How to spot when two vendor names share one address

Procurement runs into this every time a vendor master file gets audited: two supplier records, two different names, two different vendor numbers in the ERP, and then someone finally pulls the W-9 or the certificate of insurance and notices the street address is identical. Sometimes it's a unit number apart. Sometimes it's the exact same loading dock.
This is common. It happens whenever a category gets built up over five or ten years of onboarding one PO at a time, with nobody stepping back to check whether "Vendor A" and "Vendor B" are the same building wearing two different letterheads.
Why two vendor records end up at one address
A few patterns cover most cases:
A contract manufacturer runs white-label production for several brands out of one plant, and each brand's trading entity gets set up as its own vendor record. A regional distributor gets bought by a larger group, keeps its old name on invoices for a year or two, but ships from the parent's consolidated warehouse. A sales office and the actual production site share a mailing address because the sales office is the production site, just with a different legal name on the paperwork than the one your buyer used when they sourced it. And sometimes it's simpler than any of that: a supplier rebrands, nobody updates the old record, and you're carrying a ghost.
None of these are fraud. They're just normal churn in how companies register, rebrand, and get acquired. The problem is that your vendor master doesn't know any of this happened, and your risk team's "we have three suppliers in this category" count is wrong because of it.
A manual check you can run this afternoon
If you want to find these before an audit forces the issue, here's the check most category managers do by hand:
Pull the full address field (not just city and country) for every active vendor in the category, and normalize it: strip suite numbers, standardize abbreviations, geocode to lat/long if you can. Sort or cluster by that normalized address. Anything that lands on the same coordinates with a different vendor name is a candidate.
From there, cross-check the things that tend to leak the connection even when the legal names don't match: the same contact phone number on two "different" suppliers, the same bank routing details on remittance forms, overlapping tax ID prefixes for related entities, or a shared customs broker on import documentation. Company registry lookups in the country of incorporation will often show a parent-subsidiary relationship directly, if you're willing to pay for each lookup one record at a time.
This works. It's also the kind of task that gets done once during a vendor master cleanup project and then quietly stops getting done, because checking a hundred-line vendor list against a map by hand, one address at a time, isn't a job anyone wants to repeat every quarter.
What the address check doesn't catch
Even a clean version of this exercise only answers "are these two vendor records the same address." It doesn't answer the broader version of the question a risk review needs: how much of the whole category, across every supplier that's genuinely distinct, is still sitting in one industrial park, one port catchment, or one flood-prone stretch of river. Two vendors can have entirely different, correctly separate addresses and still both be two buildings down from each other on the same industrial estate, which is a concentration risk no vendor master field will ever flag.
That's a different exercise from matching duplicate names, and it's the one most category reviews skip because it genuinely requires looking at the ground, not the spreadsheet. Sourcing Concentration takes the supplier list you already have and plots it against satellite imagery once a year, so the cluster that never shows up in a vendor dedup exercise, because every supplier on it has a technically unique address, still shows up on the map.
If your last vendor master cleanup found the duplicate names, it's worth finding out what's sitting next to them.