SourcingConcentration

Blog

Country, province, or corridor: picking the right unit for a concentration map

Aerial view of a sprawling industrial park with warehouses, roads, and green space.

Every supplier concentration review starts with the same question nobody answers out loud: concentration of what, measured at what level? Spend by country tells procurement leadership something for the board deck. It tells the risk team almost nothing about whether a flood, a port strike, or a fire in one industrial park could take out half a category.

The unit you pick determines whether the map finds the risk or just repackages the ERP export.

Country rollups hide the cluster that matters

"60% of connector spend sources from China" is a true sentence and a useless one for planning. China is roughly the size of the US and has multiple industrial belts, port systems, and power grids that don't share a single point of failure. A country-level pie chart treats a supplier in Shenzhen and a supplier in Chengdu as the same exposure. They're 1,500 km apart, on different grids, reachable through different ports.

Country is the unit finance and the board want because it's fast to pull and easy to present. It's not the unit that tells a category manager whether a dual-sourcing plan is real diversification or two addresses on the same industrial estate.

Province or state is better, but it still misses the corridor

Moving down to province, state, or prefecture fixes some of this. Now "Guangdong" is more useful than "China," and "Ohio" is more useful than "United States." But administrative boundaries were drawn for tax and governance reasons, not for supply chains, and plenty of real clusters sit on or across a provincial line: a river port that serves two adjacent states, an industrial corridor that runs along a single highway and happens to cross a border partway through.

A supplier list tagged by province will split a corridor into two rows and understate it. It will also lump together two suppliers that share a province but sit on opposite sides of a mountain range, two different watersheds, two different grid operators, and call that "regional diversification."

Province is a reasonable filter for a first pass. It's the wrong unit to make a sole-source-risk decision on.

What finds the cluster

The unit that catches both problems is physical: where are the sites, not which jurisdiction do they file taxes in. A cluster worth flagging is a shared industrial park, a shared flood plain, a shared port catchment, or a shared stretch of corridor, regardless of what the address field says. Two suppliers in the same industrial park are one point of failure even if a clerical quirk puts them in different postal districts. Two suppliers 400 km apart in the "same province" are not meaningfully correlated risk at all.

This is the gap between a vendor list and a map that shows where things sit: a spreadsheet sorts by the text fields it has, and a flood plain or a port catchment isn't a text field anyone filled in when the supplier was onboarded. Someone has to lay the addresses over physical geography and look for where they bunch up, which is slow to do by hand across a category of any real size. A concentration map that plots category spend directly against supplier geography does that layer-over step once a year and hands back the clusters already found, ranked by how much spend sits in each one.

If your last supplier risk review reported diversification at the country or province level and you want to know whether that holds up once you look at where the sites actually sit, that's the gap worth checking before the next sourcing decision locks in.

← Back to the blog