Skills CX operations
Incentive design for support
Attaching pay to a metric does not reward the underlying behaviour. It rewards optimisation of the measured proxy. Goodhart's law is not a cautionary tale — it is the default outcome unless the scheme is designed around it.
npx rulebase-skills install cx-incentive-designWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “incentive plan”
- “agent bonus scheme”
- “SPIFF design”
- “pay for QA score”
- “what metric should we bonus”
How it works
The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.
Step 1: name what you are actually trying to change
Write the behaviour in plain language before touching a spreadsheet: "Agents maintain quality while handling a fair share of volume".
Step 2: classify metrics by incentive survivability
No single metric survives alone. Every primary measure needs at least one guardrail — a metric that must stay within bounds or the bonus zeroes. Typical pairs: CSAT primary → QA floor and complaint-rate ceiling.
Step 3: catalogue gaming modes before launch
For each proposed primary metric, list how a rational agent maximises pay while harming the customer or the business. If you cannot list modes, you have not understood the metric.
Step 4: design the scheme structure
Principles that hold up in practice: Use periods long enough to survive noise.
Step 5: governance and review
Before launch, agree: Sample design, Mix adjustment method, Dispute route, Review cadence, Sunset clause.
Related skills
Free and open source, and vendor-neutral — it reads the conversations from whichever helpdesk you already run. Browse all 149 skills · connect your helpdesk over MCP · source on GitHub
