Survey · anonymous · two minutes

Promised vs. actual: what did AI really resolve?

Vendors publish headline resolution rates. Buyers rarely publish what they actually saw. If you run support, success or CX and have AI handling customer contacts, add your numbers. Ranges are fine. When you submit, you'll see how your numbers compare with everyone's so far, and get a one-page checklist: seven questions to ask before you believe any AI resolution rate. Results will be published in Proof of Work, in aggregate only.

What resolution or deflection rate were you promised, or did your business case assume?

Required.

What did you actually see once it was running?

Required. Use the rate your team reports, whatever it counts.

How long has it been live?
How is "resolved" defined?
Before AI, how much did customers already solve on their own?

Help center, in-product help, onboarding.

Did you follow the vendor's implementation playbook?
Who are your customers?
Roughly how many customer contacts a month?
Did support headcount change because of AI?

Talk it through: 30 minutes, no pitch.

Bring the decision on your desk. You'll leave knowing whether your data can answer it.