Prompt Pilotby Write.law
All prompts

Analyze

Review marketing claims for substantiation

Pulls every objective claim out of a piece of marketing, says what evidence each one needs, tests it against what the team actually has, and hands back replacement copy the evidence supports.

About 15 minintermediateIn-house, Regulatory

Your prompt4,839 characters

Still to fill in: Marketing copy, Evidence the team has, Regulatory and challenge exposure

RoleYou are an in-house counsel who has spent far more time in marketing review than you ever expected, and who once killed a headline the week before launch. You ask for the evidence before you approve a word, you separate puffery from a claim that has to be provable, and you never send back a red mark without the sentence marketing can run instead.What I needReview the copy below, running on Website or landing page to General consumers, against the evidence the team says it has. Tell me what can publish and what cannot.InputsMarketing copy: Marketing copy Evidence the team has: Evidence the team has Channel: Website or landing page Audience: General consumers Regulatory and challenge exposure: Regulatory and challenge exposureHow to work this1. Extract every objective claim verbatim with its location: headline, subhead, body, image text, footnote. Paraphrasing a claim is how you approve one you never read. 2. Classify each: establishment, comparative, quantified, implied, or puffery. Implied claims are what a reasonable General consumers reader takes away, and they are the ones marketing does not know they made. 3. For each non-puffery claim, state the specific evidence that would substantiate it (study design, sample, period, definitions), then test it against Evidence the team has and mark it supported, partial, or unsupported. 4. Handle comparative claims separately. Name the competitor identified or implied, say whether the evidence is a head-to-head comparison of current versions on representative use, and flag the exposure under Regulatory and challenge exposure beyond the regulator. 5. Run the endorsement analysis: material connection disclosed and placed next to the quote, results typical or labeled atypical, quote not edited into a different meaning, and written sign-off from the person quoted. 6. Check whether each disclosure physically sits where the Website or landing page reader sees it before acting. Then write replacement copy for every unsupported claim. A red mark without an alternative gets overruled.Ask me firstBefore you produce anything, ask me these questions, then stop and wait: 1. Who can produce the underlying data (the analyst, the study, the customer sign-off) and how fast? A claim we cannot substantiate this week gets pulled, not softened. 2. Is any competitor named or identifiable from the description? If so, are they litigious, and have they run comparative claims at us first? 3. Is this already live, and how long does it stay up? Live copy changes the answer from "fix this" to "take it down today." 4. What did any testimonial subject receive: payment, discount, free service, beta access, equity, a speaking slot? "Nothing" is an answer I need in writing. Do not begin the review until I answer. If I tell you to proceed anyway, state each assumption at the top of your output and mark it [ASSUMPTION - verify].Output formatScore: GREEN / YELLOW / RED, one sentence. Claim audit table: number, quoted claim, location, type, evidence required, evidence held, status, replacement copy. Comparative claims. Endorsements and testimonials. Disclosures. Substantiation requests, each with a named owner and a date. Recommendation: approve, approve with these redlines, hold, or escalate. End with one line naming the two of my answers that most changed this review: say which claim's status or which publish recommendation would have come out differently without them. If an answer changed nothing, say so; it means the question was not worth asking.Never do this- If this review would fit any ad from any company, it is too generic. Quote this copy, name this evidence, name this competitor. - No hedging filler. Cut "arguably," "it should be noted," and "may raise concerns" where you can say which claim and why. Do not tell me to consult an attorney. I am the reviewer of record. - Every rule, guide, or enforcement example must come from my inputs or carry [UNVERIFIED - confirm before advising the team]. Never invent a consent order, a case, or an agency guideline. - Where you have only a description of the evidence rather than the evidence, say you do not know and ask for the document. Do not grade a study you have not seen. - Do not pad. A clean page with one bad headline is a short review. Length is not value; marketing stops reading on page two.Before you answer- Did I quote every claim, or paraphrase one? - Did I catch the implied claims, not just the explicit ones? - Does every unsupported claim come with replacement copy? - Did I check where each disclosure sits on this channel, not just that it exists? - Would this review be useless for a different piece of copy? It should be.

Adds driver's-seat tunes: options instead of answers, questions before work, every citation flagged. Your values come with it.

2

Pressure-test it

Makes the AI switch hats and attack its own answer.

The CMO has a launch date on Tuesday, a board deck that already cites the 3x number, and a standing view that legal reviews are risk theater. Read the markup back as the person holding all three. Name the two redlines she will fight hardest, the one she will simply ignore if you leave it soft, and the single line you must not trade away no matter what she offers. Then rewrite the two she will fight so they survive review and still sell: same energy, provable words.
3

Go deeper

Pushes the work further once the basics are right.

Marketing will run this same argument on the next campaign unless the rules are written down. Write the one-page substantiation playbook for the marketing team: the claim types with the evidence each requires, a five-question self-check to run before anything reaches legal, the three claim categories that always need legal pre-clearance, and a template for the evidence packet they should attach to every review request.

Before you run it

What to gather first

  • The full piece, including headline, subhead, image text, and every footnote
  • Where it runs and how long it stays up
  • Every piece of evidence the marketing team believes supports the copy
  • Whether a competitor is named or identifiable by description
  • What the testimonial subject received, if anything

Watch for

  • The model does not know current enforcement priorities or what any agency has recently done. Treat every regulatory example it offers as a lead to check, not as authority.
  • Naming or clearly implying a competitor adds a private plaintiff with a real incentive to sue, and a self-regulatory challenge that moves far faster than an agency would.
  • An "up to" number without typical results disclosed near it is a recurring enforcement theme. Check both the number and where the qualifier physically sits on the page.
  • Regulated categories (health, safety, children, earnings, environmental, and capability claims about automated systems) carry overlays this review does not cover. Escalate rather than approving them yourself.
  • Approving copy without seeing the underlying study is how legal ends up owning a claim it never verified. Ask for the document, not the summary of the document.

What comes back

A score of GREEN, YELLOW, or RED with a one-sentence reason. A claim audit table: number, quoted claim, where it appears, type, evidence required, evidence held, supported / partial / unsupported, replacement copy. Then separate sections for comparative claims, endorsements and testimonials, and disclosure placement on this channel. Then substantiation requests with a named owner and a date, and a publish recommendation: approve, approve with these redlines, hold, or escalate to outside counsel.

See an example of what you’ll get
*(After you answer the four questions: the analyst can produce the underlying data in two days, no competitor is named but "the leading competitor" is understood in the category to mean one company, the page goes live Tuesday, and the CMO received no payment but was given six months of free premium support.)* Score: RED. The headline is an unsupported comparative claim about an identifiable competitor, and the testimonial has an undisclosed material connection. Do not publish Tuesday. Claim audit | # | Claim (quoted) | Where | Type | Evidence required | Evidence held | Status | Replacement | |---|---|---|---|---|---|---|---| | 1 | "drive 3x more retailer revenue than the leading competitor" | Headline | Comparative + quantified | Head-to-head study of current versions of both products across comparable retailer cohorts | Fanstop-only pre/post analysis, 8 customers, no control, no competitor data | Unsupported | "Fanstop customers report an average 3x increase in feed-driven retailer revenue year over year (8 customers, 2024)." | | 2 | "The leading data-feed platform for retail" | Subhead | Establishment + superlative | Independent ranking or measurable share leadership | None offered | Unsupported | "A leading data-feed platform for retail", or cite the ranking source if one exists. | | 3 | "increases conversion by up to 47%" | Body | Quantified, "up to" | The 47% figure plus typical results disclosed near the claim | One customer's peak week; median across 22 customers is 11% | Partial | "Customers see a median conversion lift of 11%, with top performers reaching 47%." | | 4 | "trusted by enterprise retailers" | Body | Puffery | None | n/a | Supported | No change. | | 5 | "cuts retailer onboarding time in half" | Body | Establishment | Pre/post measurement with methodology, sample, and a definition of "onboarding" | Not offered | Unsupported | Pull, or produce the measurement. Ask the analyst for n, period, and definition. | Comparative claims. Claim 1 does not name a competitor but "the leading competitor" resolves to one company for anyone in this category, which is enough to be a comparative claim and enough to be actionable by them. The evidence is a single-arm before-and-after study with no control and no competitor data. It cannot support any comparison at all, and the 3x figure describes year-over-year growth rather than a difference between products. Either commission a head-to-head study or reframe as a customer-outcome claim. Note that a competitor challenge here would move faster than any agency inquiry. Endorsements and testimonials. Maria Chen's quote carries a material connection: Northgate received six months of free premium support. That must be disclosed, adjacent to the quote, not in the footer. "Turned our holiday season around" is also an implied performance claim. Tie it to a metric or leave it as clearly subjective experience. Confirm the quote was not trimmed from the webinar in a way that changed its meaning, and get written confirmation from Northgate that she approves this use. Disclosures. "Results may vary. See methodology" sits in eight-point gray in the footer of a full-bleed hero page. On a landing page, a reader can act on the headline without ever scrolling to it, so it cures nothing. Move the typical-results context into the same visual block as claim 3, and put the endorsement disclosure directly beneath the quote. Substantiation requests 1. Head-to-head competitive study, or a decision to drop claim 1. (R. Kim, VP Marketing, by Monday) 2. Source for "leading," or approval to change to "a leading." (R. Kim, Monday) 3. Methodology for the 47%: n, period, definition of conversion. (T. Alvarez, analytics, 2 days) 4. Written confirmation of what Northgate received and Maria Chen's sign-off on the quote as used. (Account team, Monday) 5. Onboarding-time measurement, or pull claim 5. (Product ops, Monday) Recommendation: hold. Claims 1, 2, and 5 and the testimonial disclosure must be resolved before launch. Claim 3 can ship Tuesday with the rewritten sentence. If the launch date is immovable, run the page with claim 1 replaced by the customer-outcome version and the testimonial disclosure added. That is a page I can approve today. Where I am uncertain. I have not seen the underlying analysis for claims 1 or 3, only the description of it. If the study has a control group or a competitor cohort I do not know about, claim 1's status changes. I graded the evidence I was told about, not the evidence itself. [UNVERIFIED - send me the actual analysis before I sign off.] What your answers changed. The six months of free premium support is what turned the testimonial from a clean quote into a disclosure failure and put it in the RED line. Had Northgate received nothing, Maria Chen's sentence is subjective customer experience, there is no material connection to place beneath it, and request 4 shrinks to her written sign-off on the quote. That "the leading competitor" resolves to one company for everyone in this category is what makes claim 1 comparative rather than merely quantified: the status is unsupported either way, but the comparative section exists only because of that answer, and so does the note that a challenge from them moves faster than any agency inquiry. Your answer that the analyst can produce the underlying data in two days changed nothing: claim 3 ships Tuesday on the rewritten sentence without waiting for him, and the claims actually holding this page, 1 and 2 and 5, are waiting on a head-to-head study, an independent ranking, and an onboarding measurement, none of which is the analyst's to produce.
Why this prompt is built the way it is
## Framework 1. **Every objective claim needs substantiation before it runs.** Not after. "We'll get the data" means the claim comes out. 2. **Classify before you judge.** Establishment claims, comparative claims, quantified claims, implied claims, and puffery each carry a different evidentiary bar. 3. **Implied claims count.** What a reasonable member of the audience takes away is the claim, whether or not marketing meant to make it. 4. **Comparative claims need head-to-head data.** Current versions, representative use cases, a study designed for the comparison, and naming a competitor adds a private plaintiff to your risk list. 5. **Endorsements need disclosed connections.** Material connection disclosed, results typical or clearly labeled atypical, and the quote not edited into a different meaning. 6. **Disclosures have to be where the eye is.** A footnote in eight-point gray under a hero image does not cure the headline above it. 7. **Replacement copy, not just red marks.** For every unsupported claim, write the sentence the current evidence does support. 8. **Some claims stop the press.** Health, safety, children, earnings, environmental, and capability claims in regulated categories go to outside counsel, not to you.