The setup
The plan was sound on paper: run the legacy CRM and the new platform side by side for six weeks, migrate department by department, then decommission the old system. Leadership approved budget and a project charter with a clear end date. What the plan didn't account for was that one large business unit — the firm's highest-revenue division — had built years of custom reporting and a commission-calculation process directly on top of the legacy system's data model, and migrating that logic to the new platform turned out to be a much bigger job than the discovery phase had estimated.
Rather than delay the whole rollout, leadership made the call to launch the new CRM everywhere else and let that one division keep using the legacy system "until the reporting gap is closed." That was the last clearly-scoped decision anyone made about the timeline.
The collapse
The reporting gap was never formally closed. It was deprioritized, then reprioritized, then deprioritized again as other initiatives took the division's attention. Two years in, the "temporary" exception had become permanent infrastructure: one division ran entirely on the legacy system, everyone else ran on the new platform, and a manual weekly export-and-reconcile process stood in for what should have been one shared system. The person who built that reconciliation process left the company. It now runs as a half-documented set of spreadsheet macros maintained by whoever inherited the responsibility, informally, alongside their actual job.
The organizational cost compounds quietly. New hires get onboarded onto "whichever system your team uses" as if that's a normal fact about the company rather than an unresolved failure. Cross-team reporting that touches both divisions requires manual reconciliation every time, introducing lag and error into numbers that go to leadership. And because the legacy system's vendor contract has technically been up for renewal twice now, the firm is paying for two CRM licenses indefinitely, with no one willing to own the migration project a second time given how the first attempt is remembered.
The autopsy
Root causes on record
- An exception was made without an expiration date or an owner. "Until the gap is closed" is not a deadline; it's a way of deferring the decision indefinitely.
- The hardest 20% of the migration was allowed to become nobody's job. Once the rest of the org moved on, the highest-complexity piece had no forcing function left.
- A manual reconciliation process quietly became permanent infrastructure. What was meant to be a stopgap outlived the person who built it, with no plan for that.
- No executive owner was assigned to actually finish the migration. Everyone could point to the unresolved state; no one was accountable for resolving it.
- The cost of the status quo was never made visible. Two license renewals and years of manual reconciliation labor never got totaled up as "the cost of not finishing," so it never competed for budget against other priorities.
Recommendation pending
Editor's note: this slot will point to a lighter-weight CRM or reporting layer suited to finally consolidating a legacy holdout division without repeating a full re-platform project.
What the post-mortem actually changed
Nothing yet — this file is still open. It's included here specifically because "zombie state" cases like this one are undercounted in most CRM war stories: they don't have a dramatic failure moment, just a slow, ongoing cost that never gets urgent enough to fix and never gets acknowledged as a failure everyone is already living with.