Regulatory filings, government statistics, licensing registries, job postings, page source. Everything here is worked out from materials anyone can open — registries pulled in full, shelves tracked week by week, financials checked against the original filings. The ingredients are the same for everyone. The cooking is what differs. Handle the same material differently, count it differently, and each piece ends up with a ruler its industry didn't have. We also write down where public data stops — beyond that line the numbers exist only in your own data, and that is the work we do.
A new piece weekly
Six of eight brand companies had the same condition. The press named a surge in marketing spend as the cause. Nowhere was there evidence that anyone had measured whether the money produced incremental revenue.
Read →Traditional liquor is the rare Korean drinks category permitted to sell online. Yet only one of six producers runs a storefront of its own. For the other five, every order record accumulates on someone else's platform.
Read →Filings break revenue out by sales channel — but every company that discloses it bundles its own storefront with third-party marketplaces into one "online" line. The own-channel share appears nowhere, and that boundary decides whose database the customer ends up in.
Read →Korea's full manufacturing registry shows about 380 new food products born every day — and no way to count the ones that die. So I measured the age of the survivors: shelf clocks differ almost ninefold by product type.
Read →Seven weekly snapshots of Olive Young's category rankings. The skincare cluster trades brands at 1.2–1.4× chance while every path into makeup sits below half of chance — getting on a shelf and widening it are different events.
Read →Korea publishes prescription data by ingredient, dose, district and month — 2,340 API applications in ten years, and not one public analysis I could find. Using it directly showed: combinations crossed half the market, and the best-selling drug's dose demand and product lineup diverge by 18 points.
Read →All fifteen fashion brands had their auditor name inventory valuation a key audit matter. That inventory is 30% of revenue sitting for 299 days — yet days-of-inventory, the only gauge available from outside, correlates with the write-down rate companies actually booked at +0.002.
Read →Seoul's coffee sales fell 11.9% in two years — transactions fell 18.6%, and an 8.2% price rise papered over half of it. The median neighbourhood's coffee shop rings up thirty-one payments a day. Only 23 of 384 neighbourhoods grew both customers and per-shop earnings — and neither Gangnam-daero, Hongdae nor Seongsu made the list.
Read →Chase the neighbourhoods where sales run high, and the rent is higher by exactly that much — the burden ratio stays flat (+0.01). Low-closure neighbourhoods aren't safe; they're places nobody opens in — the only thing closure tracks is opening (+0.45). And industry explains 37% of closures; neighbourhood, just 4%.
Read →We opened the public records of 85 consumer, F&B and hospitality companies and counted the paths along which one customer splits into separate ledgers. 84 of 85 (99%) had at least one. Fragmentation is layered — median two layers, two in five carry three or more — which is why one system never fixes it.
Read →Seven weeks of Olive Young shelf prices. The usual suspicion — that list prices are padded to manufacture a discount — did not survive: they held steady for 96.2% of products and matched the undiscounted week's real selling price to the won 97.2% of the time. What that makes visible: 1,833 of 6,630 products (27.6%) have zero observations of themselves at that price. Discount depth does not separate them either — both groups sit at a median of 20.0%.
Read →Only figures verified against the original source. Secondary aggregates and press summaries are never used as headline numbers, and no comparison is drawn without stating standalone or consolidated basis.
The limits come first. Public sources cannot establish incrementality, contribution profit by channel, or repeat purchase. What follows is a hypothesis, not a measurement.
No sector is being written off. These industries were broadly having a good year. What diverged was execution, company by company.
Lambency uses public data to diagnose where to look, and internal data to test whether it holds.
If you want to know what the same method turns up on your own numbers.
caffrey.w.lee@gmail.com