How we make sure you can trust what you read here

Compare Power makes money when a reader switches to an energy plan through our platform. That is a commercial relationship, and we do not hide it. What we refuse to do is let that relationship decide what our editorial team writes. Plan comparisons are ranked by consumer fit — the all-in price at your actual usage, the early-termination terms, the fee structure, the renewable mix — not by what a retail electric provider (REP) pays us.

Editorial and commercial sit in separate rooms. Our editors do not see partner pipeline. They do not attend REP sales calls. They do not have quotas. If a partner pulls out of the marketplace because an editor wrote honestly about a bill-credit trap or a teaser rate, that is the partner's problem — not ours, and certainly not yours.

Practical rule: no editorial piece may be modified to make a partner look better. An editor can correct a factual error. An editor cannot soften a legitimate criticism because someone asked nicely.

If a page claims an average Texas household uses 1,176 kWh per month, that number comes from the U.S. Energy Information Administration (EIA), not from the back of a napkin. If a page quotes a rate, the rate is pulled from the provider's Electricity Facts Label (EFL) and Terms of Service (TOS) — the documents the Public Utility Commission of Texas (PUCT) requires every REP to publish — and the page is stamped with the date we read it.

We have a cadence. Market-condition pieces (wholesale prices, ERCOT capacity) get reviewed monthly. Regulatory explainers (PUCT rules, ERCOT protocols) get reviewed when the rule changes, and at minimum annually. Provider plan data refreshes live through our sync pipeline; if a plan is stale by more than 48 hours, we flag it rather than publish an unverified number.

No public-facing content ships without a named editor signoff. That editor's name is attached to the piece. If the piece is wrong, you know who to call. "The team" is not a person. "The editorial bot" is not a person. A human being signed the work.

Google calls it E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness. We call it "show your work." Our editorial team combines decades of direct Texas-market experience: former REP employees who saw the churn math from the inside, journalists who covered the 2021 winter storm and its fallout, a licensed broker on staff for commercial plans, and fact-checkers whose only job is to verify the claims before anyone else reads them.

Every author on our site has a public profile. Real name, real photo, real bio, real areas of expertise. We publish credentials, licenses, prior work, and, where relevant, the schools and news outlets our writers came from. The directory lives at /about/editorial/ and it is not decoration.

For regulated or technical topics — PUCT rule interpretation, ERCOT market mechanics, solar-buyback economics — we require a second reviewer with subject-matter expertise before publication. If we cannot find one on staff, we name the outside expert who reviewed the piece. We do not fake authority by hiring a random "energy consultant" to rubber-stamp content.

Primary sources first. For Texas deregulated-market content, that means the PUCT (particularly Title 16 Chapter 25 of the Texas Administrative Code — the "Substantive Rules Applicable to Electric Service Providers"), ERCOT protocols and market notices, the EIA, and the provider-filed EFL/TOS documents on Power to Choose. When we cite a rule, we cite the section — for example, PUCT §25.475 for disclosure requirements — and we link directly to the current version on PUCT.gov.

We mark estimates as estimates. A regulated number (the PUCT-mandated energy charge at 1,000 kWh on an EFL) is different from an estimate (the cost of running a home air conditioner in August). Regulated numbers get a citation. Estimates get assumptions stated plainly: "We assume a 1,500 sq ft home at 75°F, based on ASHRAE climate zone 2B." If we cannot cite it, we do not treat it as fact.

We do not cite other aggregators as sources. If the only place a number appears is on a competitor comparison site, that is not a primary source — that is someone else's unverified claim. We go back to the EFL, the rule, or the agency.

We use AI tools in editorial production — for outlining, for first drafts of templated sections, for translation, for research summarization, and for fact-lookup against our own knowledge base. We are public about that because pretending otherwise is the kind of shell game that erodes reader trust in the industry already.

The rule is simple: no AI-generated claim goes public without a named human editor verifying it against primary sources. AI can draft; it cannot ship. When a piece was assisted by AI in a way a reader would care about — when the draft, the translation, or the data narrative started with an AI tool — we say so. Disclosure happens at the piece level, not buried in a sitewide footer.

Spanish translations on this site are initially produced with a machine-translation engine (DeepL), then reviewed by a bilingual editor for voice and accuracy before publication. A translation is not a publication event until a human signs it.

Texas deregulation started with Senate Bill 7 in 1999 and went live for most of the state in January 2002. That is the baseline context for almost everything we write. We know the difference between a utility (the transmission and distribution service provider, or TDSP — Oncor, CenterPoint, AEP Texas, TNMP) and a REP (the retailer you buy your plan from). Most of our readers have been told these are the same thing. They are not.

We write about the 2021 winter storm (Uri) honestly because the people we serve lived through it. We name the wholesale-price-exposed plans that got customers handed five-figure bills. We explain how HB2208 and related post-Uri reforms changed what REPs are allowed to offer. We do not pretend the 2024 capacity conversations at ERCOT are settled when they are not. When a PUCT rulemaking affects how a plan can be marketed, we update the affected content.

If the regulatory landscape matters to the plan you are looking at — time-of-use pricing under PUCT §25.498, variable-rate disclosure under §25.475, renewable-energy product labeling under §25.476 — the page will tell you which rule is in play and what it requires. You should not need a Ph.D. to read a rate plan.

We write for a Texan who is tired of being talked down to. That means short sentences when short sentences work, plain nouns instead of marketing abstractions, and specific numbers instead of vague reassurances. "Customers typically save" is not a claim. "On a 1,500 kWh monthly usage in Oncor territory, the median plan we list is 3.2 cents per kWh cheaper than the provider's default renewal rate" is a claim.

We name industry bad behavior when it is real. "Up to 99% savings" is not a compliment we pay to our peers — it is what we call out. Teaser rates, bill-credit usage traps, free-nights-and-weekends plans with jacked-up day rates, confusing minimum-usage fees, renewal letters designed to run out the clock: we explain the mechanism, then we explain how to avoid it.

We do not preach. We do not pretend the only honest electricity plan is the one we ranked first today. Sometimes the right move is staying put. Sometimes a competitor has a better deal for your specific usage and we will say so. The position is simple: readers win, then we win.

Our correction policy is short. If a factual error is reported, we verify it, fix the page, and publish a dated correction note on the page itself within 72 hours. Material errors — anything that affected a price, a provider claim, or a regulatory citation — also appear in a public correction log so you can see our track record. The correction note names what was wrong, what is now right, and when it changed.

We do not silently edit. If a paragraph changes in a way that alters meaning, we flag the change. Version history is kept internally and is available on request for any public piece. Stealth-updating articles is how misinformation gets laundered — we refuse to play that game.

If a correction affects a plan recommendation a reader might have acted on, we do more than edit the page. We flag the change at the top, and if a provider's EFL changed materially after we recommended the plan, we say so plainly.

We do not sell reader data. We do not broker your name or email to a list of REPs who then call you at dinnertime. Enrollment happens between you and the provider you pick; our job ends when you have the information you need to choose. If you give us your zip code or your usage data for a comparison, that data is used for the comparison — not resold.

We do not use dark patterns. No fake scarcity timers. No "3 other people are looking at this plan right now" pop-ups. No pre-checked newsletter subscriptions. No countdown banners designed to bully you into enrollment. If a page you are on does any of those things, report it — it is a bug, not a feature.

We disclose our tracking. Analytics and product instrumentation tell us which pages work and which are broken. We publish what we collect, the vendors involved, and the controls you have to turn it off. Our privacy policy is the legal version of that disclosure; plain-language summaries live alongside it.

WCAG 2.1 AA is the floor, not the ceiling. Every page is designed and tested for keyboard navigation, screen-reader semantics, sufficient color contrast, resizable text, and meaningful focus states. We run automated accessibility checks in CI, and we audit with real assistive-technology users on a rolling basis. When we find a gap, it is tracked and fixed — not deferred.

We write for a plain-language reading level. Most editorial content targets a Flesch-Kincaid grade of 8-10 — readable, not dumbed down. Where a regulatory term matters (EFL, TDSP, kWh, demand charge), we define it the first time we use it on a page.

Spanish parity is a requirement, not a stretch goal. Nearly 30% of Texans speak Spanish at home. If a page exists in English, the Spanish version is a real translation — reviewed by a bilingual editor — not a machine-translated afterthought that reads like a LinkedIn auto-reply.

Spot an error or a misleading claim? Email editor@comparepower.com. That inbox is read by our editor-in-chief, not a support triage queue. We reply within two business days with either a confirmed correction, a request for more information, or a dated explanation of why we disagree with the claim. Silence is not an option; we have a service-level commitment internally that every editorial escalation gets a named response.

If the editor-in-chief does not satisfy your concern, escalation goes to the head of the company. Compare Power is small enough that the founder's inbox is reachable. Texans protecting Texans is not a slogan — it is the org chart. If a correction requires a regulatory filing (a PUCT complaint against a REP, for example), we will tell you which section of Chapter 25 applies and how to file.

We take structural feedback seriously. If you think a policy on this page is wrong, say so. We update this document when our practice changes. The version history is not buried; the page footer will tell you when this document was last reviewed, and why.