Retire the rate codes before you replace the billing system

Most failed billing conversions I have seen were blamed on the software, the vendor or the integrator. I think the more common cause is the utility's own history. Over thirty years a utility's tariffs gather riders, pilot programmes, special contracts and grandfathered rates, and each survives because removing it requires a decision nobody is asked to make. A billing replacement carries all of them into a new system, where every one must be configured, tested and explained. My position is that a cooperative should require its board to retire each rate code it no longer needs before conversion work begins, and should accept delay, and even the loss of participants, to keep that condition.

Eleven hundred ways to bill a kilowatt-hour

In 2012 I was responsible for customer information and billing at Hollensby Electric Company, an investor-owned utility in Kansas, when its new billing system went live. In the first two bill cycles about 38,000 bills were delayed or estimated, out of roughly 560,000 customers. The review I presented to the audit committee early the following year traced most of the failures to one fact. We had converted more than 900 rate and rider combinations, many applying to a handful of accounts, and the test plan could not cover them all. The defects clustered in exactly the rates nobody had calculated by hand for years, because nobody remembered how they were meant to behave.

Merrowind's shared billing system, used by fourteen of our member cooperatives, presented the same problem in a larger proportion. Built in the late 1990s on a midrange platform, it would lose vendor support at the end of 2027. Across 236,000 accounts it held 1,140 rate codes. Some were legitimate and complicated: irrigation contracts with seasonal demand charges, oil-field pumping loads, prepaid plans and power cost adjustment riders that differed by member. Others had last produced a bill in 2014. One member had nine residential rates, and its own billing supervisor could explain four.

Rate codes in a cooperative matter beyond the bill. Capital credits, the allocation of the cooperative's margins back to the members who generated them, are usually calculated from what each member has purchased. A rate error can therefore follow a member into equity allocations years later, which is a further reason to understand every code before moving it.

The board decides, not the vendor

Retiring a rate is not a technical task. It moves members from one tariff to another, sometimes changing their bills, and in a cooperative that decision belongs to the member's elected board. Vendors and integrators are poorly placed to ask for it, because their contracts reward configuring whatever the customer specifies. Technology departments are poorly placed too. We can count rate codes, but we cannot tell a board which farmers will pay more under a consolidated irrigation tariff.

So I made rate review a condition of conversion. Before any member entered a wave, its rates staff and my analysts prepared a schedule showing every code, the accounts on it, the revenue it billed in the last year and the bill effect of moving those accounts to the nearest surviving rate. Each board then decided which codes to keep. A board that wanted to keep a rate serving only a few accounts could do so, accepting the configuration and testing cost. Most chose not to. The twelve members that stayed in the programme held 940 of the original codes and retired all but 296.

We changed the testing as well. Each wave runs two full bill cycles in parallel on the old and new systems, and any bill that differs by more than 2% is held until a person has examined it. Member service representatives, who will explain the bills at the counter and on the telephone, test the system before go-live. They found defects our analysts had missed, because they knew which members would call.

What insisting cost us

The condition cost us two participants. Our largest member, with 58,000 accounts, wanted to convert first to meet its own date for a new member portal, and its board did not want its rate decisions tied to our timetable. In 2023 it bought its own system, and a smaller member followed. We lost about US$2.3 million a year in fees, which raised the cost per account for the remaining twelve members by about 6%, and the programme will finish nine months later than first planned. The total cost will be about US$43.5 million against US$38 million approved, largely because parallel cycles and member staff overtime cost more than we estimated.

I think the cost was justified, though not every part of it. Three waves have converted ten cooperatives and 120,000 accounts, in October 2024, April 2025 and February 2026. First-cycle exceptions were 0.6% in the first wave and 0.3% in the third, and no member has had to send an estimated bill. The old system will be switched off after the fourth wave in 2027, months before its support ends.

The part I would change is how I presented the condition. I described rate review as settled policy before asking our largest member what its portal date required. A wave built around that date, with the rate review running alongside, might have kept it in the service. Holding to a principle does not require holding to its first timetable.

There are objections I take seriously. Rate consolidation can raise some members' bills, and a cooperative should not use a systems project to make tariff changes it would not otherwise defend; each change should stand on its merits before the member's board. Some legacy rates exist for good reasons, such as contracts with large customers, and should survive. And a utility facing an imminent support deadline may not have time for a full review, in which case it should at least list the codes that billed nothing last year and retire those.

For any cooperative or municipal utility planning a billing replacement, the first question is not which vendor to choose. It is how many rate codes the utility has, how many produced a bill last year and who has authority to retire the rest. If that is not answered before the contract is signed, the conversion will answer it, usually in the first bill cycle.