Methodology & Data Provenance

Every number on this site — a tipped cash minimum, a square-footage cap, a salary floor — is either traced to an authoritative source and dated, or it is not published at all. This page documents the rules we follow: where figures may come from, how the tools distinguish verified data from estimates, what the review dates mean, and the specific gaps we have chosen to disclose rather than model. If you ever need to know why a calculator refused to compute something, the answer is here.

The sourcing hierarchy#

Wage-and-hour figures decay. Rates index annually, ordinances are amended mid-cycle, and secondary write-ups quietly fall out of date while still ranking well in search. We therefore source in a strict order, and a figure's place in that order determines how much weight it can carry:

1. Operative statute, regulation, or ordinance text. The codified law itself — a Labor Code section, a CFR part, a municipal code chapter — is the only source that can establish a rule: a cap, a formula, a premium multiplier, a consent requirement. Where a city publishes its code through a hosting service (American Legal, Municode, eCode360), we cite the hosted section and treat it as primary.
2. Official agency publications. Department of Labor fact sheets and opinion letters, state labor department rate notices, city wage-office posters and bulletins. These are authoritative for current dollar figures — especially CPI-adjusted rates that the statute states only as a formula — and for an agency's enforcement position. When an agency notice and our reading of the statute diverge, we publish the agency figure and note the divergence.
3. Reputable secondary alerts — management-side firm bulletins, legal trade press, established local journalism. These are used only as pointers: they tell us that something changed and where to look. A secondary alert is never, by itself, the source of a published figure. If a firm bulletin reports a new rate and we cannot confirm it against the ordinance text or an agency publication, the figure stays out and the change is described qualitatively until it can be confirmed.

Each data record carries its own source list with access dates, and those sources are reproduced on the pages built from the record — every state guide ends with the official sources its figures came from, and the hotel ordinance brief does the same per ordinance.

Verified vs. estimated: a taxonomy the tools enforce#

The distinction between "we verified this" and "we are projecting this" is not a footnote here — it is structural. The site's data layer encodes verification status in machine-readable form, and the calculators change their behavior based on it. The concrete idioms:

The operating rule When a parameter cannot be verified from the first two tiers of the hierarchy, the tools disclose instead of compute. A wrong number presented confidently is worse than a stated gap: operators staff and budget against these figures, and an unverified cap or premium multiplier could induce a real-world violation. We accept the cost — occasionally a tool tells you less than a marketing page would — in exchange for never silently approximating.

Review cadence and what the dates mean#

Two date fields appear throughout the site, and they make different promises:

• lastReviewed (shown on every page as "Last reviewed" or "Verified as of") is the date a human last checked the page's or record's figures against the sources cited. It is not the date the law changed and not an auto-updated timestamp — a page rebuilt today with unreviewed data keeps its old date. Pages whose review date falls more than 120 days behind the newest content in the corpus are flagged internally as stale at every build.
• accessed (shown beside each source link) is the date the cited source was retrieved. A figure is only as fresh as its access date; if an agency reissues a notice after our access date, the site may lag until the next review.

Scheduled rate steps are recorded as dated entries — a rate and the date it takes effect — taken from enacted schedules, never from proposals. The build machinery validates every schedule (steps must be strictly date-ascending and may never fall below the step before them) and assembles the compliance calendar automatically from the same records, so the calendar cannot drift from the data. Where a statute indexes a rate instead of stating a dollar figure, the step is recorded as "CPI-adjusted" until the agency announces the number.

Mid-week steps get specific handling: wage floors change on dates like January 1 and July 1, which rarely land on a workweek boundary. The workweek auditor resolves the applicable wage floor per day, so a July 1 ordinance increase that lands on a Wednesday is applied to Wednesday through Sunday at the new rate and Monday through Tuesday at the old one — the same way an auditor would reconstruct the week.

What the tools deliberately do not compute#

The calculators are educational instruments that show their arithmetic, with citations, so a controller or counsel can check every step. They are not a payroll system, a damages model, or legal advice. By design they do not:

• Render legal conclusions. The exemption checker produces a structured issue-spotting summary, not a classification; the exposure modeler labels its own output an educational estimate, not a damages calculation.
• Compute rules whose parameters are unverified — the workloadRules: null and computed: false behavior described above.
• Model every state quirk. Where a rule is simplified (for example, an earnings-dependent daily-overtime trigger), the tool says so in its on-page notes rather than implying completeness.
• Substitute for the source documents. Statutes, regulations, and ordinances summarized here are simplified for orientation; the source documents control. See the full disclaimer.

Known gaps we have chosen to disclose rather than model#

The policy: when a law exists but a computational parameter is unverified, we publish the law's existence, its citation, and a qualitative summary — and we list the gap here, so the absence of a number is never mistaken for the absence of an obligation. The gaps currently known to us:

• The three California ordinances promoted on secondary sources (LA County LACC ch. 8.21, Glendale GMC §5.120.060, Long Beach Measure WW). These were previously disclosed-not-computed. As of June 2026 their parameters — tiers, special-room models, proration, and the all-hours double-pay premium — are computed, each structurally identical to an ordinance already modeled (LA City, Santa Monica/West Hollywood, Oakland) and corroborated across multiple independent legal analyses citing the adopted text. They remain on this list because the codified municipal-code text could not be retrieved directly at the review date (the repositories returned errors): the computation rests on those secondary legal sources pending raw-code confirmation, and research/penalty-regimes-notes.md records the provenance and confidence.
• Kentucky's seventh-day overtime (KRS 337.050). Time-and-a-half is owed for the seventh day when an employee works all seven days in a workweek and is permitted to exceed 40 hours. Because that premium overlaps the FLSA weekly overtime the workweek auditor already computes and turns on the 40-hour proviso, the tool flags it for review rather than pricing the same hours twice.
• The fluctuating workweek and the §7(i)/§7(g)(2) elections. The fluctuating-workweek method (29 CFR 778.114, rejected outright in CA, PA, AK, and NM) and the §7(i) commissioned-employee and §7(g)(2) rate-in-effect elections are documented but not computed in this release; the auditor prices straight-and-overtime pay under the default weighted-average method.
• Washington's Demetrio rule for non-agricultural piece-rate workers. The housekeeper auditor computes the separate rest-break pay Demetrio v. Sakuma requires, but no Washington appellate decision has extended that holding from agriculture to non-agricultural piece-rate workers such as hotel housekeepers (WAC 296-126-092); the tool computes it on the identical regulatory language while labeling it a litigation-risk position, not settled law.
• New York uniform-maintenance pay. 12 NYCRR 146-1.7 requires weekly uniform-maintenance pay (three hour-band tiers with regional dollar amounts) when the employer does not launder required uniforms. The rule is real and flagged (computed: false); the current dollar amounts live in the NYSDOL Part 146 summary and are not computed by these tools.
• Hawaii and Washington credit-card-fee silence. Neither state's law (nor controlling case law we have verified) squarely answers whether a pro-rata card-processing offset against tips is permitted, so their tip-pool records carry no ccFee rule at all — the tip-pool architect stays silent rather than importing the federal answer into a state that may reject it.
• New York meal-period remedies. Labor Law §162 violations carry NYSDOL enforcement exposure, not a formulaic California-style premium hour, so the engine records no premium formula for New York meal periods and the tools compute none.
• Washington's missed-meal premium theory. The auditor computes the verified remedy for a missed Washington rest break — the break time itself paid as wages — but the broader theory that a missed 30-minute meal owes 30 minutes of penalty pay on top of time worked (Androckitis v. Virginia Mason) remains contested, so it is disclosed in the rule's notes rather than computed.
• Penalty multipliers on tip theories in Washington, New Jersey, and D.C. The exposure modeler computes back pay on tip-misappropriation claims everywhere, but applies those states' double/treble multipliers only to the wage theories their statutes squarely cover — whether each statute's damages reach a gratuity claim is not pinned by primary text, so the multiplier is conservatively withheld.
• West Hollywood's hotel-rate indexing cap. West Hollywood's citywide minimum wage carries a verified 1–4% annual adjustment band, but we could not confirm that the band governs the separate hotel-worker rate, so the forecaster applies no cap when projecting that rate.
• The §7(i) retail-or-service-establishment status is attested, not computed. The exemption checker tests the two numeric §7(i) prongs (the regular-rate floor and the more-than-half-commissions share) but takes the threshold establishment question as a user attestation — whether a hotel spa qualifies as a distinct retail-or-service establishment is a fact-intensive judgment the regulation does not reduce to a formula.
• Multi-week bonus allocation uses the §778.209(b) equal-weekly method. The regular-rate calculator's bonus true-up apportions a deferred bonus equally across the workweeks it covers — the regulation's method when the bonus cannot be tied to particular weeks. Where it can be tied to specific weeks, the law requires allocating to those weeks instead, and the tool says so.
• Nevada's rolling-24-hour daily overtime is simplified in the tipped calculator. The tipped overtime calculator applies Nevada's earnings gate (daily overtime only below 1.5× the minimum wage) but treats the daily trigger on a calendar-day basis; the workweek auditor models the true rolling-24-hour window from shift times.

This list is maintained alongside the data itself and is expected to change: when a gap is verified, the parameter moves into the computed layer and comes off this list; when a new law arrives faster than its codified text, it goes on. If you can point us at primary text that closes one of these gaps, the about page explains how to reach us.

Numbers the site computes for itself#

Worked examples in the issue briefs and tool pages are not typed by hand from a calculator app — they are produced by running the same calculation engines that power the tools, so the prose and the tools can never disagree. If an engine's method changes, its examples are regenerated. Where an example depends on a current rate (California's $16.90 minimum wage, the federal $684 salary floor), the rate is drawn from the same verified data records described above.

Verifying a figure yourself#

Every published number is meant to be checkable. The replication recipe:

1. Find the record. A rate or rule lives in a specific data record — a state guide's sources list, an ordinance entry, an engine rule. Each page ends with the official sources its figures came from, so start there.
2. Read the source and its access date. Every source carries an accessed date; a figure is only as current as that date. If the agency reissued a notice after it, the site may lag until the next review.
3. Go to the operative text. For a rule — a cap, a premium multiplier, a consent requirement — confirm against the statute, regulation, or ordinance itself (the authorities map says where each lives); for a current CPI-indexed dollar figure, confirm against the agency rate notice.
4. If text and figure diverge, the better source controls. The site publishes the agency figure and notes any divergence from a literal reading of the statute; if you can show a published figure is stale or wrong against primary text, that text wins — and the about page explains how to reach us.

Citation conventions and corrections#

Citations follow a consistent shorthand, decoded in full on the authorities map: statutes as title-code-section (29 U.S.C. §207), regulations as title-CFR-part.section (29 CFR 531.59(b)), and cases as parties, volume, reporter, page, year. The site cites to the most authoritative layer that fixes a given figure — the statute or regulation for a rule, the agency notice for a current indexed rate — and never treats a secondary alert as the source of a number. New legal propositions are added only when traceable to primary text or official guidance; where a point cannot be verified, it is stated at the level of generality the evidence supports rather than dressed in an invented citation. Corrections are welcome and routed through the about page; when one lands, the affected record's lastReviewed date advances and, if the change is material, it appears in the developments feed.