SaaS subscription pricing was one of the best commercial structures the software industry ever invented.
Everybody talks about recurring revenue which in my view is just so powerful mainly because of the clear incentive alignment.
Perpetual licensing captured most of the commercial value upfront. Once the deal closed, onboarding, support and product improvement quietly became cost centres. The customer had already paid.
That dynamic shaped vendor behaviour whether companies admitted it or not.
Subscription SaaS changed that completely.
LTV only materialised if the customer stayed and customers only stayed if they continued seeing value.
That single shift pushed software companies toward much healthier behaviour:
- better onboarding
- continuous product improvement
- stronger customer success
- deeper focus on retention
It also gave buyers predictability, something that gets underestimated now in hindsight.
Finance teams could budget properly, Costs became operational instead of capital-heavy and commercial relationships simply became much more stable.
For the last 15 years there was an alignment, not perfect but still good, a veritable win win.
The vendor succeeded if the customer stayed and the customer stayed if they received value.
That alignment built enormous trust across the software industry and honestly a huge amount of SaaS growth came downstream from that trust. And this is why what is happening now feels less like a natural evolution and more like a structural break.
AI is not breaking SaaS pricing because the model failed. It is breaking because the unit of value it was built around, the human seat, no longer works cleanly. When software mainly acted as a productivity layer for people, more seats generally meant more value. It was not perfect, but it was a reasonable proxy.
But AI increasingly performs meaningful portions of the work itself and once that starts happening, charging per user starts disconnecting from the value actually being created.
Usage-based pricing is the obvious interim response. In cases with genuine marginal cost such as compute, inference and processing, it makes complete sense. But the market has already pushed well beyond that logic. Usage pricing is increasingly being applied where there is very little underlying incremental cost and buyers are starting to notice.
The customer logic becomes pretty simple:
“If my additional usage costs the vendor almost nothing, why am I suddenly paying significantly more?”
That is where trust starts weakening and so naturally the market starts moving toward outcome-based pricing.
And honestly, I understand why.
CFOs want to pay for results instead of access and procurement teams are tired of shelfware and underused licences. In all fairness the mutual risk-sharing sounds attractive.
And in some industries, outcome-based pricing genuinely works extremely well. Payments infrastructure is the clearest example, Stripe charges per transaction processed, the outcome is binary, and nobody disputes whether the payment went through. It works because the outcome is determined entirely by the banking network. The vendor has no ability to define or influence whether a transaction succeeded. Reality does that. The honest truth is that categories like this are rarer than people think and almost everything else being positioned as outcome-based right now has the vendor’s thumb somewhere on the scale. When the outcome is measurable and attribution is clean, it is probably one of the fairest pricing structures possible.
The problem is that most software outcomes are very hard to measure
Support and service outcomes may become even harder to measure cleanly for example.
On paper it sounds straightforward.
- Faster resolutions
- Higher CSAT
- Lower ticket volumes
- Higher automation rates
Even the metrics themselves are problematic.
Lower ticket volume might indicate a better product or it might indicate frustrated users who have simply stopped reporting issues.
Higher AI auto-resolution rates might look impressive operationally while customer frustration quietly increases underneath it. Resolution times can certainly be gamed, CSAT can be selectively optimised and escalation paths can be manipulated. Once commercial pricing starts depending on those outcomes, the incentives around measurement become even more distorted.
Humans already game it through deflection tactics, forced closure timers, and selectively routing easy tickets to the AI. Once commercial pricing pressure compounds that incentive, its going to get worse fast.
That is where outcome-based pricing becomes much more complicated than it first appears.
Take another practical example, one that perhaps sits close to home;
An AI SDR platform increases your pipeline by 20%.
Sounds straightforward, But what about quality and the conversion rates.
The vendor will of course point to pipeline growth and you will point to closed revenue.
Now multiply that disagreement across a large enterprise contract involving finance, procurement, operations and sales leadership.
That is the real measurement problem. And I think the market is massively underestimating how difficult this can become.
Attribution is almost always contested.
Your AI sales platform helped close more deals.
So did stronger reps.
Better market conditions.
Better marketing
Brand momentum.
Pricing changes.
A competitor making mistakes.
There is rarely a clean answer and both sides know that going in. The other issue of course which will also fast emerge is that the vendor increasingly control the measurement layer itself. Creating a core structural conflict which for the most part subscription SaaS largely avoided.
Your CRM vendor has very little financial incentive to inflate your pipeline numbers but an outcome-priced AI vendor absolutely does.
Whoever defines the outcome and measures the outcome holds enormous commercial power.
I do not think most buyers fully appreciate this yet.
Additionally timing comes into play, usage is measurable instantly whereas real business outcomes are not.
Revenue growth, retention improvements and operational efficiency gains can take months to materialise and even longer to attribute properly.
That lag has the potential to create disputes, commercial friction and indeed lead to contract complexity.
Ironically, many of the things that made SaaS commercially elegant start disappearing.
Outcome-based pricing does not remove the historical misalignment between vendors and customers but relocates it into measurement.
And measurement disputes are significantly harder and significantly more expensive to resolve than seat-count disputes ever were.
Outcome-based pricing doesn’t remove the misalignment that has always existed between vendors and customers. It relocates it into measurement, where it’s harder to detect, harder to dispute, and far more expensive to get wrong.
