Post copy vs published endorsement-guide factors for creators and brand teams
Methodology: disclosure patterns · deterministic placement math · human review
Paste your sponsored, review or affiliate post before you publish it. Get a scored pre-check report — is the material connection disclosed at all, is the wording clear enough to count, does the disclosure land above the platform fold, does an affiliate link go unexplained, and does the copy match the connection you declare — based on the FTC Endorsement Guides (16 CFR Part 255). A heads-up, not a legal opinion.
1. What the engine owns
Severity-ranked findings from pasted post copy against word lists and placement heuristics derived from the FTC Endorsement Guides (16 CFR Part 255 — disclosure of material connections, clear-and-conspicuous standard, endorsements by employees and influencers). Five check families: disclosure-presence, which asks whether any unambiguous connection-disclosing wording or hashtag (“paid partnership”, “#ad”, “I earn a commission”) appears at all, or only vague relationship labels the guides themselves call too vague (“ambassador”, “partner”, “collab”); disclosure-clarity, which flags a bare mid-sentence “ad” or decoder-jargon slang; placement, which locates the first clear disclosure token and, on truncated feeds (Instagram captions fold at roughly the first ~125 characters; TikTok and Shorts fold early too), flags a disclosure that sits below the fold; material-connection, which flags affiliate links and “link in bio” CTAs with no commission sentence and copy that understates the connection you declared; and claims-context advisories — quantified results, health and income wording, and platform paid-partnership toggles — that never dock the score. Scores dock 20/6/2 per HIGH/MEDIUM/LOW finding (advisories never dock); the rule snapshot is dated by the engine version — it is not synced with regulator or platform pages in real time. When no disclosure token can be located, placement math is skipped and the report says so — nothing is guessed. Demo is labelled Rule-based — never silent mock.
2. What optional AI may do
Explain or polish findings when a key is configured. Upstream failure returns 502. No invented findings, no quotes from copy you did not paste, no disclosure position the parser did not actually find.
3. What you must do
- Review findings against your own contracts and affiliate relationships — only you know whether a material connection exists and what the real deal was.
- Apply fixes in your own tooling — EndorsePrecheck does not connect to accounts, brand deals or ad platforms and never publishes anything.
- Treat output as decision-support, not an endorsement-disclosure certification or legal advice.
4. What we deliberately do not do
- Material-relationship verification — whether a claimed payment, gifted product or commission is real is a records question outside the pasted text; it appears only as a labelled advisory, never as a lookup result.
- Account, brand-contract or affiliate-dashboard connections, platform-label inspection, or continuous monitoring.
- Compliance probabilities or enforcement-prediction numbers — outside the regulator nobody can compute what it will decide, and pretending otherwise is the check becoming the pattern it flags.
- Legal review — the Part 255 checks read disclosure patterns and placement, not your full obligations. The rule set cites a public mirror of 16 CFR Part 255 that returned HTTP 200 with endorsement content from this build environment; the eCFR canonical page and the FTC summary redirected to a bot-unblock interstitial or returned 403 at build time, so those exclusions are documented in the rule-sources log (lib/rules/SOURCES.md). Non-US regimes (EU/UK UCPD, ASA) are advisory context only, and regulated campaigns need a qualified professional.
Heuristic pre-check — not an endorsement-disclosure certification or legal advice. No material-relationship verification, account connection, platform-policy measurement, or enforcement prediction performed.