Recommendation: move to a hybrid (platform fee plus usage) and raise prices

Pricing model

Switch from flat per-seat to a hybrid: platform fee plus metered usage. Two of your facts point the same way:

• Usage varies 20x between customers. The skill's decision tree says to use usage-based pricing (or a hybrid) when usage varies more than 5x. At 20x, a flat $20/seat means light users overpay relative to their value and heavy users get a large subsidy.
• Value tracks usage, not headcount. An API product's value comes from calls, work automated, or engineering time saved. A customer can consolidate onto a few seats and run heavy volume, which is the "usage goes up, revenue doesn't" failure mode of seat pricing.

Structure

• Free tier: set the limit just below where production usage starts. Hobbyists stay free and production users convert. I need your usage distribution to set the number.
• Team (self-serve): a platform fee plus a usage rate, with a few included units so the bill is predictable.
• Enterprise (sales-led, unpublished): an annual commitment with usage bands. Gate SSO, RBAC, audit logs and SLAs here, and offer Good/Better/Best options. Don't publish a number.
• Seats: drop them as the main meter. If you keep them, use them only as a minimum or an admin/feature gate.

The platform fee ensures every customer covers cost to serve. The variable part captures the value of your 20x heavy users.

Should you raise prices? Yes.

Large customers say you save them about $200K a year. Your price is $240 per seat per year, so the value ratio depends on seat count:

For any large customer under about 200 seats, you're above the skill's 5x "underpriced" line. Below about 40 seats you're above the 10x "massively underpriced" line. The skill's reference case moved from $15K to $45K a year on a similar $200K+ value and lost zero enterprise customers.

A $20 price also tells enterprise buyers you're a small-team tool. If you want larger customers, the price itself works against that.

How to do it

1. Anchor on value, not cost. A target for large accounts is roughly 10–20% of the $200K savings, or $20–40K a year. That's a 3x value ratio at the top of the range and is still healthy.
2. Raise via repackaging, not just a number. Put SSO, audit logs, support and SLAs in the enterprise tier so the higher price comes with new value.
3. Grandfather existing customers for 12–24 months. Give notice and a date. This also creates urgency for prospects.
4. Add a 5–10% annual escalator to enterprise contracts.
5. Communicate it as investment. Don't apologize.
6. Pilot first. Start with new customers and the largest accounts, and watch win rate and churn before rolling out wider.

Check before you commit

• Seat counts for the $200K customers. This sets how big the increase can be.
• Whether the $200K is validated. It's self-reported. Confirm it with two or three customers using the alternative-cost math (hours × rate, or build-in-house cost).
• Win rate and price objections. Above 40% win rate with no recent pushback means raise now. If you're losing deals on price, segment first and don't raise across the board.
• Marginal cost per call. Keep enterprise discounts above cost to serve plus 40% margin.
• Heavy-user migration. Model what each existing customer would pay under the new structure. Heavy users will see big increases, and light users may see decreases, which is fine for a free or entry tier.
• Bill predictability. Offer included units, commit bands or spend caps so customers don't get surprise bills at month-end.

If you share seat counts, the usage distribution and cost per unit, I can draft specific tier prices and the free-tier limit.