Went through 146 recorded calls with ecommerce operators. How the inventory spreadsheet actually fails
Disclosure up front: I'm the founder of an inventory software company (SKU.io), so read this with that bias. No links in this post and nothing in it needs my product. I had the data sitting there and figured this sub might find it interesting.
The data: we record our sales and onboarding calls, and 146 of them are tagged for internal research. Mostly operators doing $1M-$50M across Shopify + Amazon plus some wholesale. 33 of the 146 got into the spreadsheet or Excel layer that actually runs their purchasing and inventory. I went back through those and sorted what people said into buckets.
The thing that surprised me: almost nobody says the math is wrong. The formulas mostly work. The sheet dies operationally. Ranked by how often it came up - the top four each showed up in at least five separate calls:
It depends on one person. "We have someone three days a week who just produces manual reports." A guy quits and leaves behind five spreadsheets of tabs nobody else can read. One founder at around $25M told me his fear isn't the formula, it's that somebody forgets to update a macro and everything downstream is quietly wrong for days.
It drifts from physical reality. One buyer did a physical count during due diligence on a business and it came back at about two and a half times what the sheet said. Damaged and defective stock is the usual leak - the warehouse floor knows, the sheet doesn't. The year-end count vs the sheet is where "we thought we were profitable" goes to die.
Costs move and the sheet doesn't. This one has gotten noticeably worse since tariffs. Same product, one container lands at $5/unit freight, the next one at $3. Cost layers maintained by hand. Change one FOB price and there are four other tabs that were supposed to inherit it.
The math outgrows the sheet. A reorder pass that takes one full day, once a month. 2,400 SKUs where changing a growth assumption means editing 2,400 rows. One ops person described allocating a single PO across warehouses as "apparently asking for a Nobel prize in math."
Concurrency. Two wholesale reps entering orders at 11pm off the same sheet, both seeing the same available stock. That oversell happened without anyone typing a wrong number.
And the bill for all of the above shows up as inventory, never as "a spreadsheet problem": eight months of cover sitting on one product. Close to $100K in overstock. One team tracked weeks-of-cover by hand, weekly, and still ended up with 8,298 units past 52 weeks of cover. Stockouts on the bestseller while cash sits in the slow movers. One seller put it flatly: "I think most people don't truly calculate the cost of running out of stock."
If I had to compress the whole pile: a spreadsheet is a snapshot pretending to be a system. It doesn't know anything happened until a human tells it, and the humans are busy.
Curious whether the folks here on the warehouse and manufacturing side see the same ranking. I'd guess the failure modes are different when you control the whole building.