How do you audit utilities and energy invoices?

Margin drift is the gap between what a vendor contract says and what the invoice actually charges. On a utility or energy bill, the "contract" is usually a regulated tariff schedule plus whatever rider or negotiated rate sits on top of it, and almost nobody at the facility re-reads it after the meter goes live.

That is what makes utilities a distinct audit category rather than a subset of general AP review. The invoice is computer-generated, arithmetically consistent, and still wrong, because it is applying a rate structure that no longer matches the account.

Executive Summary

Utility and energy invoices carry a rate structure most AP teams never re-check after the account is set up: a tariff class, a demand charge, a power factor penalty, a fuel or purchased-power adjustment clause, and taxes and franchise fees that vary by jurisdiction. That gap opens quietly on a utility bill, because the meter reading is usually correct even when the rate applied to it is not.

The mechanism is structural. Utility tariffs change on a schedule set by the regulator, not by the vendor or the customer, and a facility's own load profile shifts as equipment is added or retired. Neither event triggers an automatic re-check of which tariff class or rate schedule still applies. The invoice keeps generating against whatever was set up last, and the account can sit on the wrong tariff, an unclaimed power factor exemption, or a stale contract rate for years before anyone looks.

What changes it is treating the utility account the same way a contract compliance audit treats a rate card: extract the tariff schedule and any negotiated rider, map every charge line to its authorization, and re-run that match whenever load, equipment, or the tariff itself changes. That is a control problem, not a payment-processing problem, and it is why utility drift survives normal three-way matching indefinitely.

1. How do you audit utilities and energy invoices?

You audit a utility invoice by reconstructing the rate the account should be on today, not the rate it was billed on, then comparing every line against that reconstruction. Pull the current tariff sheet or negotiated rider for each account, identify the tariff class, demand ratchet, power factor terms, and any fuel or purchased-power adjustment clause, and re-price a sample of recent bills against those terms. Any gap between the reconstructed charge and the billed charge is a finding.

The reconstruction has to start from the account's current characteristics, not its original setup. A facility that added equipment, shifted a shift schedule, or installed capacitor banks for power factor correction has a different load profile than the one the account was originally classed against. The tariff class that fit years ago may not fit now, in either direction.

Once the correct tariff and rider are established, the audit becomes line-by-line: demand charge against the metered peak, energy charge against the metered consumption, adjustment clauses against the published index for that billing period, and taxes and fees against the jurisdiction's actual schedule. Each of those is checkable independently, which is what makes utility audit different from a judgment call about usage.

This is the same discipline the diagnostic applies across every category: match the invoice to its authorization, not to itself.

2. What line items on a utility bill actually need checking?

A utility invoice bundles several independently variable charges into one total: the energy charge, the demand charge, a power factor penalty or credit, a fuel or purchased-power cost adjustment, and a set of taxes, franchise fees, and rider surcharges layered by the regulator or the municipality. Each has its own authorization source and its own failure mode, so checking only the bottom-line total against last month's total misses drift that a line-by-line comparison catches immediately.

A single bundled total hides which specific mechanism is driving a change. Splitting the bill into its component charges before comparing periods is what makes the comparison meaningful.

• Tariff class: The rate schedule the account is billed under. A mismatch after a load change re-prices every other line on the bill.
• Demand charge: Billed against the account's peak demand in the period, sometimes with a ratchet clause that holds the charge high for months after a single peak.
• Power factor adjustment: A penalty or credit tied to how efficiently the facility uses reactive power. It requires the meter data to verify, not just the invoice.
• Fuel or purchased-power adjustment: A pass-through cost tied to a published index. It should track that index, not drift independently.
• Taxes and franchise fees: Set by jurisdiction and exemption status. An exemption the facility qualifies for but never filed does not apply itself.

3. Why does utility rate drift go undetected for so long?

Utility rate drift survives because the invoice is internally consistent: the math is correct for the rate applied, the meter read is accurate, and the total is close enough to prior months that nothing trips a variance alert. Three-way matching in the ERP checks that a purchase order and receipt exist for a charge; a utility account rarely has either, since there is no purchase order for a kilowatt-hour. The control that would catch a wrong tariff class simply does.

Utility spend is usually coded to a recurring general ledger account and paid on autopay or a standing approval, which removes the human review step that catches drift in categories like freight or contract labor. Nobody opens the tariff sheet again once the account is live.

Variance alerts, where they exist, compare this month's total to last month's or to a seasonal baseline. A wrong tariff class produces a bill that is wrong in a stable, repeatable way, which looks like a normal bill to a variance check. It takes a rate reconstruction, not a trend comparison, to surface it.

The same blind spot shows up in other indirect spend categories where a charge is technically correct against the wrong reference table.

4. How do power factor penalties and demand ratchets create hidden cost?

A power factor penalty charges extra for reactive power the facility draws but does not use productively, and a demand ratchet holds a percentage of the highest peak demand as the minimum billing basis for months afterward, even if actual demand drops. Both mechanisms are legitimate parts of industrial tariffs, but both are also correctable: capacitor correction addresses power factor, and load scheduling addresses peak demand. An account paying either without knowing it is a candidate for correction.

Both mechanisms recur automatically on every subsequent bill until someone acts, which is why they carry disproportionate weight over a multi-year period.

A. Power factor correction

Reactive power draw depends on the equipment mix at the facility, particularly motors and induction loads. A power factor below the tariff's threshold triggers a penalty that recurs every billing period until corrected, with capacitor banks sized to the load. The audit question is whether the penalty on the bill matches the facility's actual power factor as metered, and whether correction equipment already installed is being credited.

B. Demand ratchet exposure

A ratchet clause sets the billing demand as the higher of current demand or a percentage of the highest demand recorded in a trailing window, often 11 or 12 months. One unusual peak, a startup event, a temporary process change, can set a raised billing floor that persists long after the load returns to normal. Checking whether the ratchet basis still reflects a genuine peak, rather than an anomaly, is a distinct check from the monthly bill review.

5. Which utility accounts are worth auditing first?

Start with accounts where the load has changed since the tariff was last reviewed: a facility that added or retired production equipment, changed shift patterns, or completed a capital project affecting electrical or gas demand. Accounts on a demand ratchet or a power factor sensitive tariff class are worth checking next, since those mechanisms compound quietly. Accounts billed on a flat, unreviewed contract rate for multiple years without a tariff comparison are the last category, because staleness itself is a.

A facility rarely tracks when an account's tariff class was last checked as a field anywhere, which is part of why the check gets skipped. Building that list means cross-referencing utility accounts against known capital projects, equipment changes, and shift schedule changes over the audit period.

Multi-site organizations should also compare rate structures across sites in the same utility territory. Two accounts on the same utility, similar load, and different tariff classes is a flag worth investigating before pulling meter data.

This prioritization mirrors how a maintenance and repair invoice audit starts: find where scope or terms changed and the billing never caught up.

6. What does a fuel or purchased-power adjustment clause actually track?

A fuel or purchased-power cost adjustment passes through the utility's own input cost changes to the customer's bill, tied to a published index rather than a fixed rate. Per the US Bureau of Labor Statistics Producer Price Index for Industrial electric power, series WPU0543, read September 6, 2026, the July 2026 index value was 341.966, up 1.2% year over year, showing this input cost moves independently of any single facility's usage. An adjustment on an invoice should track.

The audit check here is not whether the adjustment exists. It is legitimate and contractual. The check is whether the adjustment applied on a given invoice matches the index or fuel cost basis stated in the tariff or rider for that billing period, rather than an older or substitute figure.

Because these adjustments recalculate every billing cycle, an error introduced once, a wrong base rate or an outdated index reference, repeats on every subsequent invoice until someone reconstructs the clause and re-checks it against the current published basis.

This is the same logic behind a software true-up audit: a formula that recalculates automatically still needs its inputs checked periodically, not just its arithmetic.

For the wider pattern this sits inside, start with the margin drift guide.