CST Grading · For Ross's review

Break "Product" out of Q&A

Make product a real, graded category so we can train Pam on it directly.

Proposal from Anthony · please review and mark each boundary rule agree or change · nothing changes until you sign off

Reviewing as Mark each rule Agree or Needs change, add a note, and hit Save. Your answers go straight to Anthony, no Slack needed.

TL;DR

Product questions (sizing, fit, materials, specs) are where Pam is most likely to make something up, and right now they are hidden inside the Q&A score so we cannot see or train them.
Proposal: make PRODUCT its own graded category, slim Q&A down to account and informational, and make GENERAL a true catch-all.
The boundary rules below are yours to set. That is the whole reason for this page.
Your audit grader and the shared grading sheet have to move at the same time as ours, or the two blind graders stop agreeing.
Goal stays 95%, account-wide and per category.

Why this matters

Product is the #1 fabrication risk. Pam inventing a fit, a fabric, or a spec is the worst kind of miss. Today it is buried in Q&A, so a product weak spot is invisible on the scoreboard.
The product buckets are empty today. One collected ticket total, because the classifier currently sends "what size should I get" to Q&A on purpose. That was your call. We want to change it, with you.
The new dashboard grades every category vs 95% across last night, 7-day, and 30-day. Product deserves its own row so you can train it on its own.

Today vs proposed

TODAY

5 graded intents. Product is just a label inside Q&A.

WISMO
ORDER ISSUE
RETURN ISSUE
PAYMENT
Q&A → sizing, materials, fit, account, affiliate, solicitor, taxes, password
PRODUCT CATEGORY → empty (label only)
GENERAL → empty (label only)

PROPOSED

Product becomes real. Q&A slims. General becomes a true catch-all.

WISMO same
ORDER ISSUE same
RETURN ISSUE same
PAYMENT same
Q&A → account, affiliate, solicitor, taxes, password, gift cards, donations product removed
PRODUCT CATEGORY → sizing, fit, materials, specs, product-line questions now graded
GENERAL → ambiguous or multi-intent only catch-all

The new PRODUCT playbook is seeded from your existing Q&A sizing and product write-ups, so it launches with real content, not from scratch.

The boundary rules we need you to set

These decide where each ticket lands. Please mark each one agree or change. The two marked your call are genuine judgment calls.

"What size should I get?" / fit / "true to size?" your call

→ PRODUCT CATEGORY (today it is Q&A / sizing)

Materials, fabric, care, "does this glow under UV?", product specs

→ PRODUCT CATEGORY

Product-line questions (tops, bottoms, bodysuits, shoes, chains/harnesses, hats, outerwear, masks, accessories)

→ PRODUCT CATEGORY

"Exchange this for a bigger size" unchanged

→ stays RETURN ISSUE / exchange. Asking about size is product. Swapping the item is a return.

"Is the M in stock?" / restock your call

→ proposed: stays ORDER ISSUE / oos. Tell us if you want it under PRODUCT instead.

Q&A keeps

account, affiliate / collab, solicitor, taxes & duties, forgot password, gift cards, donations

GENERAL = genuinely unclassifiable or multi-intent only

If GENERAL volume ever grows, that is a signal to spin out a new real category, not a dumping ground.

What changes under the hood, and the one cost

We add PRODUCT CATEGORY as a 6th graded intent in the classifier and move the sizing and product subcategories under it.
Your audit grader and the shared "Pam Grading Audit" sheet (Grading Rubric tab + Playbooks tab) update to the same tree at the same time, or our two blind graders stop agreeing. We do that lockstep step together right after you approve.

One-time cost

The per-category 30-day history gets a break point on the day we re-slice (before and after are not comparable). The overall account score is unaffected: same tickets, same grades, just re-labeled.

Overall thoughts

Anything not covered by the 7 rules: naming, sequencing, concerns, or a flat yes.

The ask

Go through the 7 boundary rules and mark each agree or change. Once you are aligned, we update your grader and the shared sheet to match, then it goes live. Account-wide and per-category goal stays 95%.

Level 1 · Static Doc · Generated by Claude for Anthony