Zendesk now charges per seat for human agents and separately per automated resolution for AI ones, starting around a dollar per resolution. Microsoft is testing outcome based pricing for Copilot Studio. This model is arriving in renewals, and most buyers are evaluating it on price per unit when the important questions are elsewhere.
Paying for results rather than for logins is genuinely more honest than a seat licence for people who never sign in. It also transfers a risk you did not previously carry, and that risk is worth understanding before it is in a signed contract.
The definition is the whole negotiation
A resolution sounds self explanatory until you try to write it down.
If a customer asks a question, the agent answers, and the customer replies "that is not what I meant," was that a resolution. If a ticket is closed automatically after seven days of silence, was that a resolution. If the agent answers correctly but the customer opens a new ticket about the same problem an hour later, how many resolutions have you bought.
Each of those is a real billing dispute and each has a defensible answer in either direction. The vendor's default definition will be the one in their metering code, written by people with an interest in a generous reading. Getting your own definition into the contract is the single highest value thing you can do in the negotiation, and it costs nothing.
What to insist on: a reopened ticket within a defined window does not count. A conversation escalated to a human does not count, or counts at a reduced rate. Automatic closure through inactivity does not count. These are reasonable and most vendors will accept them, because refusing looks exactly like what it is.
You now need to audit a number you cannot see
Per seat billing is verifiable. You know how many people work there. Outcome billing rests on the vendor's own count of events happening inside their system, and you have no independent record.
That is not an accusation of dishonesty. It is a normal control problem: any metered charge you cannot reconcile is one you are accepting on trust, and finance teams do not enjoy discovering that mid year.
Ask for an exportable log of billable events with enough detail to sample. You will not check all of them. You will check twenty at random in month one, and the fact that you can is what keeps the count honest. A vendor unwilling to provide that detail is telling you something.
Also ask what happens when the metering is wrong. Every usage system has billing bugs eventually. The question is whether the contract provides for correction and credit, or whether their invoice is definitionally correct.
We have sat in renewal conversations where nobody on the buying side could say what the vendor's counter actually counted, and the answer arrived three invoices later as a dispute. The question costs nothing at contract stage and is expensive to ask afterwards.
Read More - WordPress Patched Twelve Flaws Across Twenty-Three Branches: What You Still Have to Do
The cost line moves in the wrong direction at the wrong time
Per seat costs are flat and predictable. Outcome costs scale with volume, and volume scales with the things that are usually bad for you.
A product outage generates support tickets. A confusing release generates support tickets. A pricing change generates support tickets. Under outcome billing your software bill rises during exactly the weeks you are already having a difficult month, which is a correlation nobody notices until it happens.
Seasonal businesses feel this most sharply. Retail in December, accountancy at year end, travel in summer. The bill peaks with the season, which is defensible commercially and still needs to be in the budget rather than a surprise. It is the same shape of problem as a pricing tier that jumps without warning, experienced from the buying side.
Negotiate a cap, or a committed volume with overage at a known rate. Vendors expect this. Uncapped variable spend is the thing procurement will object to, and finding out at signature is slower than raising it early.
Compare the two models on your own numbers
Vendors present outcome pricing as savings, and it often is, and the comparison is usually run against your current seat count rather than the seat count you would have next year.
Model three cases. What you pay today under seats. What you would pay under outcomes at current volume. What you would pay if volume rises 40 percent, which is a normal year of growth or a bad quarter of product quality.
That third number is the one that matters and the one nobody produces in the sales cycle. If it is uncomfortable, the answer is not to reject outcome pricing; it is to negotiate the cap that makes it acceptable.
That third case is the one we find missing almost every time a client brings us a vendor's comparison. The sales model runs against today's volume, which is precisely the number about to change.
What this signals about your vendor
There is a positive read worth taking seriously. A vendor charging per successful outcome has tied their revenue to their product working. That is a stronger alignment than a seat licence, which pays the same whether or not anyone benefits.
It also predicts where their engineering effort goes. A vendor billing per resolution is motivated to increase resolution volume, which is good when it means better automation and less good if it means classifying marginal interactions generously. Both incentives are real and the contract is where you choose which one you get.
Around 48 percent of B2B SaaS companies now run hybrid pricing as their primary model, so this is not a niche experiment you can wait out. It is worth building the evaluation once and reusing it, because you will see this structure repeatedly across renewals over the next two years. We looked at the same shift from the vendor's side in what agents did to software valuations.
Frequently Asked Questions
Is outcome pricing usually cheaper than per seat?
At current volumes, frequently yes, which is why vendors lead with it. The comparison that matters is at your expected volume in eighteen months, including a bad quarter. Cheaper today and uncapped later is a different deal from cheaper overall.
What should a resolution definition include?
At minimum: what counts as resolved, what happens if the customer reopens, how escalation to a human is treated, and whether automatic closure through inactivity is billable. Get all four in writing. Ambiguity here is not a technicality, it is the difference between a predictable bill and a quarterly argument.
How do we verify the vendor's count?
Ask for an exportable event log at contract stage rather than after the first invoice. Sample twenty entries in the first month against your own records. You are not auditing everything, you are establishing that the numbers can be checked, which changes the relationship.
Should we push back on this entirely and stay on seats?
You can, and some vendors will accommodate it, usually at a price. It is worth asking whether staying on seats actually serves you, because if your headcount is falling while your volume is not, a seat licence is quietly becoming the worse deal for you rather than the safer one.
What if the vendor will not agree to a cap?
That is useful information delivered early. Some genuinely cannot within their billing platform, which is a maturity signal about the vendor rather than a negotiating position. Others simply prefer not to. Either way, uncapped variable spend against a metric you cannot audit is a real risk, and it should be escalated rather than absorbed.
Does this apply to smaller vendors too?
Increasingly. The large vendors moved first and the model spreads downward quickly because it solves a real problem for anyone selling software whose customers are automating work. Expect to see it from smaller suppliers within a year or two.
How does this change what we should measure internally?
You now need your own count of the thing being billed, even roughly. Ticket volume, transaction counts, whatever the unit is. Without an internal figure you cannot forecast the bill, cannot check the invoice, and cannot tell whether a rise reflects growth or a change in how the vendor counts.
If you are building metering into your own platform rather than buying it, our SaaS development team treats caps, forecasting and an auditable event log as part of the pricing change rather than a later addition.
Talk to our team if outcome pricing has appeared in a renewal and you want the contract questions answered before you sign.