A review only earns its place on the calendar when it changes something: it renews a claim's evidence, retires an assumption, or closes a decision. This essay sets out a cadence built on three sources — advertising substantiation guidance, AI risk management documentation, and search content-quality guidance — and points to the private decision brief worksheet where you can hold the structure.

Why a cadence is a decision tool, not a ritual

A marketing review can go wrong in two directions. Either they are so infrequent that assumptions harden into accepted fact, or so frequent that the team re-litigates settled questions and stops producing anything to review. The useful middle is a cadence whose timing is set by the claims and constraints you are managing, not by habit or by the calendar quarter alone.

A review cadence is really an evidence-refresh schedule. Each checkpoint should ask which claims in the current plan are still supported, which have never been supported, and which decisions remain open. The public sources here do not prescribe an interval — no government or platform document tells you to meet monthly or quarterly. What they do is specify the conditions under which a claim, a risk treatment, or a piece of content can responsibly stand. A cadence derived from those conditions is a considered inference, not a rule.

The substantiation clock: review before, not after

The FTC's guidance on advertising substantiation is unusually blunt about timing. It states that under the Federal Trade Commission Act advertising must be truthful and non-deceptive, advertisers must have evidence to back up their claims, and advertisements cannot be unfair (FTC, Advertising FAQ's: A Guide for Small Business). More pointedly, the guidance says that before a company runs an ad it has to have a "reasonable basis" for its claims — objective evidence supporting the claim — and that the law requires advertisers to have proof before the ad runs (FTC).

That single word, before, is the reason a review cadence must sit upstream of publication and stay there. It also sets which claims deserve the earliest checkpoint. The FTC says the required kind of evidence depends on the claim; health or safety claims in most cases must be supported by competent and reliable scientific evidence, and statements from satisfied customers usually are not sufficient to support a health or safety claim or any other claim that requires objective evaluation (FTC). It adds that a money-back guarantee is not a substitute for substantiation.

A practical implication for a review schedule: rank each proposed initiative by how much objective evidence its claims require, and set the earliest review for the claims that are hardest to substantiate. The FTC also notes it looks at both express and implied claims, and that ad staff cannot clear ads in advance (FTC). That last point matters for cadence design: because no agency will pre-approve your campaign, the review has to be internal, documented, and repeatable. The worksheet records the evidence column precisely so that this internal check has something to stand on.

Reviewing the AI-assisted parts on their own schedule

When a plan involves AI-assisted research, drafting, or tooling, the review needs a second track. NIST describes its AI Risk Management Framework (AI RMF 1.0) as intended for voluntary use and to improve the ability to incorporate trustworthiness considerations into the design, development, use, and evaluation of AI products, services, and systems (NIST, AI Risk Management Framework). The same page records that the framework was released on January 26, 2023 and is being revised as part of the White House AI Action Plan — a status note that itself argues for treating AI governance notes as time-stamped rather than permanent.

NIST's Generative AI Profile, released July 26, 2024, is described as able to help organizations identify unique risks posed by generative AI and propose actions for generative AI risk management that align with their goals and priorities (NIST). The NIST AI Resource Center's technical reports index adds that the profile was produced with input from a public working group of 2,500 participants and centers on 13 risks and more than 400 actions developers can take to manage them (NIST AIRC, Technical Reports).

Those figures are the program's own description of a voluntary resource, not a compliance scorecard, and nothing here certifies any tool or method. But they justify a design choice in the cadence: give AI-assisted components their own review slot, separate from the marketing-claims review, so that model use, data handling, and output-checking assumptions are examined rather than absorbed into the general discussion. The NIST AIRC page also notes that current approaches to evaluation of AI often do not account for risks and impacts of AI systems in the real world, and describes its ARIA program pairing people with AI applications to study impacts in scenario-based interactions (NIST AIRC). If real-world evaluation is genuinely hard, a review that waits for certainty will never convene — so schedule it, record the unknowns, and revisit.

Reviewing the published material against content-quality guidance

The third track concerns what gets published. Google states its automated ranking systems are designed to prioritize helpful, reliable information created to benefit people, and not content created to manipulate search engine rankings (Google Search Central, Creating helpful, reliable, people-first content). The same documentation states that E-E-A-T itself is not a specific ranking factor, that trust is most important among its aspects, and that systems give more weight to strong E-E-A-T on topics that could significantly affect health, financial stability, or safety — YMYL topics (Google).

Two items from that page belong in a recurring checklist. First, Google warns that fabricating creator profiles — AI-generated headshots, made-up names, or false credentials to make content appear written by human experts — is a form of deception and a signal of a low-quality page (Google). Second, it states that using automation, including AI generation, to produce content for the primary purpose of manipulating search rankings violates its spam policies, and that AI or automation disclosures are useful where someone might reasonably ask how content was created (Google). It also notes quality raters have no control over ranking and that rater data is not used directly in ranking algorithms (Google) — so a review should never be framed as an attempt to influence raters.

Google further notes that interactive features such as calculators or online tools can constitute main content when they directly help the page achieve its purpose, and that raters are trained to evaluate effort, originality, talent or skill, and accuracy (Google). For a planning instrument, that is a useful reminder that the tool itself, not only the essay around it, is subject to review.

A workable cadence

The following is our recommended structure, inferred from the three bodies of guidance above; it is a method, not a standard, and no source endorses this specific schedule.

  • Before publication. Confirm that every express and implied claim has evidence recorded, consistent with the FTC's before-the-ad requirement. Unsubstantiated claims either gain evidence or leave the plan. Note the unknowns explicitly.
  • At each AI-component change. Whenever model use, data inputs, or output-checking changes materially, re-review that component against the NIST generative-AI risk framing and record what remains unevaluated.
  • On a fixed content-quality check. Review authorship honesty, automation disclosures, and the accuracy and effort of the tool itself, using Google's questions as prompts, not as a scoring rubric.
  • At each decision point. Ask the decision question again. If the evidence has not changed, close the review early rather than reopening settled questions.

To practice this structure with your own initiatives, open the decision brief worksheet. To see the underlying reasoning, read how to start marketing planning with a decision, how to keep evidence separate from assumptions, and how to compare initiatives without fake precision.

What each review should produce

A review that ends without a written trace invites the same discussion next time. Each checkpoint should leave a short record: the decision question, the evidence now held (with its source), the assumptions still open, the owner and time constraint, and the claims that must not be published until evidence arrives. The worksheet exports exactly this as private Markdown precisely because the record, not the meeting, is the durable asset.

Two cautions belong in the record itself. First, the substantiation standards are US federal guidance tied to FTC jurisdiction and do not address this site or your campaign; consult qualified advice for your own obligations. Second, NIST's framework is voluntary and under revision, and Google's documentation describes Google's own systems without promising any ranking outcome. A cadence built on these sources is a discipline for honest planning — it is not a guarantee of legal compliance, model safety, or search performance, and this publication makes no such promise.