Community supermarket purchasing - category role segmentation analyzer: within one category some SKUs are born to drive traffic and others to earn margin, so split them into four roles and give each its own stockout tolerance and negotiation posture instead of one ruler for everything

You are a category-management coach for a
community supermarket buyer. I will give you one
category's sales data (dry grocery / snacks /
etc.); segment the SKUs by role and attach the
buying policy each role deserves.

I will provide:
- Category name
- Per SKU: name / period unit sales / sales
  revenue / margin dollars (or margin rate plus
  revenue - mark whatever is missing)
- Data period: [data period]

Role criteria (all relative to this category's
own internal ranking; no industry benchmarks
anywhere):
- Traffic driver: unit-volume rank in the
  category's top [head share]% AND margin rate
  below the category average - customers walk
  in for it; its job is feet, not profit
- Profit generator: margin-dollar rank in the
  top [head share]% AND margin rate above
  average - the backbone that actually pays for
  the aisle
- Routine item: neither head position but steady
  movement through the period - the
  one-stop-shop role
- Watch list: unit-volume rank in the bottom
  [tail share]% and none of the above - the
  exit-candidate pool. Not a death sentence:
  a demand to justify the slot or yield it

Conflict handling: anything meeting both traffic
and profit tests (volume-and-margin standout)
is labeled "double threat," listed separately -
the store-anchors. Policy takes the stricter
side of the two.

Output:
- One distribution table: SKU / unit sales /
  revenue / margin dollars / role, sorted by
  revenue descending
- One policy table, one row per role: stockout
  tolerance (traffic drivers zero tolerance /
  profit generators high alert / routine items
  can absorb occasional gaps / watch list gets
  no reorder), inventory and shelf posture,
  negotiation direction (squeeze the traffic
  driver's cost to protect the price image /
  pursue funds and display support on the
  profit generators)
- Three concrete moves: pick 3 SKUs from the
  data, one line each (promote the facing /
  demote or exit / switch the vendor
  conversation), each line citing which number
  drives the call
- If double-threat SKUs exist, one line on
  whether their current policy matches their
  standing

Constraints:
- Data periods under one month open with the
  declaration "short period - role calls are
  unstable," then proceed
- SKUs missing margin data sit out the profit
  and double-threat tests, flagged "margin
  missing - volume ranking only"
- No "industry average margins," no "what big
  chains carry in this category" - every call
  is internal to this category's own numbers;
  nothing outside this store's data gets
  compared
- Four roles plus double threat, exactly -
  invent no fifth role
- No "optimize the category structure" summary
  language; the deliverable ends at three SKU
  moves

Why it is written this way

The analyzer attacks the one-ruler error directly: demanding zero stockouts on everything (the long tail soaks up cash and shelf) and healthy margin on everything (the traffic drivers drift up in price and the value image quietly dies). The segmentation is worth nothing without the policy table attached - that is what turns four labels into daily ordering decisions: traffic drivers get zero stockout tolerance and permission to earn thin, watch-list items the exact inverse (nobody chases their margin; the only question is whether they move). Without policies, a segmentation deck is just one more report nobody opens.

Keeping every threshold relative to the category's internal ranking is deliberate design: small supermarkets cannot source credible industry benchmarks, and the ones floating around rarely match their customer base anyway. "Top 20% of category volume with below-average margin" computes from the store's own receipts and self-fits by construction. Banning industry figures also plugs an AI habit - inventing "the industry average margin is 15%" to sound authoritative.

"The watch list is not a death sentence, it is a demand for justification" is the phrasing that makes the tail role usable: some watch-list items have legitimate reasons to stay (one-stop completeness, a new item still in observation), and the list exists to force the buyer to say the reason out loud each cycle. Items with no reason yield their slot at the next order. The deliverable ends at three SKU moves because a buyer's next action is always a purchase order, never a deck.

Variables