Most AI customer support platforms charge a per-resolution fee on top of the base subscription. The pitch is intuitive: you only pay for AI when it works. The reality is that this pricing model penalizes you when AI works well, which is the opposite of what you want.
The math gets worse as you scale
A team with 1,000 monthly conversations and a 40% deflection rate is paying for 400 AI resolutions. At $1 per resolution, that is $400 a month — often more than the base seat price. Double the volume and the AI cost doubles. Improve the deflection rate and the AI cost goes up, not down. That is exactly backward.
Incentives matter
When a vendor charges per resolution, they are incentivized to count loosely — every AI message becomes a "resolution," even when the customer was not actually helped. We do not want that incentive in our business, and you do not want it in your vendor.
What we do instead
AI agents are included in every Savena plan at no extra cost. Pricing scales with conversation volume — the same metric your business already plans against — not with how often AI works. The cost is predictable, and the incentive is aligned: we win when AI deflects more, you win when AI deflects more, and the customer wins when their question gets answered fast.
When per-resolution pricing makes sense
To be fair: per-resolution pricing can be the right model for very large enterprises with predictable volume, sophisticated procurement, and the ability to negotiate favorable rates. For SMB and mid-market teams, it almost never is. The math punishes you precisely when AI is working.
Bottom line
Predictable pricing is not a feature, but it is a quiet driver of which AI investments actually pay off. If you are evaluating platforms, ask the vendor to model your bill at 30%, 50%, and 70% deflection. The shape of that curve tells you whether their pricing rewards or punishes your success.