Good Pricing Grows With the Value You Deliver
At our portfolio company Notehouse, AI agents run our recurring BI check-ins, do SEO research and manage parts of our outbound campaigns. This kind of work gets done every week, with some initial time investment upfront and regular check-ins since then.
If you think this through, you will notice that it breaks the oldest assumption in how software and services are priced: that value scales with human time.
Agencies and consultants bill hours, because hours were the best available proxy for value. SaaS companies did the same thing, except they would typically charge per seat.
Yet both models share the same hidden assumption: that value grows with headcount. But AI decoupled the two.
For a seat-based software business, that's a strange position to be in: your customer gets more value out of your product while paying you less, because they need fewer seats. Your pricing now punishes you for your customer's productivity.
I call this the efficiency penalty.
The same squeeze hits anyone selling time. If an AI agent drafts the contract, the audit report or the campaign in minutes, billing by the hour stops making sense to the customer.
They don't care how long it took. They care more that the deliverable exists and is correct.
The right pricing strategy grows your business
I think pricing will shift toward two models:
Outcomes. You pay for the thing you actually wanted: a finished document, a resolved support ticket, a qualified lead. This is the cleanest alignment possible. The customer's cost tracks the customer's value, and the vendor is incentivized to deliver results, not to maximize usage or seats.
Consumption. You pay for what you use: tokens, runs, tasks. This works, but it carries a trap I've written about before. If your price tracks your compute cost too closely, you're not running a software business anymore, but reselling compute with a markup.
Overall, I think, as founders and operators, these are the next steps we should be looking at to adapt our software businesses:
- Understand where value is created. Ask what your customer is actually buying from you. If the honest answer is "results" but your invoice says "hours" or "seats", the gap will become wider.
- Test new pricing models on new customers. You don't have to migrate your existing base. Onboard new customers on an outcome or consumption plan and compare.
- Pricing is never done. A good pricing strategy is never static, but changes over time and adapts to your business.
The shift won't happen overnight, but the businesses that course-correct first will have an easier conversation with customers who just cut their team in half and doubled their output.
When did you run your last pricing test?
Does your pricing survive AI?
How much does one customer pay you per month?
Your average monthly revenue per customer, across all seats or hours billed.
What does it cost you to serve that customer per month?
Hosting, compute, support, third-party services – everything that scales with customers.
How much of that cost scales with usage?
Compute, AI tokens, API calls – the part of your costs that grows when your customer does more.
The same customer, one year into AI adoption
Scenario from this post: your customer halves their team and doubles their output. Your usage-based costs double with their output, your fixed costs stay flat.
Growing Ventures