Flat markup
You add a fixed percentage on top of underlying AI cost and bill the customer. Simple line item, simple support conversation, simple gross margin math.
Revenue share
You charge the customer a transparent rate (often "cost + X%") and disclose underlying cost. Procurement loves it; it gives them an audit trail. Heavier ops cost on your side.
When to pick which
- SMB tier — flat markup. They want one number per month.
- Mid-market — flat markup, with usage explorer for transparency.
- Enterprise — revenue share. Procurement wants the math.
- BYOK customers — neither. Charge a platform fee; they pay AI directly.
One thing this playbook is not about
Command+K itself does not support BYOK: every call the widget makes runs through our managed AI Gateway on prepaid credits, which is what keeps per-model cost reporting exact. The advice above is about how you price your own product, not about how you pay us.
The 20% floor
A 20% markup barely covers retries, caching infra, support, and a per-user cost cap. Most mature operators land at 30-40% — you're selling reliability and reseller economics, not raw tokens.
FAQ
- Which model do most operators pick?
- Flat markup is more common for SMB tiers (simpler invoice line). Revenue share dominates Enterprise where the customer wants to see the math.
- Is a 20% markup competitive?
- It's the floor. 30-40% is normal once you bundle inference, caching, retries, and a per-user cost cap. You're selling reliability, not raw tokens.
- Can I mix both?
- Yes. Flat markup on baseline usage, revenue share on overage. Anchors the invoice but rewards heavy users.