Vendor websites are written in dry-erase marker. The Wayback Machine kept the originals, and if you administer a TMS that's been through a couple of corporate owners, the archive knows your platform's trajectory better than your account manager does. May is network-redesign and migration-planning season — the quiet stretch when you can actually schedule a dual-run before peak makes everything urgent — so let's take the walking tour of what the redesigns forgot.
Stop one, 2003: iShip's homepage. Go read it; it's a time capsule of the first multicarrier pitch — compare carriers, pick smart, ship from the browser. A quarter century old, and the promise on that page is recognizably the same promise on every shipping-platform homepage today, ours included. Worth a moment of humility from everyone in this industry: the pitch has been stable for twenty-plus years. What's differed is who survived delivering it, and under whose ownership.
Stop two, 2016: Kewill's homepage — confident, global, present tense — and the same year's Kewill blog, "The Seven Million Mile Man", a working content operation with personality. A real company, in other words. The Kewill name has since dissolved through merger and rebrand into the BluJay lineage and onward into E2open's portfolio. The blog post survives only in the archive. The customers' Clippership-descended installs survive in production. Both outlived the brand.
Stop three, 2018: ConnectShip's homepage — and its quiet present as a UPS subsidiary. No scandal there; UPS has owned it for years. But here's my favorite exhibit in the whole museum: a ConnectShip Progistics portal still answering today on a Pitney Bowes sendsuite subdomain. Production software from a previous era of the industry, still standing at its post like nobody told it the war ended. Somewhere, freight documentation depends on that page. If that sentence made you check something in your own stack, good.
Stop four, 2020: Logistyx's homepage — independent, expansive, future tense. Then the paper trail: E2open acquires Logistyx for $185 million (S4), and later, E2open's legacy Logistyx SLA page describing maintenance-grade support (S11). Homepage, press release, maintenance page — the full arc in three public documents, each one written for a different audience, only the last one written for you.
Add it up and the picture is plain. Six legacy TMS brands now sit inside one company. Their customers are running on borrowed time. Not because the software stops working — that's the trap, it keeps working — but because maintenance mode is a slow leak, not a bang. Carrier compliance drifts a cert at a time. Surcharge tables update a release late. The engineers who knew the old engine's folklore retire or get reassigned, and the talent pool that can even read the configs shrinks every year. By the time the pain is undeniable, you're migrating under duress, which is the most expensive way to migrate. And on the "modern" alternative some vendors offer: single-tenant hosted instances of the same old code aren't a migration at all. Single-tenant cloud is just outsourced legacy — your problem, moved to someone else's data center, at rent.
So here's the May agenda for the TMS admin who recognized their own stack on the tour. The realistic benchmarks: a cloud TMS deployment for a mid-market operation runs a couple of months; a full legacy replacement runs a year or more [S-cite: S26 — cloud TMS 8–12 weeks mid-market; legacy replacement 12–18 months]. Count backward from next January and the math is unforgiving — a migration that should finish before peak freeze needs its dual-run starting now-ish, and its decision made yesterday-ish.
And insist on the dual-run, because the method matters more than the vendor. Ninety days minimum, old and new systems rating and labeling in parallel, with daily reconciliation of rates, labels, and manifests — every discrepancy logged and dispositioned, not waved off. Rip-and-replace is a bet that nothing in fifteen years of accumulated rules mattered; around here I've watched that bet lose in buildings within sight of the I-35 corridor, where the freight kept arriving while the new system learned on the job. The network-redesign work most teams are doing this month — mode mix, injection points, the build-out math along the Alliance corridor — is the natural moment to run the software question in parallel, because redesigning lanes on top of an engine in maintenance mode is installing new plumbing in a condemned building.
The tradeoff, stated straight: a 90-day dual-run costs real money — duplicate licensing, integration hours, and an admin team running two systems through a Texas summer. It is still the cheapest insurance in enterprise software, because the alternative is discovering reconciliation gaps in production, in November, one at a time.
Concrete next step: the migration-window planner. One page — your contract dates, peak freeze, dual-run span, and cert-renewal calendar laid against a January-backward timeline, plus the honest TCO worksheet covering what vendors leave out: cert maintenance, surcharge-update cadence, integration drift, and talent risk. Request both, fill them in an afternoon, and you'll know whether your window is this year or already behind you. The archive is patient. Your rate tables aren't.