Influencer campaign reviews: lessons from 12 FTC orders

The FTC’s current Endorsement Guides explain how material connections and advertising claims should be handled. Its staff FAQ also discusses training, monitoring and intermediaries. The Guides were revised in 2023, after all the orders in our selected set. Historical orders show particular arrangements imposed on their respondents; they do not become a universal checklist simply because several contain similar terms.^[2]

We coded a combination of six expressly stated elements: disclosure; instructions directed to endorsers; signed acknowledgment and agreement; a monitoring system; termination or exclusion; and campaign-specific records. Eight of the eleven campaign families contain that combination. We call it a bundle for counting purposes, not a certification that a program is sufficient.^[1]

The figure separates several other questions that an “approved” label can hide. A review schedule is different from a payment condition. A creator’s opportunity to correct an inadvertent failure is different from oversight of an agency. The rows overlap because one order can address several of these matters.

In Deutsch LA, Lunada and Genesis Today, we did not identify the separate campaign-monitoring system used in that bundle. That finding concerns the reviewed main-order text. It does not mean those orders imposed only disclosure duties or that the respondents had no other responsibilities. Deutsch, for example, also contains a removal provision.^[8]

A monitoring obligation can identify the work without prescribing a calendar. Five of the eight families with a monitoring system do that in the reviewed text. Three—Legacy Learning, Machinima and Teami—provide more specific timing. Their arrangements are different enough that one generic “monthly check” would misdescribe them.

Legacy Learning’s 2011 order addresses affiliates. It requires monthly checks of the websites of the top 50 affiliates by revenue and 50 selected at random from the remainder. It also specifies how the top group is established and refreshed. This combines a revenue-based selection with a random selection; it is not a direction to review only the largest affiliates.^[3]

Machinima’s 2016 order distinguishes an initial review from another check after publication. The later check must occur no earlier than two weeks and within 90 days after final compensation, with an exception where the video is not publicly or reasonably accessible. The relevant date is final compensation, not the day someone first approved the draft.^[4]

Teami’s 2020 order instead calls for prompt review after publication of each specifically contracted video or post. A separate provision says monthly review of the top 50 endorsers by the previous month’s dollar sales is sufficient for a qualifying program limited to free or discounted goods or referral payments of $20 or less per month. Those conditions matter: the alternative is not a general exemption for gifted campaigns.^[5]

The practical lesson is to put the trigger beside the task. “Check the live post promptly after publication” identifies a different event from “review after final payment” or “review the selected affiliate sites monthly.” A campaign team should establish its own appropriate process using the applicable guidance and agreements, rather than borrow a historical schedule without its scope.

Machinima is the only family in this selected set that expressly combines an initial compensation-linked video review with a later review. Its initial rule allows an advance-payment route if the endorsement remains unpublished until it has been reviewed and verified. It would therefore be inaccurate to summarize the order as either “never pay before review” or “every post must always be approved before publication.”^[4]

Seven families expressly describe stopping, withholding or suspending payment as a response to a specified failure. That is a different control from checking something before the first payment. A payment team needs to know which event it is being asked to act on, who can authorize the action and what evidence shows that the condition has been met.

Lord & Taylor illustrates why the distinction belongs in the dataset. Its order contains termination language and permits an opportunity to correct a reasonably inadvertent disclosure failure, subject to specified conditions. We did not code an express payment-stop clause from that termination language. This is a distinction in the wording, not a conclusion that payments may continue.^[6]

Across five families, the order text provides an optional notice-and-correction route for an inadvertent endorser disclosure failure. “Optional” and “inadvertent” are material. A workflow should not silently turn a conditional exception into a promised second chance for every failure. The exact agreement and applicable rules still determine what the people involved should do.

Warner Bros.’ 2016 order explicitly addresses a retained campaign entity as well as individual endorsers. It requires oversight of that entity’s compliance and records of the monitoring. This is more specific than identifying an agency as the party that handles a campaign.^[7]

The order also separates the response paths. Its endorser provisions include an optional cure route for an inadvertent disclosure failure. Its retained-entity provision requires suspension of payments unless and until the identified failure is cured, alongside further terms for repeat failures. Combining both into one “allow correction” box would erase who the provision concerns and how it operates.

For a business, a useful handover question is: who checks that the agency performed the agreed review? For an agency, it is: what evidence will the brand receive? For a creator, it is: who is the named contact when a live post needs attention? These are suggested operational questions. The historical comparison does not establish that any one arrangement prevents a violation.

Imagine a small homeware brand working with an agency and three creators. The campaign promotes a new storage range. Its initial tracker has one row per creator, a draft link and an approval date. The brand and agency use the questions above to make the handover more specific. This example is invented; it is not a surveyed campaign or an interpretation of any named respondent’s conduct.

For each agreed post, the agency adds the intended account, format and publication window. It identifies who sends the creator the disclosure and claim instructions and who records acknowledgment. The brand’s product lead handles product-fact questions. The creator has a named contact for anything unclear before publication.

The team then distinguishes the draft from the live item. The draft reviewer records which version was checked. A separate owner receives the published link, records when it was reviewed and notes any change that needs attention. If a post cannot be accessed, the row stays unresolved with an assigned follow-up, rather than silently becoming complete.

Finally, the brand and agency write down how an issue reaches the person responsible for deciding the next action. They keep any payment decision tied to the actual contract and appropriate advice. The creator can see who will communicate that decision and how a correction will be checked. The example makes no promise of compliance or improved campaign performance; it makes the handover understandable.

Use one row per agreed deliverable, with a linked record for campaign-wide instructions. The following fields are a suggested planning aid, not a form tested by this study or a list of universally required legal records. Keep only information the team is authorized to hold, and set retention using its applicable requirements.

Content and relationship. Record the creator or affiliate, the responsible brand and agency contacts, the deliverable, the intended account and the relevant commercial relationship. Link the agreed instructions and acknowledgment. A campaign name alone is too broad to identify the post someone must review.

Review event. Record what is being checked, who checks it, and what starts the task. Distinguish a draft version, a published URL, a scheduled follow-up and a review of the agency’s work. Put the actual date beside the evidence of the completed check. If the trigger has not occurred, leave that task pending.

Issue and next action. Describe the observed problem plainly, identify the decision owner, and record the next step and follow-up date. Keep a correction request separate from confirmation that the correction was made. Where payment or termination is involved, refer to the controlling agreement and decision rather than infer an automatic consequence from a generic status.

A fictional row could read: “Storage-range video; published URL supplied by creator; live review assigned to agency campaign lead after publication; finding recorded against the viewed version; any unresolved disclosure question sent to brand contact; correction rechecked before the row is closed.” The team would separately fill in its agreed timing and payment arrangements. The row explains work without pretending to settle those choices.

For creators, the same worksheet can be shorter: what was agreed, who reviewed which version, who receives the live link, and who answers questions afterward. The aim is that the creator and business can both explain the next action without reconstructing it from several message threads.

Training, monitoring and intermediary responsibility are established subjects. Bladow’s 2018 legal scholarship already discusses those principles and differences between selected orders. Our contribution is an openly reproducible comparison that keeps the content, recipient, timing and payment conditions separate, with a source location for each code.^[9]

The counts help locate examples; they do not rank practices by effectiveness. Eight families sharing a bundle does not prove that the bundle prevented harm. One family containing a particular initial-payment review does not make it unusual across all enforcement or unnecessary elsewhere. This selected set cannot support either conclusion.

The useful output is a better question at campaign handover: what does “reviewed” mean here? If everyone can identify the content, the person, the trigger, the record and the next action, the process is easier to follow. Whether it meets the campaign’s legal requirements remains a separate assessment against current rules and its actual circumstances.

We selected 12 identifiable final FTC administrative or court orders concerning individual online endorsement campaigns. Their issue or entry dates span 2011–2020. Searches covered a broader period through October 17, 2026, using the FTC case index, known cases and primary complaints to check scope. The index was incomplete as a retrieval frame, so this is a named historical set, not a census or representative sample.^[1]

The set covers Legacy Learning, ADT, Deutsch LA, Machinima, Lord & Taylor, Warner Bros., CSGOLotto, Creaxion, Inside Publications, Teami, Lunada and Genesis Today. Creaxion and Inside concern the same campaign and form one family for the primary counts. Counting orders instead gives nine bundle-bearing orders out of twelve, rather than eight families out of eleven.

We coded express future obligations in the final orders, separately from allegations about past conduct. Each code has an order part, page reference and paraphrased explanation. An independent pass reviewed the full main-order text of all twelve original PDFs and visually checked material pages. Disagreements about scope, clause references and creator-versus-agency remedies were adjudicated before the corpus was frozen.

Four cases have broader boundaries: ADT mixes broadcast and online activity; Legacy concerns affiliates; Lunada concerns a brand-controlled personal-style blog; and Genesis combines mixed media with an owner-endorser. Removing all four leaves six bundle-bearing families out of seven. We retain both results so readers can see how the selection matters, rather than present a percentage as an industry benchmark.

NourishLife remains a documented pending candidate: official material confirms an entered order, but the exact entered text and campaign boundary were not resolved. It is not coded as a negative or silently treated as ineligible. If later included, the denominator would rise to twelve families and the bundle count would be eight or nine, depending on its verified clauses.

A negative code means the selected control was not expressly identified in the reviewed main-order text. It does not establish that no duty existed. We use exact descriptive counts without confidence intervals: these are selected legal documents, not a probability sample. We do not assess present enforceability, compliance outcomes or the truth of every complaint allegation. The downloads provide the coding, source references, selection record and reproducible calculations.