Inventory Positioning Analysis — Demand, Stock, Baskets & Freight

Contents

Executive Summary
Flow Map
Stocking Recommendations
Basket-Affinity Analysis
Freight Economics
Delivery-Time Impact
Investment & Payback
Seasonality
Address-Book Audit
Methodology & Limitations
Other Factors & Next Steps

1 · Executive Summary

1,461

US fulfillments analyzed (12 mo)

~28M

Excess unit-miles / year

$43,971

Shipping billed to customers (12 mo)

Unknown

True carrier cost — not in NetSuite

2–4 days

Faster delivery, ~600 shipments/yr

Phase-1 capital (transfers only)

The finding in one sentence: the two coasts are fulfilling each other's backyards. Boston ships BOM100 to San Francisco customers (3,500 units/yr across a 2,692-mile gap) while San Francisco ships PCB100L to Newark, NJ (3,410 units/yr) — despite the Queens warehouse sitting 15 miles from that customer.

The economics are a margin story, not an expense story. Shipping charges billed to customers ($43,971/yr) come from a largely flat-rate table (~$31 per shipment appears on wildly different routes), while carrier cost certainly scales with distance and zones. Since shipping revenue is distance-insensitive and carrier cost is distance-sensitive, every needless cross-country mile comes straight out of margin — invisibly, because actual carrier cost is not captured in NetSuite. The fix is the same regardless of the exact magnitude: position stock near demand.

Top three moves:

Transfer the SAF200 6-SKU kit + PR100-A to Denver. All from San Francisco's deep overstock. Zero capital. Serves Park City (355 shipments/yr — the steadiest demand stream in the network) and most of St Louis's basket from 785 mi instead of 1,735.
Stock the PCB100L 4-SKU basket in the Northeast (Queens or NJ) — PCB100L + SMK001 + KAP001 + COP001, which co-ship in 69 of 69 orders. After resolving the §8 lifecycle question (shipments stopped in April).
Audit BOM100 & SBP100S at Boston for transferable excess to San Francisco — the SF-metro cohort is the single most expensive-looking flow (avg $108/shipment billed, 2,692 mi).

2 · Flow Map — Current vs. Recommended

Solid red arcs = wasteful current flows (ship-from → destination). Dashed green = recommended source. Blue squares = warehouses; circles sized by annual units.

Warehouse Demand cluster (annual units) Current flow (wasteful) Recommended flow

3 · Stocking Recommendations

Recommendations are expressed as baskets, not single SKUs (see §4 for why). Investment = ~3 months territory demand at item average cost.

Priority	Basket (anchor + companions)	Stock at	Annual demand	Transit gain	Capital	Notes
PHASE 1	PR100-A (+ I360-Impeller, NS-SCCT, NS-SA-EIA-Inv already at Denver)	Denver	3,194 u / 355 shpts	−1 day	$0 — transfer ~800 u from SF (7,665 avail)	Companion coverage already good. One gap: I360-Assembly (2234) is at zero network-wide — purchasing issue, not positioning.
PHASE 1	SAF200 kit: SAF200 + CTL200 + FRM200 + BLD200 + FWH200 + FSP200	Denver	292 u anchor + kit	−2 days	$0 — all 6 SKUs transferable from SF depth (BLD200: 4,932; FWH200: 9,385; FSP200: 1,620)	Ships as a kit in 28/28 co-orders — move all six or don't move any.
PHASE 1	SBP100S + ASY107S (co-ship 35×)	San Francisco	480 u	−4 days	Likely $0 — audit Boston excess first	COM002S/COM003S companions already at SF (39/120).
PHASE 2	PCB100L basket: PCB100L + SMK001 + KAP001 + COP001 (co-ship 69/69 orders)	Queens	3,410 u anchor	−4 days	~$110K (4 SKUs; anchor $99K)	GATE Resolve April demand-stop (§8) first. Companions currently SF-only (SMK001: 13,680; KAP001: 682; COP001: 9,000) — likely transferable, cutting capital substantially.
PHASE 2	BOM100 + COM001/2/3/4 (co-ship 89×)	San Francisco	3,944 u anchor	−4 days	$34.7K anchor; companions mostly at SF but COM002 (96) and COM004 (85) are thin — top up	Highest billed-freight flow ($108/shipment avg from Boston).
PHASE 2	SSP100CTP/CTSM + SSP100P/SM assemblies	New Jersey	1,200 u	−2 days	$13.6K	Packaging companions (cartons 7,480 / labels 49,920) already at NJ. But demand is project-spiky (§8) — consider transfer-on-demand instead of stocking.
PHASE 3	CTM200 + OPR001/OPR002 (62×)	Queens	410 u	−4 days	$38.4K + companions	OPR001/002 at SF (157) and Boston (40), zero at Queens. Service play only.
PHASE 3	Gauge family (items 105–122)	NJ / Queens	~470 u / 8 SKUs	−3 days	~$82K	High unit cost, thin volume — weakest case, defer indefinitely.

4 · Basket-Affinity Analysis

For each recommended SKU, 12 months of sales orders were queried for items appearing on the same order (≥3 co-occurrences). The result shapes the recommendations fundamentally: these items are baskets, not loners — and stocking an anchor without its companions converts clean single-origin orders into permanent 2-origin splits.

Anchor	Ships with	Orders together	Companion stock at target?	Verdict
PCB100L	SMK001, KAP001, COP001	69 / 69 — always	None at Queens/NJ — all three SF-only	4-SKU MOVE Anchor alone would split every order
SAF200	CTL200, FRM200, BLD200, FWH200, FSP200	28 / 28 — always	None at Denver; all six deep at SF	6-SKU KIT Transfer as a set — still $0 capital
BOM100	COM001, COM002, COM003, COM004	89 — always	Mostly yes at SF; COM002 (96) & COM004 (85) thin	Anchor purchase + top up two companions
CTM200	OPR001, OPR002	62	Zero at Queens (SF 157, Boston 40)	3-SKU move if pursued
SBP100S	ASY107S (35×); COM002S/003S (25×)	35	COM-S items at SF; ASY107S zero avail at SF	Include ASY107S in the Boston→SF transfer audit
PR100-A	I360-Assembly, I360-Impeller, NS-SCCT, NS-SA-EIA-Inv	21	Impeller (42), NS-SCCT (20), EIA (15) already at Denver	Best-prepared target. FLAG I360-Assembly: zero at every warehouse
SSP100CTP/CTSM	SSP100P/SM assemblies + cartons/labels	10–13	Packaging already at NJ (49,920 labels, 7,480 cartons); assemblies not	Move assemblies together; packaging solved

Why this matters: the naïve version of this project ("move the top SKU by unit-miles") would have made service worse for PCB100L and SAF200 customers — every order would arrive in two boxes from two coasts. Basket-adjusted, the moves are bigger but honest. Bonus finding: I360-Assembly (item 2234) has zero available at all seven warehouses despite appearing in 21 PR100-A co-orders — an active backorder generator hiding in plain sight, worth a purchasing review regardless of any repositioning.

5 · Freight Economics

What the billing data shows

NetSuite holds 2,408 shipping-charge (ShipItem) lines on the 12 months of fulfillments, totaling $43,971. Regressing per-shipment charge against shipping distance across the flagged routes yields a slope of $0.0044 per mile — statistically flat. Peabody MA, Tempe AZ, Wilmette IL, Buffalo WY — radically different distances, identical $31.00 charge. A handful of outliers exist (SF-metro-from-Boston averages $108.06; San Mateo one-offs at $1,304), but the dominant pattern is a flat-rate table.

The decisive fact

ShipItem charges are what customers are billed, not what carriers charge the business. True carrier cost is not recorded in NetSuite. This has two consequences:

1. A precise "freight savings" number is unknowable from NetSuite data alone. Any modeled figure — whether from public carrier-rate heuristics or from the billing data — measures the wrong thing. Public-rate modeling suggests a magnitude in the $8K–$31K/yr range, but that is an estimate, not a measurement.

2. The margin argument is stronger than the savings argument. Shipping revenue is flat (~$31) regardless of distance, while carrier cost certainly is not. A $31-billed parcel trucked 2,555 miles from SF to Newark almost certainly costs more than $31 to ship; the same parcel from Queens (15 mi) almost certainly costs less. Every wasteful mile is a silent margin transfer to the carrier — invisible in NetSuite precisely because the cost side isn't captured.

What can be said with confidence

~28M excess unit-miles/yr exist; carrier pricing is distance- and zone-sensitive; therefore repositioning reduces real cost. Direction certain, magnitude unmeasured.
The flat-rate billing table itself may deserve review: if customers at 2,500 miles pay the same $31 as customers at 15 miles, the pricing model subsidizes exactly the customers who are most expensive to serve.

Recommended instrumentation (closes the gap permanently)

Start capturing actual carrier cost per fulfillment — a custom transaction body field (e.g. custbody_actual_freight_cost) populated from carrier invoices or a shipping-platform API (ShipStation/EasyPost/carrier webhooks). Even 60 days of data would calibrate this entire analysis.
Alternatively, pull one quarter of carrier invoices (CSV) and match to fulfillments by tracking number/date — a one-time exercise whose matching can be automated.
Then re-run this model with real costs: the ranking of moves will hold, and the dollars will be defensible for budgeting.

6 · Delivery-Time Impact

Ground transit approximated at ~500 mi/day. Improvements by cohort:

Demand cohort	Shipments/yr	Current route	Est. days	Proposed route	Est. days	Δ
Newark, NJ (PCB100L basket)	14	SF → NJ (2,555 mi)	5	Queens → NJ (15 mi)	1	−4
SF metro (BOM100, SBP100S)	~42	Boston → SF (2,692 mi)	5	SF local (<10 mi)	1	−4
NYC metro (CTM200, SPL001, gauges)	~40	Boston/SF → NYC	2–5	Queens local	1	−1 to −4
Winchester / Nicholasville, KY	~25	SF → KY (2,067 mi)	4–5	NJ → KY (567 mi)	2	−2.5
Park City, UT (PR100-A)	355	SF → UT (615 mi)	2–3	Denver → UT (355 mi)	1–2	−1
St Louis, MO (SAF200 kit + multi)	~95	SF → MO (1,735 mi)	4	Denver → MO (785 mi)	2	−2
FL Gulf Coast (SSP100, BOM100)	~36	SF/Boston → FL	5	NJ → FL (966 mi)	2–3	−2

Roughly 600 shipments/yr improve by 1–4 business days. Split-order reduction (§4) compounds this: fewer, fuller boxes arriving sooner.

7 · Investment & Payback

Phase	Scope (basket-complete)	Capital	Benefit case	Verdict
PHASE 1	Transfers only: PR100-A (~800 u) + SAF200 6-SKU kit → Denver; Boston excess audit for SBP100S/ASY107S/BOM100 → SF	~$0 + one-time transfer freight	−1 to −4 days on ~490 shipments/yr; split reduction for St Louis; margin leak reduction (unquantified but real)	Do now. No inventory risk — stock is already owned, just mispositioned.
PHASE 2	PCB100L 4-SKU basket → Queens (gated on §8 lifecycle check); BOM100 + companion top-up → SF; SSP100 assemblies → NJ (or transfer-on-demand)	~$60–150K depending on how much is transferable vs. purchased	−4 days on the two biggest Eastern flows; removes the largest unit-mile block (~20M)	Pilot after Phase 1 + carrier-cost instrumentation (§5) proves the margin case with real numbers.
PHASE 3	CTM200 3-SKU basket → Queens; gauge family → NJ/Queens	~$120K+	Service-level only; thin volume	Defer. Revisit only with carrier-cost data and/or customer SLA pressure.

Carrying-cost framing (the key nuance): capital figures are working capital, not expense; the real cost is carrying (~15–25%/yr). Because total network demand is unchanged, the goal is repositioning — the network should hold roughly the same total units, placed differently. Phase 1 embodies this perfectly: it moves existing overstock (SF holds 7,665 PR100-A against ~3,200/yr total demand) to where the demand actually is, simultaneously fixing overstock and mispositioning.

8 · Seasonality & Demand-Pattern Check

Monthly shipment units for key items, Jul 2025 – Jun 2026. Pattern matters: flat demand justifies permanent stocking; spiky demand suggests project-driven buying that shouldn't drive stocking decisions.

PR100-A FLAT — IDEAL

Park City: 23–32 shipments/mo, all 12 months. Textbook steady demand.

BOM100 GROWING, WINTER-HEAVY

Oct–Dec ~1,000/mo → Jan–Mar ~2,200/mo. Trend up; size the SF position on recent months, not the annual average.

PCB100L GATE — APRIL STOP

1,200–2,100 u/mo Oct–Mar, zero after March 2026. Resolve before the $110K basket move: seasonality, lost customer, or product transition?

SSP100CTP / CTSM SPIKY — PROJECT-DRIVEN

Feb–Apr burst (940 u in March), then quiet. Prefer transfer-on-demand over stocking.

9 · Address-Book Audit — All 400 US Addresses

Every active US customer address (400 total) was validated: ZIP/state consistency against USPS prefix ranges, format checks, and placeholder detection, supplemented by US Census geocoder spot checks. 93 addresses flagged — a 23% defect rate. No records were modified; this is a findings inventory for review.

Severity	Count	Issue	Examples	Impact / suggested handling
HIGH	15	ZIP leading zero lost ("1824", "2816", "8330")	Cust 229 Chelmsford MA "1824"→01824 · Cust 391 Tewksbury MA "1876"→01876 · Cust 528 Mays Landing NJ "8330"→08330	Breaks carrier label validation & rate lookup. Classic spreadsheet-import artifact concentrated in MA/RI/NJ/CT. Mechanical to fix once approved.
HIGH	19	ZIP/state contradiction	"Seattle, WA 40010" (KY ZIP) · "Dublin, OH 90010" (LA ZIP) · "Pendleton, AK 97801" (that's Pendleton Oregon) · "Las Vegas, NE 45644" (NV intended? OH ZIP)	Cannot tell which field is wrong — needs human review per record, ideally against the customer's real location.
MED	23	State spelled out ("California", "Wisconsin", "Washington DC")	Customers 3476–3500 — one contiguous import batch	Breaks state-filtered saved searches, reports, and tax logic that expect 2-letter codes. One bad import; batch-normalizable.
MED	2	Missing ZIP entirely	Cust 516, 518 (San Francisco)	Undeliverable as-is.
LOW	3	Placeholder / test data · PO Box	"123 Main St, Anywhere CA" · "123 Germany, Austria CA" · PO Box 351321	Inactivate test records; confirm PO Box is billing-only.
INFO	29	Intersection-style addresses ("West 96th and Central Park West")	All NYC-metro	Some carriers accept these; USPS standardization does not. Fine for now; blocks any future address-automation project.
INFO	2	Cosmetic (lowercase city; addr1 contains full city/state/zip)	"san mateo" · Cust 510 Lakewood NJ	Normalize whenever touched.

Why it belongs in this report: positioning analysis is only as good as the coordinates, and 23% of the address book can't be reliably geocoded or rate-shopped. The 15 leading-zero ZIPs and the 23 spelled-out states are mechanical, low-risk fixes when the business is ready to approve them; the 19 ZIP/state contradictions need a human eye. Until then, treat any per-customer distance/zone calculation involving flagged records as suspect.

10 · Methodology & Limitations

Data pipeline

NetSuite (SuiteQL): 1,461 US item fulfillments (type ItemShip, Jul 2025–Jun 2026) joined to transactionshippingaddress + transactionline for item, quantity, and actual ship-from location; inventoryitemlocations for availability; item records for cost/weight; sales-order line co-occurrence for basket affinity; ShipItem lines for billed freight; full customeraddressbook for the audit.
External (US Census Bureau Geocoding API): warehouse and top ship-to addresses geocoded to lat/long; city-centroid coordinates for smaller cohorts.
Computation: haversine distance from each ship-to to every warehouse; excess miles = (distance from actual ship-from) − (distance from nearest warehouse); flows >100 excess miles flagged and ranked by excess unit-miles; linear regression of billed freight vs. distance; USPS ZIP-prefix rule validation.

Limitations — read before acting

True carrier cost is absent from NetSuite (ShipItem = customer billing). All dollar-savings figures are estimates from public rate heuristics; direction is certain, magnitude is not. See §5 instrumentation plan.
Straight-line miles, not road miles or freight cost; centroid noise ±10–30 mi on small cohorts.
"Nearest" assumes every warehouse could stock every item. Subsidiary restrictions, hazmat rules, or licensing may forbid some placements — not verified per item/location.
Single-shipment anomalies excluded from recommendations (e.g., NS-PTD100 ×1,220 to St Louis — a one-off project) but present in totals.
Basket affinity uses ≥3 co-orders as the threshold; rarer pairings not captured.
Warehouse capacity/labor assumed adequate at Queens/NJ/Denver — pick/pack absorption not assessed.
Address audit is rule-based (ZIP-prefix ranges + format); a full geocoder pass on all 400 would additionally catch street-level errors (a 10-address geocoder sample found ~40% street-level issues).
Sample-account artifacts exist in this dataset (NS-* test items, placeholder addresses); recommendations were built from flows that look like genuine commerce, but a human pass over the item list is advised.

11 · Other Factors & Suggested Next Steps

Root cause worth fixing first: Auto-Location Assignment is OFF in this account. Without automated or region-defaulted location assignment at order entry, repositioned stock will still get bypassed. Fixing this multiplies the value of every stocking change.
Carrier-cost instrumentation (§5): the single highest-leverage data improvement available. Also enables reviewing whether the $31 flat billing rate under-recovers on long-haul customers.
Purchasing review for I360-Assembly (2234): zero units network-wide against recurring co-order demand.
PCB100L lifecycle question: the April demand stop gates ~$110K of Phase-2 spend. One conversation with sales likely resolves it.
Supplier inbound geography: if PCB100L's vendor is West-coast, stocking Queens means paying the cross-country leg anyway — but on cheaper, consolidated inbound freight instead of parcel outbound. Usually still a win; worth confirming vendor origin.
Replenishment settings: stocking a new location requires reorder points / preferred stock levels per location in NetSuite, or the items go stale.
Contractual/SLA drivers: if any flagged customers have delivery-time commitments, service-level gains may dominate the financial ones.
Returns flow: closer warehouses also shorten return legs — small but real.

Recommended sequence: (1) Phase-1 transfers — zero capital, immediate service gains; (2) start capturing carrier cost; (3) resolve the PCB100L lifecycle question; (4) revisit Phase 2 in a quarter with real cost data and a proven Phase-1 result.