The January GRI announcement is the worst possible document to organize your peak post-mortem around, and it's the one most logistics teams wait for. By the time the carrier tells you what next year costs, your memory of what this year actually broke has gone soft, your best evidence has been overwritten by holiday averages, and the negotiation frame belongs to the other side of the table. The post-mortem window is now — while the exception queue is still warm and the people who worked the waves still remember which night the wheels wobbled.
At the terminal we wrote ours on the dock whiteboard before the decorations came down. Not because we were disciplined. Because we'd learned that by February, everyone's recollection of December improves remarkably, and improved recollections negotiate poorly.
Here's the method, five chapters, in order.
Chapter one: exceptions by cause, not by carrier. The lazy cut is a league table of carriers ranked by failure count. It feels rigorous and tells you almost nothing, because it buries the distinction that matters: which failures were the carrier's capacity and which were handoffs — your dock to their linehaul, their hub to a partner's final mile, your data to their system. Most peak failures are handoff failures wearing a carrier's name. Recut your exception file by failure point. If a carrier missed pickups, that's their chapter. If your wave finished ninety minutes after their trailer left, that chapter is yours, and no rate negotiation fixes it.
Chapter two: surcharge realized versus surcharge modeled. Pull what you projected in September against what the invoices say now. The gap is your modeling error, and it's also your preview of the GRI game: published headline rates understate realized increases by 100 to 200 basis points once the surcharge mix shifts beneath them (S2, CNBC on why surcharges persist). Knowing your own model's miss, by carrier and surcharge type, is what lets you read January's announcement as a decomposition exercise instead of a press release. Write the miss down while the invoices are still arriving.
Chapter three: cross-border lanes get their own chapter, never a footnote. If your network takes Mexico-origin freight through Laredo and up I-35, blending those lanes into network averages hides exactly what you need to see. Border dwell by week. Broker release performance against whatever was promised. Which customs exceptions repeated, and which document field caused them. The northbound flow through this corridor doesn't take Decembers off, and the buildings around Alliance that took first touch on that freight have their own story — told in yard moves and door turns, not in the network summary. Interview the dock leads before the holidays scatter them. Fifteen minutes each. The whiteboard version is fine.
Chapter four: the decision log audit. Somewhere in October or November, somebody overrode routing rules by hand — diverted volume, forced a carrier, held a wave. Every override is one of two things: a rule that should exist and doesn't, or a panic that shouldn't repeat. List the overrides, sort them into those two buckets, and you have next year's rules backlog written by your own peak, for free. This chapter takes an afternoon and routinely produces more value than the other four. It is also the one nobody writes, because overrides embarrass the people who made them. Make it blameless and make it mandatory.
Chapter five: convert findings to January positions. Each chapter ends in a sentence you can say across a negotiating table. "Your hub missed sort on these eight nights, here are the timestamps" is a position. "Our surcharge model missed DAS drift by this much, so this year's cap proposal looks like this" is a position. "We're moving these two lanes to a regional unless the pickup commitment changes" is a position. Adjectives are not positions. The carriers arrive in January with data; the only question is whether you do.
The tradeoff, honestly stated: doing this in the last week of December costs you the one quiet week your team gets, and I won't pretend otherwise. So scope it tight — the five chapters above fit in three working days with the data you already have. What I'd push back on is deferring it. A post-mortem written in February is a different document: same template, worse evidence, softer numbers, written after the GRI has already framed what everyone thinks happened. The cheap week is now. The expensive version of this work is called "renegotiating in March."
One more thing the post-mortem buys you, beyond January: an honest read on time-to-confidence in your own systems. If chapter two shows your surcharge model missing wide, or chapter four shows your team routing around your own rules engine at midnight, those are implementation findings, not peak findings. They belong on the same one-page summary you give your leadership, because they're the difference between buying software and trusting it.
Concrete next step: we keep the post-mortem template as a one-page-per-chapter worksheet — the five chapters above, with the data pulls specified so your analyst isn't guessing. Request it, run it before the year turns, and if you want a second set of eyes, send us chapters two and three and we'll pressure-test your numbers before you sit down across from anyone in January.