SaaS Pricing in 2026: Beyond Per-Seat

Per-seat pricing was the perfect model for the workforce-collaboration era. Every new user was both a cost line for the customer and a revenue line for the vendor. Easy to explain, easy to forecast, easy to upsell.

In 2026, it's quietly falling apart — and not because of any single dramatic event. It's a slow squeeze from two directions at once.

From below: AI agents now do work that used to require seats. A team of five with a good AI stack outproduces a team of fifteen from 2022. Buyers are looking at their per-seat invoices and asking, reasonably, why they should pay for users that no longer exist.

From above: AI inference costs are real, and they don't scale with seats. A single power user running a heavy AI workflow can cost a vendor more than ten light users combined. If your pricing doesn't reflect that, your margin gets quietly eaten while your top-line stays flat.

Here are the four models I see replacing per-seat in 2026, and the trade-offs of each.

1. Outcome-based pricing

You charge per resolved ticket, qualified lead, generated report, processed invoice — whatever "the thing the customer actually values" is. Intercom's Fin and a wave of AI customer support products have made this the headline model.

Pros: the strongest possible alignment between your revenue and your customer's value. Procurement loves it because the ROI math is obvious.

Cons: brutally hard to forecast on both sides. Hard to instrument cleanly without disputes. Requires you to be willing to take risk on outcomes you don't fully control.

2. Usage-based pricing

You charge for compute, tokens, API calls, storage, transactions — the underlying resource. Snowflake, Twilio, and most of the AI infrastructure layer use some flavor of this.

Pros: easy to instrument, scales naturally with customer growth, aligns price with cost.

Cons: punishes power users (the people you most want to delight), creates billing anxiety, and can produce wild monthly variance that procurement teams hate.

3. Platform fee plus consumption

A predictable monthly platform fee plus metered overage above a baseline. Stripe, Datadog, and most modern infrastructure vendors land here.

Pros: customers get predictability for budgeting; you get upside as usage grows. The best of both worlds for most B2B contexts.

Cons: requires you to size the baseline carefully — too low and customers feel nickel-and-dimed; too high and you leave money on the table.

4. Tiered value pricing

You price based on a proxy for the customer's outcome size: revenue, employees processed, properties managed, transactions facilitated. Toast, ServiceTitan, and most vertical SaaS use this.

Pros: simple to explain, scales with the customer's success, easy to forecast.

Cons: the proxy is only as good as its correlation with the actual value you deliver. Pick the wrong proxy and your largest customers will quietly underpay you while your smallest ones overpay.

How to pick

The right answer depends on your value metric — the single thing that, when it goes up, the customer is unambiguously winning. Ask:

1. What's the one number that goes up when the customer wins? That's your candidate metric.
2. Can you measure it cleanly without disputes? If not, you're going to pick a proxy.
3. Does it scale with your cost? If your cost scales with tokens but you charge by seat, you're building a margin time bomb.

The wrong answer is staying on per-seat because it's what you've always done. The buyer's mental model has moved on. Yours should too.