How we compute true cost

Inputs, and where they come from

Every input to a assessment is a document you can independently check.

• Your interval data — 15-minute readings from Smart Meter Texas, the state-mandated system your utility reports into. Access is read-only and authorized by you directly through SMT's own confirmation email. We never hold a utility login.
• Plan terms — the Electricity Facts Label (EFL) each retail provider is legally required to publish. We sync the state's Power to Choose export and each provider's own official listing daily, archive every EFL version, and verify each one: we re-derive the document's own published 500 / 1,000 / 2,000 kWh prices from what we read. If they don't reproduce, the plan is not served. Your assessment cites the exact document it priced.
• Delivery charges — your utility's current residential delivery rate (a fixed monthly charge plus a per-kWh charge), taken from the Public Utility Commission's monthly rate report and cross-checked against what current EFLs print. Applied identically to every plan, because you pay it identically on every plan.
• Your current plan — from the bill or contract you upload: provider, plan, end date, early-termination terms. The assessment's savings number is measured against it, and only when it can be priced honestly.

The calculation

For each candidate plan we simulate the bill you would have received in each of your actual billing cycles:

• Time-of-use windows are applied hour by hour from your interval data. A “free nights” plan gets its free hours honored exactly — and its daytime rate applied to your actual daytime hours, which is where these plans win or lose.
• Bill credits are evaluated against each of your actual months. A credit that requires 1,000 kWh counts only in the months your home crossed 1,000 kWh — never assumed every month the way the advertised price assumes.
• Base fees and minimum-usage fees are applied per billing cycle, including the months your usage would have triggered a penalty.
• Short-term plans (under 12 months) are never annualized. Each is priced over its real term using the matching months of your history and compared with the top plan over that same window.
• Variable-price plans are priced from the provider's own published rate history for the months after the first. A provider with no usable history is listed, not ranked.
• Delivery charges are added to every plan. Taxes apply equally across plans and are not ranked on.

Plans are ranked by true cost. The advertised “average price at 1,000 kWh” never appears as the ranking number — mostly you will see how far it sits from yours.

The five traps, and the test that catches each

• Bill-credit plans. A large credit at exactly 1,000 kWh makes a plan look unbeatable at 1,000 kWh and terrible everywhere else. Test: price only on your actual monthly usage; never surface the advertised rate.
• Time-of-use mismatch. Free-nights plans assume a night-shifted home. Test: compute your actual share of each plan's own hour windows from interval data, and say so when the plan's assumption doesn't match you.
• Short-term annualization. A 5-month plan covering the cheap half of the year looks best in market when annualized. Test: price it over its real months and surface the re-shop date as a risk.
• Switching-window timing. Early-termination fees are often waived only inside a window before the contract ends. Test: read the end date and the fee terms, and compute the window where switching is free.
• The incumbent's renewal offer. The renewal a provider mails is often worse than its own new-customer plan. Test: always price the incumbent's entire current lineup, not just the renewal.

A worked example, line by line

One home, one plan, every step of the arithmetic. The home uses 14,400 kWh a year. The plan is fixed-rate with a usage credit: energy charge 11.2¢/kWh, no base fee, a $25 bill credit in any month at or above 800 kWh. Delivery is $4.39/month plus 3.7351¢/kWh.

Shaded months fall below the 800 kWh threshold — the credit is forfeited. It lands 8 of 12 months; the advertised math assumes 12 of 12. That gap is $100 a year before any rate difference.

This example is a flat-rate plan, so monthly totals suffice. For time-of-use plans the identical procedure runs at interval granularity — the free/paid window assignment happens per reading, which is why two homes with the same monthly total can get opposite assessments on the same plan.

Assumptions, stated plainly

• Your next year looks like your last. Rankings assume your coming months resemble the history we priced. When you tell us something changed — an EV, a pool, working from home — chat models it as an explicit scenario against your real hours and shows both versions. We never silently assume your life changed.
• Monthly data is a fallback. Without interval data, time-of-use plans are estimated from monthly totals with an assumed hourly pattern and carry a plain-words disclaimer until your intervals arrive.
• Prices go stale in days. Every quote is timestamped and re-verified against the current EFL before you enroll.

What we can't price, honestly

• Plans whose EFL we can't verify. If our reading of a document does not reproduce its own published prices, the plan is listed as unpriced — never guessed at.
• Variable plans with no published history. Listed, not ranked: the price after month one cannot be estimated honestly.
• Prepaid plans. Not ranked; their fee structures depend on payment behavior we don't model.
• Solar export. We price your consumption side. Export credits are not priced, and homes with generation see that noted.
• Multiple meters. Each meter is analyzed on its own intervals. We never blend meters into one verdict.

Corrections

Find an arithmetic error in any assessment and we will fix it and say so here, with the date. A methodology that expects to survive audit should invite one — please try to break it. Write to support@kilowhiz.com.