AI is changing what software does. If the work moves from people into the platform, the unit of value has to move with it.
This is the second in a three-part Overcast series exploring the shift from Software as a Service to Service as Software. In part one, we looked at Service as Software: When Your MediaTech Stack Starts Doing the Work for You*. Next, we’ll look at* Stop Buying Video Software. Start Buying Video Outcomes. Links will be added as the series is published.
There is a rather awkward contradiction sitting at the heart of enterprise AI.
We are promising customers they will need fewer people to do more work.
Then we send them a proposal based on how many people need a login.
Something has to give.
Imagine 20 people use a platform to process 10,000 assets. AI and automation allow those same 20 people to process 50,000.
The software has become dramatically more valuable. The vendor earns exactly the same amount.
Now imagine the customer can process those 50,000 assets with ten people.
The software has become more valuable again.
And under pure seat pricing, the vendor gets a pay cut.
That isn’t a criticism of SaaS pricing. Seats were an excellent proxy for value when humans did most of the work.
But AI is changing the unit of production.
And when the unit of production changes, I think the unit of value has to follow it.
Seats had a good run
Per-seat pricing wasn’t some historical mistake the technology industry has only just discovered.
It was clever.
When software made individuals more productive, the number of users broadly correlated with the amount of value being created.
More CRM seats meant more salespeople selling. More editing seats meant more editors editing. More project management users meant more of the organisation benefiting.
It was also wonderfully easy to understand.
Number of people × price = bill.
Procurement understood it. Finance could budget it. Vendors could forecast revenue without employing a small meteorological department.
The problem isn’t that seats have stopped existing.
It’s that software has started working when nobody is sitting in them.
The empty-seat problem
Think about a modern media operation.
Content arrives and is automatically enriched. Transcripts are generated. AI creates contextual metadata. Media is processed in the background. Workflows trigger automatically. Content is discovered and reused without somebody painstakingly maintaining every metadata field.
Much of that value can be created at 3am while every employee is asleep.
No additional user logged in.
No additional seat was purchased.
But the platform did more work.
This is the empty-seat problem: as software becomes more autonomous, human access and software value begin to separate.
And I think that distinction is going to reshape enterprise software economics.
Four ways to count
The obvious response is: “Fine. Let’s charge for outcomes.”
Not so fast.
There are actually four useful places to measure software value:
- Access — who can use it.
Seats, users, licences. - Consumption — what resources were consumed.
Compute, minutes, tokens, API calls, credits. - Output — what the software actually did.
Assets processed, workflows completed, checks performed, tasks executed. - Outcome — what changed for the customer.
Costs removed, turnaround reduced, risk avoided, capacity created.
Each moves closer to business value.
And each gets progressively harder to price neatly.
That is the interesting part of the transition. We’re not replacing one universal pricing model with another.
We’re finally being forced to ask what the customer is actually buying.
Nobody wants credits
AI credits are rapidly becoming the airline miles of enterprise technology.
Everybody has them. Almost nobody knows what they’re worth.
Credits are useful because vendors need a common way to meter expensive AI activities. But they solve a vendor problem more elegantly than they solve a customer problem.
Nobody arrives at work thinking:
“I’d really like to consume 50,000 AI credits today.”
They want to process more content. Find something faster. Remove repetitive work. Get a campaign live. Reduce cost.
Eventually the CFO asks the perfectly reasonable question:
“What did those credits actually do for us?”
If answering that requires a spreadsheet and three people from Customer Success, the pricing model is probably measuring the wrong thing.
Outcomes have problems too
Outcome pricing sounds wonderfully clean.
Pay for what works.
And where the result is discrete, measurable and substantially controlled by the software, it can be.
A transaction completed. A check performed. A defined process successfully finished.
But suppose a media platform helps launch a campaign two days earlier.
What’s that worth?
And, more awkwardly, who gets the credit?
The technology? The creative? The agency? The media plan? The product? A particularly successful TikTok?
The broader the outcome, the messier attribution becomes.
So I don’t think the future is simply “everything becomes outcome-priced.”
The better principle is:
Price as close to the work as you can measure credibly.
Sometimes that’s an outcome. Sometimes an output. Sometimes capacity or consumption.
And sometimes a platform fee remains exactly the right answer.
Predictability has value
There is another inconvenient fact that gets lost in some predictions about the death of SaaS pricing.
CFOs like knowing what things cost.
A perfectly aligned pricing model isn’t terribly useful if Finance has no idea whether next year’s bill will be £200,000 or £2 million.
Good enterprise pricing therefore has two jobs:
- Align price with value.
- Make the bill predictable enough to budget.
Those objectives don’t always sit comfortably together.
That’s why I expect the future to be hybrid: platform fees, capacity bands, usage, outputs and outcomes combined according to the type of work the software performs.
Pricing gets more sophisticated because the software has become more sophisticated.
That seems fair.
Divide by users? Don’t.
This is particularly relevant to Overcast MAX.
MAX isn’t designed to create value every time another person logs in.
It creates an operational layer across an organisation’s video estate: connecting media management, production, AI-powered discovery, workflows, governance and automation.
The platform carries the value.
Seats provide access to it.
That means dividing the annual MAX investment by the number of users misses much of the economics.
A better calculation asks:
What did it cost to operate this process before, and what does it cost now?
Include the people. Agency effort. Storage. Processing. Manual hand-offs. Duplicated work. Rework. Searching. Delays. And the additional capacity the organisation can now absorb without adding equivalent operational cost.
It’s messier than $X × 47 users.
Unfortunately, tidy maths and useful economics aren’t always the same thing.
YETI bought capacity
YETI is a good example.
As its demand for video grew, content was spread across hard drives, ageing equipment and departmental systems. Teams struggled to find existing material, collaborate globally and maintain a consistent view of their video estate.
Overcast MAX created a common environment covering more than half a million video files, from original RAW material through to finished campaign assets.
Half a million files sounds impressive.
But it’s an implementation metric.
The operational outcomes tell us more.
Yeti reported:
- Up to 80% lower storage costs.
- 10x faster processing from post-production to publishing.
- Immediate search across centralised video.
- Greater reuse and reversioning of existing footage.
- Less dependence on manual technical processes.
That’s infrastructure cost removed, time compressed and operational capacity created.
We can’t responsibly turn those figures into a precise ROI multiple without YETI’s complete cost base and commercial terms.
But we can ask a much more useful question than “How much did each seat cost?”
What would it cost to build an operation capable of producing the same result without the platform?
That’s the comparison enterprise buyers increasingly need to make.
Follow the work
I don’t think seats disappear.
They simply stop being the default answer to every pricing question.
Persistent enterprise platforms create continuous value through infrastructure, integrations, governance, availability and shared capability. A platform fee makes sense.
Other software performs discrete pieces of work with an obvious beginning and end. There, transactions, outputs or outcomes may make more sense.
And access can still be priced by seat where human access genuinely drives value.
The mistake is forcing all three through the same commercial model.
The pricing model should follow the work.
At Overcast, we increasingly believe customers should be able to buy specific video operations outcomes as easily as they currently buy cloud compute.
That’s a bigger shift than changing a price list.
It changes what we’re selling.
Count something better
If you’re evaluating enterprise software, try a different calculation.
Don’t start with the licence.
Start with the operation.
Take one commercially important workflow and calculate what it costs today: people, agencies, infrastructure, manual work, duplicated activity, rework and delay.
Then model what happens when software starts absorbing that work.
You’ll end up with a number considerably less tidy than “£X per user per month.”
You’ll probably end up with a much better business case.
Because AI is making the number of people operating software less interesting.
The more important number is becoming the amount of work the software operates.
And once work becomes the unit of value, enterprise software pricing has little choice but to follow.
Next: Stop Buying Video Software. Start Buying Video Outcomes.
FAQs
Is per-seat software pricing dying?
No. Seats remain useful where individual access correlates strongly with value. The problem arises when software performs substantial work independently of the number of users. In that world, seats measure access better than they measure value.
What is outcome-based software pricing?
Outcome pricing links some or all of the charge to a defined result: a process completed, cost avoided, risk reduced or another measurable business outcome. It works best when the result is clear, attributable and substantially influenced by the software.
What’s the difference between usage and outcome pricing?
Usage measures what you consume: compute, minutes, tokens or API calls. Outcomes measure what you achieved. Using twice as many tokens to achieve the same result may cost the vendor more; it doesn’t automatically create twice as much value for the customer.
Are AI credits outcome pricing?
Usually not. Credits are principally a meter for consumption. They’re useful commercially, but “we used 50,000 credits” tells a CFO remarkably little about what the business achieved.
What is capacity-based pricing?
Capacity pricing charges for an agreed level of operational capability or volume. It can provide a useful middle ground: price moves with the scale of the operation while the customer retains reasonable budget predictability.
Why does AI break the logic of per-seat pricing?
Because AI allows software to perform work without adding another human operator. When the same team can process dramatically more work—or a smaller team can produce the same output—headcount becomes a weaker proxy for software value.
Should everything eventually be outcome-priced?
No. Broad business outcomes are often difficult to attribute. The better rule is to price as close to the work and value as can be measured credibly, while maintaining enough predictability for customers to budget.
How should procurement compare outcome-led and seat-priced software?
Compare operations, not just licences. Include people, agencies, infrastructure, manual intervention, duplicated work, rework, delay and the capacity needed to achieve the same result. The cheaper licence doesn’t necessarily create the cheaper operation.
How should media teams calculate ROI?
Start close to the workflow: processing time, search time, manual interventions, asset reuse, infrastructure cost, turnaround, rework and additional volume handled. Attribute broader revenue outcomes only where the relationship can genuinely be defended.
Why does Overcast MAX have platform-led pricing?
Because MAX creates value across the video operation, not simply when an individual logs in. The platform represents the shared operational capability—media, workflows, intelligence, automation and governance—while seats provide access to that capability.
Will enterprise software become completely transactional?
Unlikely. Persistent platforms and discrete automated tasks create different kinds of value. That’s precisely why hybrid models are likely to become more important: platform economics for persistent capability, with usage, output or outcome economics where the software performs measurable work.
Still have questions? Contact our team
