Every fixed asset implementation contains one beautifully optimistic moment. Somebody says it out loud, usually in week two: once we are in the new system, the data will be clean.
It will not. I wish it worked that way, and I have wished it in a lot of conference rooms. A conversion is a transfer, not a treatment. Whatever goes in comes out the other side — same errors, same gaps, same undocumented decisions, now wearing the credibility of a new platform and considerably harder to question.
What legacy fixed asset data usually contains
After enough conversions, the inventory of problems is almost comfortingly consistent:
- Assets that no longer exist. Ghost assets from disposals never recorded — often 5-15% of line count in an environment with no disposal trigger.
- Descriptions that identify nothing. "EQUIPMENT," "2014 ADDITIONS," "VARIOUS." No serial, no location, no tie to anything you could point at.
- Bulk lines hiding many assets. One row for a $2.4M rollout across nineteen sites. Unretirable, untransferable, unverifiable, and technically balanced.
- Lives applied by habit. Whatever the default was when the prior system was configured, honoured faithfully ever since regardless of asset type.
- Book and tax that no longer agree. Two registers maintained in parallel, reconciled once a year with adjustments nobody can now explain.
- In-service dates equal to invoice dates. Convenient, extremely common, and wrong — with knock-on effects running through depreciation and tax for years.
- Accumulated depreciation that does not tie. The sub-ledger and the general ledger differ by an amount that got plugged during a close in 2017.
Why implementations skip the cleanup
Not out of ignorance, and not because anyone is lazy. The pressure is structural.
An implementation is scoped, budgeted and staffed around configuration and go-live. Data cleanup has no natural owner in that scope: the vendor's job is to load what you give them, and the internal team is already absorbing a project on top of a full-time role. Cleanup is the one workstream with no deadline of its own, so it is the one that slides.
Then go-live arrives, the balances tie to the old system — which is the acceptance test — and the errors are now baseline. Nobody revisits them, because revisiting them means standing up in a meeting to explain why the numbers in the shiny new system are moving.
What clean conversion actually involves
- Profile before you plan. Count nulls, duplicate serials, zero-cost assets, fully depreciated assets still active, assets with in-service dates that predate the entity itself. You cannot scope the work until you know its shape, and the shape is always a surprise to somebody.
- Reconcile the sub-ledger to the GL first. If they do not tie before conversion, they will not tie after — you will simply have imported an unexplained difference and given it a new home.
- Attack ghost assets with targeted verification. Not a wall-to-wall inventory. Sample by class and location, extrapolate, and verify the high-value populations directly.
- Decompose bulk lines by materiality. The big ones are worth breaking into real assets with real locations. The immaterial ones can carry forward with a documented rationale and a clear conscience.
- Rebuild the life and class mapping deliberately. Conversion is the one moment when reclassifying is cheap. After go-live it is a change with an audit trail and an explanation attached to it.
- Fix book/tax divergence at the source. Land both registers from a reconciled position instead of importing a historical gap and inheriting the argument.
- Document every decision. The rationale behind each rule is what makes the resulting register defensible to an auditor two years from now. Undocumented cleanup looks identical to an unexplained adjustment, and auditors are not obligated to assume good intent.
The consulting part is not optional
The reason this work needs judgement rather than a mapping spreadsheet is that most conversion decisions are not technical at all. Whether a bulk line gets decomposed, whether a class gets re-lived, whether an unverifiable asset gets written off — each one is a judgement about materiality, audit exposure and how the register will actually be used. A migration specialist cannot make those calls, and should not be asked to. Someone who understands the accounting consequence has to be in the room.
Clean data also depends on identity work upstream. If projects and assets have no consistent identity to begin with, cleanup has nothing to group on.
A new fixed asset system is a real upgrade in control, reporting and effort — and the tooling arriving now, automation and AI-assisted analysis included, is genuinely worth being excited about. None of it is a filter. Whatever is loaded on day one is what the organization will be reporting from for the next decade, and every clever thing you layer on top will be reasoning from it. Day one is the day worth spending on.
