Stop reporting deflection to your board
Deflection records a customer who gave up as a success. If it is the headline figure in your AI report, your board is looking at the wrong measure.
Read →
AI Support Health Check
Find Out What Your AI Agent Is Really Resolving
A two-week, fixed-fee review for leadership teams with an AI support agent in production. You’ll see what it actually resolves, what it costs you in customer satisfaction, and the five changes that would improve both.
Is this you?
Your dashboard says it’s working. Your customers may not agree.
For organisations with an AI agent live for three months or more, whichever platform you use.
- Your vendor reports high deflection or “resolution”, but ticket volumes haven’t really fallen
- CSAT has dipped, or nobody can say how AI-handled conversations score
- The board wants evidence of a return on the AI spend
- Nobody has reviewed the help centre since the AI went live
- Customers complain about repeating themselves after the handover to a person
What I check
Six checks, using your own data
Resolution
Your vendor’s “resolved” figure against a true resolution rate, counting customers who come back about the same issue within seven days.
Satisfaction
CSAT split three ways: AI only, AI to human, and human only. The middle group shows where handovers fail.
Knowledge
The 20 help articles your AI relies on most, checked against your product as it is today, each with a named owner.
Handover
A sample of conversations passed from AI to your team, checking whether agents get the full history.
Scope
Which contact reasons the AI handles well, and which should return to your team for now.
Cost and reporting
Cost per issue instead of cost per contact, and what your board sees today.
How it works
Two weeks, read-only, about 4–6 hours of your team’s time
01
Before we start
You share a data export and read-only access to your helpdesk and AI reporting. I send a one-page checklist of exactly what’s needed.
02
Week one: measure
Recalculate resolution, split CSAT, pull the most-used articles and sample handovers. Three short interviews with your support lead, a team lead and the owner of the AI tool.
03
Week two: conclude
Check the top articles, map which contact reasons stay with the AI, and work out cost per issue.
04
Report and briefing
A findings report and a 60-minute briefing for your leadership team, then a follow-up 30 days later.
What you get
Clear numbers your board can trust
- A one-page AI scorecard: true resolution, repeat contact rate, CSAT for each group and cost per issue, each rated red, amber or green
- Your vendor’s figures set against the real ones, explained in plain terms
- A written definition of “resolved” you can adopt
- A list of contact reasons to keep with the AI and to return to your team
- A help centre fix list with an owner for each article
- The five changes that would make the biggest difference, and how to measure each one
Fixed fee
£3,950
One AI agent, one language. Includes the findings report, a briefing for your leadership team and a 30-day follow-up.
More than one agent, channel or region? I’ll quote after the intro call.
- Read-only: I don’t change your setup, so there’s no risk to live service
- Vendor-neutral: I don’t sell or implement AI tools, and take no referral fees
- Built on your numbers, not industry averages
The thinking behind it
All Practical CX articlesNobody owns your help centre, and your AI shows it
If your AI support agent is underperforming, look at your help centre before you look at the model.
Read →The CSAT drop nobody puts in the business case
Most AI business cases account for savings but overlook a key cost: customer satisfaction.
Read →Is Your AI Really Resolving?
A 30-minute conversation about your AI agent and whether a Health Check would help. No obligation.
Book an intro call