The Manufacturing System That Wholesale Distributors Were Never Meant to Use
Picture this: It's 1964, and Joseph Orlicky is about to revolutionize manufacturing at Black & Decker with something called Material Requirements Planning (MRP). The concept is elegant—forecast what you need, calculate your production runs, and manage inventory accordingly. Throughout the 1970s, manufacturers embraced this methodology with fervor:
SAP was founded in 1972, adding supply chain planning to MRP.
Oracle’s ERP systems debuted in 1977.
These manufacturing-focused enterprise systems became the craze, adopted by most of the Global 1000 companies.
The Great Migration—And Why It Was Doomed
Like most successful innovations, forecasting-based inventory planning didn't stay confined to manufacturing. Through the 1970s and 1980s, wholesale distributors looked at their manufacturing partners' shiny new systems and thought, "We need something like that too."
The software industry responded. GAINS was founded in 1971 and Smart Software in 1981. In the late 1970s, what would eventually become Infor SX.e debuted as GAP, was rewritten into a product called Trend, and became SX Enterprise in 1989. Eclipse began development in 1988 by Clark Yennie and incorporated as Eclipse, Inc. in 1990, specifically targeting wholesale distribution. These weren't manufacturing systems retrofitted for distribution—they were built from the ground up for the distribution business model.
But here's where the story gets interesting: even distribution-native ERP systems adopted the forecasting paradigm that had proven so successful in manufacturing. Why wouldn't they? It was the state-of-the-art approach to inventory management.
By 1980, entrepreneurs like Anders Herlitz saw an opportunity. He founded E3 Corporation in Marietta, Georgia, as one of the first third-party forecasting and replenishment tools designed to bolt onto ERP systems. E3 would eventually grow to employ more than 250 people with offices in twelve countries before being acquired by JDA (now Blue Yonder) in September 2001, specifically to accelerate JDA's move into collaborative planning, forecasting, and replenishment between suppliers and retailers.
Slimstock released Slim4 in 1993. Thrive Technologies released its first forecasting and replenishment system in 2001 focused on distribution companies. Blue Ridge Global was founded in 2007. Netstock entered the market in 2009.
The forecasting-industrial complex was in full swing. There was just one problem for wholesale distributors: every software vendor assumed the forecast-based approach would work for distribution like it did for manufacturing.
The Tale of Two Inventory Worlds
The uncomfortable truth that's been hiding in plain sight for three decades is this: wholesale distribution and manufacturing are fundamentally different animals, and what works brilliantly for one can fail spectacularly for the other.
The difference comes down to a simple reality that has profound implications:
Manufacturers: Small SKU counts, high volume per SKU
Distributors: Large SKU counts, low volume per SKU
When Black & Decker forecasts demand for a particular drill model, they're dealing with potentially thousands or tens of thousands of units of sale volume for that one SKU. Forecasts work quite well for high-volume smooth demand history.
Wholesale distributors, on the other hand, live in a completely different universe. An automotive parts distributor might carry thousands of SKUs spanning everything from door handles to exhaust pipes for multiple makes and models. When 80% of your inventory consists of slow-moving items that sell 10 or less times in that branch—as research from Thrive (2022) confirms—you're playing a completely different game.
Eclipse, SX Enterprise, and their competitors brought industrial-strength ERP systems to distribution. E3 and others layered sophisticated replenishment buying logic on top of simple forecasts. Forecast error was used as an input into safety stock. But the math (and resulting inventory levels) doesn’t work well with a SKU that sells 4 times a year.
The Killer Statistic Nobody Talks About
As mentioned above, Thrive Technologies has analyzed the data of hundreds of distributors across North America over the last 25 years and discovered that 80% of wholesale distributors' SKUs sell fewer than 10 times per year.
Let’s read that again.
You're trying to forecast demand for items that sell less than once a month. Traditional forecasting methods need volume to work—they need patterns, trends, seasonality signals. They need the statistical significance that only comes from regular, repeated transactions. What they don't handle well is radio silence punctuated by occasional bursts of demand.
When Forecasting Becomes Fortune-Telling
Research published in the journal Operational Research (September 2020) confirms what distributors have suspected all along: daily SKU demand forecasting is "particularly challenging when dealing with irregular series characterized by intermittency and erraticness, especially for slow-moving items." That's academic-speak for "traditional forecasting doesn't work here."
The data gets even more damning. In one study examining demand patterns, researchers found that only 23% of SKUs exhibited "smooth" demand—the kind of predictable patterns that traditional forecasting loves. A staggering 78% fell into intermittent, erratic, or lumpy demand categories. And here's the kicker: forecast accuracy is only a meaningful metric for that smooth category.
In other words, for the vast majority of a distributor's inventory, traditional forecasting accuracy measurements are essentially meaningless. Current supply chain planning vendors who claim that their forecast accuracy is over 90% for distribution companies are not telling you the truth, or they don’t know how to measure forecast accuracy correctly.