Four Numbers on Your Electric Bill Decide If Solar Pays

Georgia Power's residential tariff prints a summer energy charge of 15.0828 cents per kWh for usage above 1,000 kWh. EIA puts the average price of residential electricity in Georgia at 15.84 cents per kWh for May 2026 (Electric Power Monthly, Table 5.6.A, released 23 July 2026, read 20 August 2026). Two numbers within a cent of each other, arrived at by completely different routes, and a proposal built on either one is wrong.

A kWh on that tariff costs about 21.29 cents. The gap is four rider schedules that the residential sheet names and does not price.

That kind of gap used to get absorbed. With a 30% federal credit knocking a third off the installed price, a rate error of a few cents moved the payback year by a fraction and nobody noticed. The credit is gone — section 25D closed on 31 December 2025 — so the whole payback now rests on the savings side of the fraction, and that side is built out of four numbers that come off your own bill.

Number one: twelve monthly kWh figures, not one annual total

The annual total is the number every proposal asks for and the least useful of the four.

Solar production is seasonal and so is your load, and they are not seasonal in the same shape. A house that burns 1,800 kWh in July and 750 in January has a very different relationship with a rooftop array than one running a flat 1,050 every month. Both total 13,200 kWh a year. You need the twelve figures.

They are on the bill, usually as a bar chart of the last thirteen billing periods. Better, take the download: many utilities publish interval data through Green Button Download My Data as an XML or CSV file at whatever resolution the meter records, commonly 15-minute or hourly (Green Button Alliance, read 20 August 2026). That is the same file sizing a battery from an actual load profile runs on, so pulling it once serves both jobs.

Four things spoil a twelve-month history, and in all four the kWh column looks perfectly ordinary:

• Budget billing. Georgia Power's R-31 offers a budget bill that "has the effect of leveling the Customer's monthly billing amount" (Schedule R-31, effective with bills rendered for the billing month of June 2026, read 20 August 2026). Twelve levelled dollar amounts say nothing about seasonal shape. The kWh column still does.
• Estimated reads. A month estimated and then trued up puts a trough next to a spike that never happened.
• A load that arrived mid-year. An EV, a heat pump, a second freezer. The last twelve months then describe a house that no longer exists.
• Existing PV. EIA flags this one explicitly: net-metered systems reduce purchases, so in states with a lot of rooftop solar "the amount of household electricity consumption may be a lot higher than household electricity purchases" (EIA FAQ, How much electricity does an American home use?, read 20 August 2026). Your bill measures purchases.

That same FAQ gives the sanity check: 10,791 kWh a year, the 2022 average for electricity purchased by a U.S. residential utility customer, with Louisiana highest at 14,774 and Hawaii lowest at 6,178. If your twelve figures sum to something far outside that band, the reason is worth knowing before it becomes an input.

Number two: the cents that vanish when a kWh vanishes

EIA defines average revenue per kilowatthour as total monthly revenue divided by total monthly sales (EIA glossary, read 20 August 2026). Fine statistic, bad input. It contains the fixed charge, which no kWh you avoid will reduce, and it flattens seasons and tiers that differ by a factor of two.

What you want is the marginal rate: the amount the bill falls by when one more kWh does not pass through the meter.

Georgia Power shows this clearly because the price is split across five schedules. The residential sheet carries the base energy charge; four riders sit on top, each its own PDF (all read 20 August 2026):

EIA's glossary defines the base bill as "a charge calculated by taking the rate from the appropriate electric rate schedule and applying it to the level of consumption" (EIA glossary, read 20 August 2026), which is what the two percentage riders attach to. So, for a secondary-distribution customer outside city limits, one summer kWh above 1,000:

base                       15.0828
ECCR   15.0828 x 0.130205 = 1.9639
DSM-R  15.0828 x 0.011969 = 0.1805
FCR                       + 3.8069
                          --------
subtotal                   21.0341
MFF    21.0341 x 0.011995 = 0.2523
                          --------
marginal rate              21.2864 c/kWh

That is 41% above the printed figure. Inside city limits it is 21.68 cents, and county sales tax sits on top of both, so treat these as floors.

Run the winter charge — 8.2116 cents per kWh, all usage, October through May, with that season's fuel rate of 3.8561 — through the same stack and you get 13.39 cents. Same house, same meter, same tariff, and the summer figure sits 59% above the winter one.

The last line is the one step the schedules do not settle. ECCR-15 and DSM-R-16 both attach to "base bill calculations." MFF-11 says its rates apply "to the total revenues of each bill" and then prints them as a percentage "times usage revenue" — and none of the three sheets defines usage revenue or says whether it takes in the fuel rider. The stack above uses the wider reading. The narrower one, franchise fee on the base energy charge alone, gives 21.2150 cents: about seven hundredths of a cent lower. Everything above that line is identical either way, and your own bill settles which reading your utility applied.

California packages the same complexity differently. PG&E's E-TOU-C sheet prints total bundled rates, so there is nothing to stack — but plenty to split. Summer peak is $0.52240 per kWh and summer off-peak $0.39940; winter peak $0.39757 and winter off-peak $0.36757; usage inside the baseline quantity gets a credit of $0.08140 (Schedule E-TOU-C, Cal. P.U.C. Sheet No. 61364-E, Advice 7921-E, Decision D.26-04-036, effective 1 June 2026, read 20 August 2026). Peak runs 4 to 9 p.m. every day including weekends, and summer is service from 1 June through 30 September.

So on one tariff a kWh is worth anywhere from 28.617 cents (winter, off-peak, inside baseline) to 52.240 cents (summer, peak, above baseline). A factor of 1.83, decided by the hour and the month. "Your rate" is not a number on a bill like that. It is eight of them.

That marginal figure also decides whether electrifying anything pays. It is the E in the breakeven COP formula for a heat pump, and feeding an average into that produces the same class of error there.

Number three: the part of the bill solar cannot reach

Fixed charges are billed per month or per day whatever the meter says. Generation on your roof does not touch them.

Georgia Power's basic service charge is $0.4603 per day on R-31, and the tariff sets the minimum monthly bill as that charge plus ECCR plus DSM-R plus the franchise fee. Push the percentage riders through and the daily figure becomes $0.53205, or $194.20 a year — what the utility bills a house whose annual consumption is zero.

PG&E turned the same idea into a base services charge, sorted by income tier and stated per customer per day. The income-tier language entered E-TOU-C with the revision effective 1 March 2026; the amounts below sit on the rate sheet effective 1 June 2026:

The tariff is explicit that this is not new revenue: customers "will pay a monthly base services charge, with the base services charge revenue used to reduce their $ per kWh energy rates." For a payback calculation that trade is not neutral — it moves dollars out of the column solar can eliminate and into the column it cannot. The same sheet carries a California Climate Credit of $36.18 per household, described there as a semi-annual payment occurring in the August and September bill cycles, which arrives whatever the roof does.

So a proposal claiming to eliminate the electric bill overstates annual savings by at least the fixed charge. Subtract it before dividing into system price. For a Tier 3 PG&E household that is $289.60 a year no array can reach — $7,240 across a 25-year system life at today's rate, before any escalation.

Number four: the code that says which tariff sheet is yours

The first three numbers all come out of a document, and the code tells you which document.

Bill layouts are not standardised, so there is no one spot to point at. On most bills it turns up in the detail-of-charges section rather than the summary, on a line labelled Rate, Rate Schedule, Service Plan, or Tariff, and it looks like R-31 or E-TOU-C. If nothing on the paper resembles a code, the account portal names the plan you are on. Either way, once you have it the utility's tariff library turns everything above into something you can check.

Two things about the code are easy to miss.

It carries a version. PG&E's E-TOU-C rate sheet is Cal. P.U.C. Sheet No. 61364-E, cancelling 61249-E, filed under Advice 7921-E, effective 1 June 2026. Georgia Power's residential sheet says "With Bills Rendered for the Billing Month of June, 2026," and the riders carry their own dates that do not all match: FCR-27, ECCR-15 and DSM-R-16 also read June 2026, while MFF-11 still reads January 2025. A tariff PDF saved last year is a different document under the same name, which is why every figure here carries the date it was read.

And the code is usually a choice, or a default you were placed on. E-TOU-C states that it "is available to residential customers on an opt-in basis." Going solar can change it again: the schedule lists "Customer transitions to Net Energy Metering Successor Tariff" among the circumstances under which its twelve-month bill protection credit is paid out automatically. The four numbers you extract today describe the tariff you are on now, and the array will very likely be priced under a different one.

A 13,200 kWh year, priced three ways

Take the Georgia house from the first section: 1,800 kWh in each of four summer months, 750 kWh in each of eight winter months, R-31, secondary distribution, outside city limits, 30-day billing periods, sales tax excluded.

Summer tiers after riders run 13.9940 cents for the first 650 kWh, 20.6980 for the next 350, and 21.2864 above 1,000. Winter is a flat 13.3939.

Divide and the average price is 17.65 cents per kWh. Not the July rate of 21.29, not the January rate of 13.39, and not the rate at any moment of the year.

Now the part that catches people sizing up. Offset kWh come off the top tier first, so the value of each one falls as the array grows:

An array covering the first 800 kWh of a July month is genuinely worth 21.29 cents a kWh. An array covering the whole month is worth 18.54 — a 13% haircut that never appears on the quote, because the quote multiplied one rate by total production. Worse, the error is largest on exactly the oversized systems that produce the most impressive annual figures, and it stays invisible unless you rebuild the bill tier by tier.

The fifth number is not on your bill

All of the above prices the kWh you stop buying. It says nothing about the kWh you send back, and those are priced by a separate document — a net metering, net billing, or buy-all/sell-all tariff, depending on the state — that a pre-solar bill has no reason to mention. Skipping it is how a proposal ends up valuing exported kWh at retail in a state that pays an avoided-cost rate a fraction of that size. Find that tariff before accepting any annual savings figure.

Four files and one spreadsheet column

Twelve monthly kWh figures from the bill history or the interval download. Your rate schedule code. That schedule's PDF, plus every rider it names by title — Georgia Power's names four; a utility publishing total bundled rates will name none.

Then build the marginal rate for each season and, on time-of-use, each period. One row per rate block, arithmetic written out the way it is above so somebody else can check it. Keep the fixed charge on its own line in dollars per year, because it belongs on the other side of the ledger.

When a proposal arrives, find its rate assumption and compare. A single flat cents-per-kWh figure applied to a tiered or time-of-use tariff means the payback came out of a formula that cannot be right; only the direction of the error is still open. Load reduction moves the same inputs: the measures ranked by an energy audit run before any panel goes up come off the top tier too, at the same 21.29 cents, and cost far less per kWh removed.