Skills Quality assurance
Quality across languages
Multilingual support quality data has a structural problem: the measurement is usually weaker in exactly the languages that score worse, so a real difference and a measurement artefact look identical in the output.
npx rulebase-skills install cx-multilingual-qualityWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “compare quality across markets”
- “our Dutch team scores lower”
- “quality by language”
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.
Establish instrument parity before comparing anything
Do this first. If it fails, the comparison does not happen.
Separate the four things that produce a language gap
When a market scores lower, it is one or more of these. They have entirely different remedies and the analysis has to distinguish them: Instrument, Mix, Rubric fit, Genuine capability.
Check the supporting material, not just the agents
A market scoring low very often has less to work with: Macros and templates, and whether they were translated well or at all, Training material availability, Escalation paths staffed in that language and timezone.
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
