Mesa Ridge Supply Co., a $23M industrial distributor in Texas: fasteners, hydraulic fittings, hose and MRO from three warehouses, run on a cloud inventory system.
Inventory climbed while sales fell, and the bank asked which decisions put the cash on the shelf. Every report could name the stuck items. None could say why each one was bought.
The question it answers
Messy exports in. Every fix logged, every note read, every unit traced to the order that bought it.
Purchases arrive through four decisions, and no report says which one. Vendor programs and the buyer's own bets were a fifth of the buying and half of the trapped cash.
The four exports were cleaned, Claude read every note the buyers typed, and code traced each unit on the shelf back to the order that bought it.
A decision list with a dollar value on every line, and one field on the purchase order that keeps it visible every month.
No system has a field for why an order was placed. Claude reads the notes the buyers typed, and the order's own facts carry the rest.
The cleaning, the trace and the totals run in code, so the same exports give the same answer every time.
Each line is a decision with a dollar value: what to return, what to stop buying, what to run down.
Find the cash · Mesa Ridge Supply Co.
A $23M industrial distributor in Texas: fasteners, hydraulic fittings, hose and MRO from three warehouses. Four standard exports, as they came out of the inventory system. US dollars.
Inventory grew 48.9% while sales fell 9.1%. The bank's question was not which items are stuck. It was which decisions put $1,897,269 on the shelf, and will they do it again next year.
Act 1 · What the reports already said
Aging, ABC and stock on hand all group by a field that exists: supplier, category, location, date. No system has a field for the reason an order was placed. That is the column this dashboard fills in.
Act 2 · Where it broke
| Why it was bought | Purchases, 24 months | Still on the shelf | Trapped | Share of shelf trapped |
|---|---|---|---|---|
| Vendor program | $6,326,939 | $1,336,648 | $731,801 | 54.7% |
| Buyer's call | $1,334,675 | $462,878 | $248,143 | 53.6% |
| Reorder point | $27,990,772 | $3,787,870 | $860,640 | 22.7% |
| Customer order | $1,059,703 | $189,319 | $33,815 | 17.9% |
| All buying, incl. stock bought before the window | $36,712,089 | $5,777,544 | $1,875,052 | 32.5% |
A dollar bought on a vendor program is 2.4 times as likely to be stuck as a dollar bought at the reorder point. 43% of program dollars were committed in the last 21 days of a quarter, against 26% of all buying: the supplier's calendar, not Mesa Ridge's demand.
Trapped by supplier
| Supplier | Trapped |
|---|---|
| Red River Hose | $418,482 |
| Mesa Ridge Pro (import program) | $255,290 |
| Trinity Components | $238,750 |
| Caprock Lubricants | $157,261 |
| Lonestar Industrial | $140,003 |
| Pecos Seal & Gasket | $117,394 |
How each order got its reason
| Source | Orders | Purchases |
|---|---|---|
| Note read by Claude | 1,776 | $14,715,557 |
| Order facts only | 1,135 | $20,083,389 |
| Reference field | 111 | $1,913,143 |
Claude read every distinct note the buyers typed. The order's own facts (line count, timing, cover, a customer order first) carried the rest. Checked against the known reason for every order: 79% right on the order's facts alone, 90% with the notes read.
Act 3 · What to do
| Decision | Cash released |
|---|---|
| Return program stock still inside supplier terms | $185,076 |
| Stop buying the items with no sale in 180 days | $183,795 |
| Run down the buyer's bets that never found demand | $52,714 |
| Clear what is left over from customer orders | $6,108 |
| Recoverable within two quarters, incl. $164 of stock bought before the window | $427,857 |
Every quarter the buying runs unchanged adds about $52,000 more to the shelf. The fix that keeps it visible is one dropdown on the purchase order, four choices, filled in when the order is raised. From then on this is a monthly report, not a project.
What the exports looked like, and what was fixed
| Fix, logged line by line | Count |
|---|---|
| Purchase order lines exported twice, removed | 473 |
| Supplier names unified | 4,872 |
| Suppliers after unifying | 14 |
| Location names unified | 2,479 |
| SKU spelling fixed (purchase order lines) | 1,650 |
| SKU spelling fixed (sale lines) | 3,006 |
| Purchase order dates in a second format read correctly (OrderDate) | 5,779 |
| Sale dates in a second format read correctly (SaleDate) | 10,668 |
| Purchase order amounts stored as text converted (LineTotal) | 7,994 |
| Sale amounts stored as text converted (LineTotal) | 14,954 |
| Purchase order lines where quantity x price does not equal the line total | 0 |
What it does not count
The rebate earned on each program is set against the cost of carrying the stock it left behind before a program goes on the stop buy list. The release is cash out of stock, not profit. Recovery figures assume returns and run-down within two quarters, and are replaced by your actual supplier terms.
Your turn
Run the same first step on your own four exports. Or have one decision tested before it is signed: the next quarter-end vendor program.
Test one decisionBack to Find the cash