What Happens to Your Unitization Backlog When Your Most Seasoned Accountant Retires

Every plant accounting group I talk to has the same dependency. The person who knows, without looking it up, that a certain vendor’s line item needs splitting before they’re capitalized, that the regulator station rebuilds get coded one way in the southern district and another way in the north, and that a work order from a specific crew is going to arrive thin on material detail, so somebody should call before it posts.

That knowledge isn’t in the capitalization policy. It isn’t in the property unit catalog. It’s in one person’s head, and it took them twenty years to build that knowledge.

When they retire, the team doesn’t lose one person’s worth of capacity. It loses more than that, because everyone left behind now has to research what that person simply knew. The backlog doesn’t hold steady. It grows.

FERC has already documented what happens to the unitization backlog when experienced staff leave.

FERC has already put this on record. In a 2016 audit of Westar Energy, the agency found that completed construction balances hadn’t moved from Account 106 into Account 101, Electric Plant in Service. The findings pointed to employee turnover in the property accounting department as one of the causes, alongside process inefficiencies and system issues.

Turnover in property accounting, named in a federal audit a decade ago is a reason capital sat unclassified. And the replacement pipeline has not caught up since. The IEA’s World Energy Employment 2025 report puts the industry’s entry-to-exit ratio at roughly 2.4 workers nearing retirement for every worker under 25 coming in, and tighter still in grid-related professions. The people who hold twenty years of coding judgment are not being replaced one for one.

Why unitization depends on judgement no policy document captures

Unitization is uniquely exposed to knowledge loss because most of its decisions were never written down, and the standard continuity plans don’t replace what leaves.

Unitization is one of the most judgment-heavy steps in the close. Charges arrive from the field as vendor and PO data that was never structured for accounting, and someone has to interpret them line by line and allocate them to specific retirement units and utility accounts. On a single large work order, that can mean working through thousands of individual charges to arrive at a few dozen asset records. We have seen a work order carrying roughly 3,500 charges resolve into about 30 component lines. Each of those allocations is a judgment call, and most of them aren’t answered by any policy document.

That is why the usual continuity plans underperform here:

• Hiring ahead of retirement buys you capacity, not judgment. A new plant accountant is productive on the clean work orders quickly and slow on the exceptions for years.
• Documenting the process captures the rules that were already written down. The value of the retiring accountant was everything the rules didn’t cover.
• Bringing in outside help clears a portion of the current backlog during the engagement. When it ends, the knowledge leaves with the consultants, and the backlog starts building again.

The cost, meanwhile, is real and recurring: most mid-size utilities carry two to ten full-time people, or outside consultants, on unitization alone, a standing line item in every annual budget. Under month-end pressure, the exceptions get deferred first, so the hardest judgment calls are exactly the ones that pile up.

How Intelligent Unitization captures that judgment before it walks out the door

Intelligent Unitization reads your team’s own coding history to draft line-by-line unit estimates, so a senior accountant’s judgment keeps working after they’ve left

The reason I am optimistic about this problem is that the judgment isn’t actually lost. It is recorded, thousands of times over, in the decisions your team has already made.

That is the premise behind Intelligent Unitization. The model reads a work order’s transaction and material detail against your property unit catalog, your FERC account structure, and your historical asset records, including how your team has coded similar charges before, and returns a recommended unit estimate line by line. Each line carries a confidence score and a plain-language explanation of why it was predicted that way.

Three things about that approach that matters for the continuity problem specifically:

1. It learns your organization, not a general one. Each model is trained only on your data, in a single-tenant environment. The patterns it encodes are your senior accountants’ patterns.
2. The accountant stays in control. Recommendations are drafted, never posted. A reviewer accepts, adjusts, or rejects each line, and nothing reaches the system of record until it clears PowerPlan business rules and lands in a reviewable audit trail.
3. Corrections compound. When a reviewer overrides a line, that correction feeds back into the model. Institutional knowledge keeps accumulating instead of walking out the door.

What I keep hearing from accountants in early engagements is that the confidence scores and the written rationale are what turn skepticism into adoption. Nobody wants to approve an answer they can’t interrogate. Being able to open a line and read why it was predicted is what makes the recommendation something a reviewer can stand behind in front of an auditor.

It’s also the fastest way to evaluate the product: take a work order your team has already unitized, and compare the model’s line-by-line recommendation against what your senior accountant decided.

The last contribution

The most durable thing a retiring accountant leaves behind isn’t the backlog they cleared. It’s the pattern set that keeps clearing it.

There is a version of this where the senior plant accountant’s retirement is a capacity gap you staff around, and the backlog quietly gets worse for three years.

The difference shows up in the first close after that person is gone. Instead of a queue of exceptions waiting on the one person who knew the answer, every line arrives with a recommendation, a confidence score, and the reasoning already attached, and the reviewer spends the month judging the hard calls instead of reconstructing them.

There is another version where that person’s most durable contribution isn’t the work they clear before their last day. It’s the pattern set they leave behind, still working, still improving, still explaining itself to whoever reviews it next.

If you want to see how the recommendations and rationale actually look against real work orders, the on-demand launch webinar that Jim Dahlby and I hosted is available.

Intelligent Unitization is available to early adopters today. Reach out to us to schedule a demo and learn more.

---

At PowerPlan, we work with the largest utilities in North America and support their management of more than $4 trillion in regulated assets. What makes that scale significant is the integrity of the data underneath it. The PowerPlan NXT platform seamlessly carries each asset’s regulatory, accounting, and tax details forward without re-entry or re-mapping, so the integrity of the data is defensible before regulators and tax authorities. When terms change and cost allocations shift with them, that asset-level detail is what lets you turn the change into information the rates and regulatory group can actually use – in hours, not weeks. Start the conversation with us today.