For 15 years, seats worked.

When you sold access, expansion meant more headcount and it was all quite simple.

Then AI happened. Your product doesn't just give people a tool to use, it does the work. A single user with AI features can generate 10x the output of a team of five without AI. So when your pricing page says "$59/seat/month," your prospect is asking: why am I paying for access when I'm buying results?

And if you're a CRO, you should be asking: "why are my reps still quoting licenses when the product delivers outcomes?"

There are many possible issues, but one that we've come up with again and again is that companies run two disconnected revenue systems. PLG customers through Stripe, enterprise deals through Salesforce CPQ.

At the moment a self-serve customer wants to upgrade to a committed spend with seats plus credits plus usage tiers, the whole thing collapses into a spreadsheet and a Slack thread with finance and the deal-desk.

PLG/SLG is a false binary. It really isn't one!

Very few companies scaled on pure PLG. Eventually, customers' CFOs want control and formal contracting. And on the other side, enterprise customers don't want to talk to a salesperson for routine expansion, they want self-serve interfaces for adding volume. The motion is quite fluid between both, sometimes in the same account, same period.

But the billing infrastructure forces a binary that doesn't match reality. We called this the split stack problem.

That's because when billing can only model seats, reps can only sell seats. The system literally can't express "200K credits at $0.03/credit with a 15% volume discount, plus 10 seats, plus a base platform fee, with credit rollover into Q2". It's too complicated for those systems.

Sales reps will either be creative and push deal desks to approve (and run manual billing), or they'll sell what the system can quote and leave the AI value off the invoice entirely.

When switching costs disappear, only value alignment remains

Seats created switching costs, because every user configured and every workflow customized was friction working in your favor.

It's now very clear that AI is dissolving that friction. Products are increasingly interchangeable at the user layer. The new generation of AI-native tools can onboard in hours.

And at worst, you can vibe-code a connectivity layer to migrate.

The risk is that customer dissatisfaction with the pricing model (not the software itself) is also now becoming a main driver of churn. When the product is good but the pricing feels misaligned, customers don't complain. They just leave - because it's easier.

Time to align revenue to value

This will need to happen in several steps:

1. Instrument value before you price for it

Build visibility into what your product delivers per customer.

Usage patterns, AI outcomes, inference costs, and more - these have to be measured and demonstrated.

You can't price for value you can't measure.

Bring your CFO in early: teams that model new pricing as "different products" instead of different plans on the same product end up configuring their financial system backwards and that really hurts (trust me).

2. Introduce credits alongside seats, not instead of them

PricingSaaS data shows 126% year-over-year growth in credit-based models.

I don't think credits are a full replacement for seats, they're a complement - at least for the time being.

Seats anchor the deal, credits capture the value (if done correctly). There are some exceptions to this, but this is how it's perceived at the beginning of 2026.

The evolution is toward committed spend: the customer commits a dollar amount and pays as they go, gradually adopting new services without re-contracting. It's the model OpenAI, Anthropic, Cursor, and Harvey have adopted, and that's not because they figured out perfect pricing. It's because they built infrastructure flexible enough to change it.

Lovable completely got rid of seats and moved only to credits.

You also must allow credits must be pooled across the org or across teams. This is a ledger design decision, not a gift card hack!

3. Never cancel-and-replace

Yes, PLG customers expect transparent pricing. When they move to SLG, complexity grows but from their perspective, they shouldn't feel like they're canceling one subscription and replacing it with a completely different one.

Structure transitions as incentives: new model offers more value with better economics. No forced migration. And because the motion might reverse, the infrastructure should support movement in both directions.

4. Arm reps with quoting that can express the hybrid deal

If your rep opens Google Sheets to model "seats + credits + volume tiers + annual commit" the deal takes weeks instead of "within the same call". Deals ghost during that lag.

Companies are switching to unified platforms because they literally cannot execute their pricing strategy with disconnected tools.

5. Make expansion follow adoption

When credits are the growth vector, expansion happens naturally.

The emerging pattern: embed professional services hours inside committed-spend contracts as a usage product.

Customers add onboarding help or consultative support without re-contracting. In the AI era, this "GTM engineering" function is becoming the expansion vector reps didn't know they had!

The bottom line

The seat model trained an entire generation of sales teams to sell access. Demos focused on features. QBRs focused on login rates. Expansion followed org charts. None of that translates to AI.

Your reps can only sell what your systems can quote. And right now, most systems can only quote seats. That's the bottleneck. Not the sales team. Not the product. The revenue architecture.