Supplier Replacement Time Is a Spread

Most resilience plans have a number in them. Three approved alternates. Two backup carriers. Dual-source on the critical parts.

The number's the least useful bit.

Ask how long it takes to actually buy from each one and you don't get a number back. You get a spread, and it's wider than anyone expects.

The spread

A partner already on EDI whose spec you've worked before is a couple of weeks from decision to first live transaction. One with custom segments and odd validation rules is three or four. Run the whole traditional cycle with certification testing and you're at six to ten weeks.

And a partner who isn't on EDI is a different kind of project rather than a dead end. The default is somebody retyping their documents into your system forever. The alternative is capturing what they do send, PDFs or email or a portal export, and generating the EDI from that. Which is often the faster of the two, because the clock stops depending on their certification queue and starts depending only on you.

Same word on the slide. Two weeks at one end, an entirely different project at the other.

Same standard, different mapping

The bit nobody asks about is that two partners can sit on the same X12 version and still need completely separate mapping, because each publishes its own implementation guide on top of the standard. Segments the standard calls optional become mandatory. Code values get restricted to a subset. Conditional rules appear, so one segment being present makes another one required. And they disagree about how to identify your item, whether that's UPC, your vendor number, a GTIN, or two of those in a particular order.

Write down your worst case

So the honest figure in a resilience plan is your worst alternate rather than your count of them. That's the one you'll be reaching for on the day you have no choice.

Takes an hour to find out. Write the alternates down the page, put a real number next to each name. Not a target. What it actually took last time, or whatever your integration lead says when you ask them cold.

The list usually gets shorter.

Making the long end of that spread predictable is roughly what we do, but the spreadsheet hour costs nothing and you should do it either way.