. Same number every month.

Change the assumptions and the gap moves. Drop to 100 tickets and the metered bill falls to about $95, which is still more than the flat plan but much closer. Raise the bot rate to $0.99 or the resolution rate to 75 percent and the metered bill climbs while the flat one sits still.

The point of running it yourself is not to find the cheapest month. It is to see how far your bill can swing between a quiet January and a bad December, and decide whether you want that swing on your books at all.

The comparison people get wrong is judging the models on unit price. A cheaper per-resolution rate can still produce a bigger invoice than a flat plan, and a flat plan can be poor value if your bot only closes a handful of tickets. Run your own volume through both before deciding. Our breakdown of cost per ticket gives you the per-ticket figure to compare against, and the hidden costs of help desk software covers the fees that sit underneath both models.

Five questions to ask before you sign

1. What triggers a billable resolution? A timeout, a closed status, or a confirmed fix. These produce very different bills from identical support.
2. Is it billed in advance or in arrears? Arrears billing means your first real invoice arrives a month after you have already spent the money.
3. Can I set a spending cap, and what happens when I hit it? Find out whether the bot goes quiet or whether you get charged an overage rate.
4. Do reopened tickets get billed twice? Ask specifically about a customer who comes back the next day.
5. What else is metered? Some desks charge separately for agent-facing tools, extra analytics or additional channels, all on top of seats and resolutions.

Ask all five by email so you have the answers in writing. A vendor who will not put its billing definition in an email is telling you something.

How SupportifyGPT prices this instead

We do not charge per resolution. The support bot is included on every SupportifyGPT plan, with no fee per answered question and no cap on how many questions it handles.

Plans are $15 a month for one seat, $39 for up to three, and $79 for up to ten. Every feature is on every plan, including the $15 one. The tiers differ only by how many people you need on the desk. There are no onboarding fees and no ticket overages, so the number you see is the number you pay. You can compare the tiers on our flat monthly pricing page.

Worth being straight about how the bot works, because it is the thing you are paying for either way. Hand-written rules do the finding: they read the customer's intent, spot when someone needs a person, and search your own help articles. Then a model writes the reply in plain sentences using only what those articles say. If your articles do not cover the question, it says so and passes the customer to you. It cannot invent a policy you never wrote down. If anything goes wrong it drops back to quoting the article directly, so a bad day makes it less useful rather than wrong.

When it drafts a new help article from a solved ticket, that draft waits for you to approve it. Nothing it writes goes live to customers unreviewed.

The honest limit of a flat fee is the same one every flat fee has. If you get very few tickets, a metered plan might cost you less in a given month. What you get in exchange is a bill you can plan around and no reason to hope your bot does less work. You can start a free trial and run your own tickets through it for fourteen days before you decide.