Skip to main content

Trust signals · Methodology

Methodology

How OBBBAtips calculates the qualified-tips and qualified-overtime deductions, where every citation comes from, how the site updates when the law moves, and how the decoder protects your inputs.

The pipeline

90d

Review cadence

30d

Update on new law

Math runs in your browser; every constant pins to 26 USC §224 / §225.

Personalized calculation

Compute

How we calculate

The TTOC decoder runs entirely in the browser. The math sits in a pure-function module sourced from src/components/decoder/decoder-math.ts. Every constant — the $25,000 tips cap, the $12,500 / $25,000 overtime caps, the $150,000 / $300,000 MAGI thresholds, the 10% phase-out reduction rate — is sourced from policy-state.json which in turn pins to 26 USC §224 / §225 on Cornell LII and the IRS final reg in the Federal Register.

The output card renders the math line-by-line. The estimated tax savings is shown as a band (12% low to 32% high marginal rate) and is flagged “estimate, not measurement” in the surface text — your real refund depends on your full tax picture.

Statute-cited expansion

Citation pipeline, update cadence, and AI Overview defense

Citation

How we cite

Every policy claim on this site has a claim_id that resolves to an entry in src/data/policy-state.json. Each entry carries a plain-English claim, the statute or regulation citation, the primary URL, the last-verified date, and the reviewer-signoff status. A pre-build CI script (policy-state-check) blocks any deploy that references a claim_id missing from this file.

Inline citations render as small superscripts that link to /sources; the policy-state pin under each verdict shows the last-verified date and the source URL. Primary sources are IRS.gov, Treasury.gov, federalregister.gov, congress.gov, and law.cornell.edu — never an aggregator-blog substitute.

Cadence

How we update — quarterly review (90-day cadence)

The site runs a 90-day review cron. Every claim_id older than 90 days is flagged for re-verification against its primary URL. A quarterly cadence means an in-flight final regulation, an IRS notice update, or a state DOR publication does not sit on the site for more than three months without a fresh last-verified stamp.

When Congress acts (extension, modification, sunset), this site updates within 30 days and re-pins all affected claim_ids.

AI Overview

How we built around AI Overview

Before authoring a single page, the build ran a 30-query SERP audit covering topline policy, form-mechanics, per-occupation eligibility, statute-threshold, procedural / amend, and tooling / brand classes. The pre-build audit measured 43.3% of representative OBBBA-tip search queries (13 of 30) as HIGH-cannibalization risk for Google's AI Overview — above the 40% pass threshold that was set as the heuristic gate.

Rather than fight AI Overview on the queries it owns (topline policy and boolean “is X a tipped occupation” informational lookups, both of which IRS and Big-Four sources already dominate), the build was structured around five surfaces AI Overview structurally cannot render:

  1. A 3-tier per-occupation structure. Tier 1 (5 deep, ~1,500–2,200-word pages on SSTB-edge or contested-organic occupations: dancers, sports-and-recreation instructors, massage therapists, hairdressers, taxi / rideshare drivers). Tier 2 (18 medium, per-trade pages on thin-organic occupations where top-3 organic is patchy). Tier 3 (48 long-tail navigational pages with the decoder pre-filled). The depth tier defends the highest-value occupations from AI Overview compression.
  2. A personalized-math wedge. The decoder accepts the worker's tax year, filing status, base wages, tip wages, qualified overtime, and MAGI and renders the §224 and §225 deduction lines with the phase-out math shown step-by-step. AI Overview cannot substitute a paragraph for an interactive calculator that honors the user's actual numbers.
  3. A 50-state conformity matrix. AI Overview can answer a single-state conformity question but cannot render 50 states comparatively. The matrix at /state-conformity is the highest-ROI wedge identified in the audit.
  4. An employer-side Box 14b validator. The validator at /box-14b-validator cross-checks a three-digit code against the canonical 71-trade TTOC list and emits a plain-English W-2c request template. That's a tool, not a definition page; AI Overview doesn't render tools.
  5. An amend-prior-year decision tree. The walkthrough at /amend-prior-year covers the §6511(a) window math, the “was your TY2025 transition method correct or did you miss it” decision tree, and the e-file vs paper-file tradeoffs. Procedural walkthrough is a category AI Overview compresses badly.

The audit is reproduced in full at .ops/research/aio-audit.md. The 43.3% finding is documented honestly rather than framed away — the wedge this site occupies is narrower than an open-greenfield assumption would have suggested, and the build doctrine is shaped accordingly.

Actionable artifact

Reviewer policy and decoder privacy

Reviewers

Reviewer attribution

Tax content on this site is YMYL (Your Money, Your Life). Every reviewer attribution this site publishes is a real human with a real CPA license (state + license number) or a real IRS Enrolled Agent enrollment number. The reviewer's LinkedIn or firm page is linked; their last-reviewed date is stamped on every page they reviewed.

Until the reviewer is secured, the attribution block on every page renders the placeholder state honestly — “Reviewer pending.” This site never publishes a synthetic reviewer.

Privacy

The decoder doesn't log your inputs

Your occupation, base wages, tip wages, overtime wages, MAGI, and filing status are processed in your browser. They are never sent to a server, never written to localStorage, never posted to analytics. The only thing the analytics layer ever sees is an aggregated event — a verdict bucket, an industry bucket, a tax year — never tied to identifying inputs and never tied to a session that could be re-identified.

Full details on /privacy.