Support automation ROI, measured honestly
Support automation ROI is simple in theory: the value of the tickets the AI resolves, minus what the AI costs. In practice most teams get the number wrong, because they measure deflection instead of verified resolution and forget that a per-resolution price rises with the very volume that is supposed to prove the ROI. This page shows how to calculate it honestly.
Why most ROI numbers are wrong
Gartner's 2026 data is sobering: only 24 percent of service leaders reported positive financial returns on AI, and the report notes most optimized the wrong number. The most common mistake is counting deflection, tickets that did not reach a human, as resolution. A deflected ticket where the customer gave up has no positive value. It often has negative value, because the customer comes back angrier or churns.
Gartner also found only 9 percent of customers say a chatbot fully resolved their issue, and chatbot-only CSAT is 28 percent versus 82 percent for humans. If your ROI model assumes every deflection is a win, it is measuring a number that does not exist.
The honest ROI formula
Measure value on verified resolution, not deflection. A verified resolution is one confirmed by survey, confirmation, or no 72-hour recontact. The formula is:
- Value: verified resolutions times the fully loaded cost of a human handling that ticket (labor, tooling, overhead).
- Cost: the AI platform cost plus the human cost of everything it did not resolve, plus admin and setup time.
- ROI: value minus cost, divided by cost. Watch reopen rate as the honesty check, because reopens turn a counted resolution back into a cost.
How pricing model changes ROI
Here is the trap. Under per-resolution pricing, the cost side of the equation grows with the resolution count, so as you scale the volume that is supposed to prove ROI, you also scale the cost that eats it. Under flat pricing, cost is fixed per band, so every additional verified resolution is pure upside. Flat pricing is not just cheaper on average, it makes ROI improve as automation improves instead of flattening out.
How Resolve prices instead
Resolve decouples price from resolutions entirely. You pay a flat monthly rate for a band of inbound contact volume, with unlimited seats and unlimited AI resolutions inside that band. The published tiers are Launch at $250 a month for roughly 750 inbound contacts, Growth at $900 for roughly 3,000, and Scale at $2,500 for roughly 10,000, with a custom flat Enterprise tier above that. If you go over a band, you bump to the next band, never a per-unit surprise.
The promise is one sentence: your bill never goes up because the AI did its job. It only changes when your business grows. That is the opposite of a per-resolution meter, where the better the automation performs, the more you owe.
Frequently asked
Why do most teams see poor ROI on support AI?
Gartner 2026 found only 24 percent of service leaders reported positive financial returns, largely because they optimized the wrong number, counting deflection as resolution. A deflected ticket where the customer gave up has no real value. Measure verified resolution and watch reopen rate instead.
How should I calculate support automation ROI?
Value equals verified resolutions times the fully loaded human cost per ticket. Cost equals the platform price plus the human cost of unresolved tickets plus admin time. ROI is value minus cost over cost. Use verified resolution, not deflection, and treat reopens as negative value.
Does per-resolution pricing hurt ROI?
Yes. It puts your cost on the same axis as the volume that proves your ROI, so scaling automation scales cost. Flat pricing fixes cost per band, so every added verified resolution is upside and ROI improves as automation improves.