How-To Guide

Which posts should require customer review before they go live?

A risk-based matrix separating post classes that return to the customer from the classes a standing account-level posting permission already covers.

8 min read

By the Superdirector editorial team. FINRA Rule 2210, the FTC Endorsement Guides pages, the approval-workflow guides and Superdirector's own pricing and how-it-works pages were read November 15, 2026. Every worked example is a labelled hypothetical, not a customer account.

At a glance

Which posts should require customer review before they go live?

Two questions decide it. First, what has the customer already agreed to for this account, because a standing posting permission is itself a decision and the classes it covers are exactly the ones that do not come back for review. Second, what does the post claim: Tareno's risk rule compresses the rest into one line, low-risk content needs speed while high-risk content needs control. The classes that normally come back to the customer are regulated or specialist claims, money and price claims, commitments to a named third party such as a sponsor or a partner, and synthetic imagery presented as proof. Evergreen tips, routine brand posts and replies to comments usually stay inside the standing permission instead of earning one review each. Write the split down as a small matrix keyed to the account, name what re-opens review after approval, and name the default for when the customer never answers, so silence becomes a rule instead of an argument.

  • Start at the account, not the post: the scope the customer actually agreed decides which classes even can go live without one review each.
  • Four classes keep coming back to the customer: regulated or specialist claims, money and price claims, third-party commitments, and synthetic imagery presented as proof.
  • Re-approval keys to the claim, the price and the asset, not the punctuation: most guides exclude typo fixes and formatting cleanup from a new round.
  • Set the silence default in writing before you need it, because unapproved content that ships quietly is the failure every source names as the one unforgivable move.
  • On Superdirector's Growth plan, posting turns on for an account after one warning, and after that its eligible Instagram carousels and single X text posts publish by themselves; the classes above stay the customer's own scope decisions.
A five-row matrix placing four claim classes in the reviewer column and routine posts in the standing-permission column.

Four claim classes each carry a named reviewer, while routine posts resolve inside the account's standing permission. Original diagram, drawn from this page's matrix table and the cited approval guides, November 15, 2026.

Superdirector

Make it work for your brand

Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.

The matrix at a glance

Keep two columns and let each row carry one decision. The post class names what the post is about, the authority names who decides, and the publish gate states whether the post can leave at all before a named person reads it. The class rule comes from the sources this page analyses; the permission column is where the account rule takes over, and it is filled in from the agreement the customer signed.

Post classExamplesWhere the authority livesTypical route
Regulated or specialist claimsFinance and returns claims, health statements, pharma and public-sector contentNamed specialist or compliance reviewer under the industry's own rulesAlways reviewed before it can go out; FINRA pre-approves the strict form
Money and price claimsPrices, discounts, offers, availability, earnings wordingThe person who owns the price, verified against the current sourceAlways checked; a wrong price becomes a correction after the fact
Third-party commitmentsSponsor and partner posts, client content, paid-partnership wordingsThe named third party, together with the customerAlways agreed with the party it binds; disclosure wording checked
Synthetic imagery as proofAI product shots, AI product screens, synthetic people as customersThe customer, on the accuracy decision beside the assetPublish, revise or replace before it meets an audience
No-claim routine postsEvergreen tips, brand-voice posts, comment repliesThe standing permission the account already carriesPublishes inside the permission without one review each

Read the rows as two regimes, the way Tareno's matrix does: routine rows need nothing but the permission they already sit inside, and claim rows need a named decision before the post can leave. Kontentino puts the always-clause in plain words for the class it knows best, campaigns, announcements and anything with claims get sign-off, and Montr reaches the same split from the agency side with a chain that skips a stage for small clients and adds a legal check for regulated ones.

Start from the scope the customer actually agreed

Before routing a single post, read what the account has been granted. On Superdirector's Growth plan the agreement is concrete: for 199 dollars a month the account's eligible Instagram carousels and single X text posts publish by themselves once posting is turned on for an account, and the page that states it also states the other half, after one warning that names the account, the public effect and the cost. Once posting is on, no post asks for approval after that; the customer turns it off any time, and edits and corrections still happen in the drafts.

That single switch is why this article exists in that shape. The customer already made the decision a week of per-post queues would otherwise spread out, so the review question moves from this post to the four classes named next: does the account's permission reach this kind of claim at all? Classes the customer wants kept close should be written as exceptions before the switch flips, because after turn-on the machine fills the calendar instead of a queue holding each post.

The plan tier matters too, and the pricing page keeps them distinct. Starter at 29 dollars a month covers live social data, ideas, scripts and downloads from Claude or ChatGPT and does no publishing, so on Starter nothing in this matrix publishes anywhere and every post is manual. Growth carries the publishing permission, the 2,000 monthly credits, and the account switch this section described.

Sort the four classes that return to the customer

Class one is regulated or specialist claims. Finance is the written case: FINRA Rule 2210 requires that an appropriately qualified registered principal approve each retail communication before the earlier of its use, with the duty lifted only for the interactive-forum class of posts and supervision still applying elsewhere. Hootsuite's approval guide compresses the same class across industries: in finance, healthcare, government and pharma it routes content to compliance, healthcare content to privacy review, and it names the finance claim line directly, claims about performance or returns require pre-approval.

Class two is money. A price, a discount, an offer or an availability promise publishes a fact the customer will be asked about, and a pricing claim should not publish without verification is the whole reason Tareno's guide gives it its own row. The accountability behind the class is legal rather than stylistic: under the FTC's endorsement guides your company is ultimately responsible for what others do on your behalf when a network member says something false about the product.

Class three binds someone else. A sponsor post, a partner feature or a client post promises a thing that party never wrote, so the review loop runs through both the customer and the named party, and the disclosure wording travels with it. The FTC's brochure for the influencer case tells the customer what the wording must do, the endorsement message should make a material connection obvious, in the same language as the endorsement itself, and it rejects the half-measures: vague shorthand like abbreviations, or assuming a platform's own disclosure tool is good enough on its own.

Class four is proof-shaped imagery. Soar's AI-image guide sorts each asset into publish, revise or replace, and the replace row is the class's hard edge: a generated product screen showing an unavailable feature, or a synthetic person presented as a real customer also needs replacement or a clearly different presentation. The context rule travels with it, because a harmless illustration can become misleading beside a headline that presents it as a real installation. Where the image claims something about the product, the decision belongs above the publication line.

Name the classes the standing permission covers

Everything without a claim in it stays inside the account's permission, and naming those classes is what stops the queue from filling with work nobody reads. Evergreen tips, routine brand-voice posts and replies to comments have no price to verify, no party to bind and no specialist rule to satisfy, so their row in the matrix points at the permission and not at a person.

The vendor sources agree with each other from three directions at once. Kontentino lets recurring low-risk formats run on pre-approved templates or themes, reviewed monthly, so the client's attention goes to genuinely new work. Emplifi keeps routine posts out of the legal queue so the queue holds the posts that need a legal read, and Montr keeps the chain short by default and lets the customer lengthen it only where their own industry demands it. None of them treats review as a virtue on its own.

A middle band exists and the matrix should name it rather than blur it. Repurposed winners, rewritten captions and template mixes carry no new claim of their own, but they can resurrect an old one, which is why Tareno's rule is that repurposing follows the current risk level, not the original one. An evergreen tip rebuilt from a post that quoted last year's price inherits the money row until the claim is checked out of it.

A two-column split where routine posts stay inside the standing permission while the claim classes leave it for the customer.

The turn-on switch opens the permission that holds routine posts; the claim classes still leave it for the customer. Original diagram, drawn from the standing-permission section and the cited pricing page, November 15, 2026.

Write the matrix down and bind it to the account

The matrix earns its keep only where the account meets it. Put the table in one file the whole team can find, with the reviewer, the final approver and the publish gate named per class, the way Emplifi writes it into the governance document and Hootsuite writes it into the onboarding pack. Montr's version is the customer-facing half: a short onboarding sentence that states when the batch arrives, how long review takes, and what happens when no reply comes.

Bind the rows to the account record, not to a person's memory. The two fields that make an approval stand later are who approved it and which version they saw, the pair Emplifi's audit trail carries, and the mechanical part, the record fields, checksums and the destination check, is the neighbouring guide's subject. This matrix only decides which posts travel through that machinery.

Set re-approval triggers and a default for silence

Approval binds a version, so the matrix needs the rule for the version that follows. Tareno excludes typo fixes, formatting cleanup, minor punctuation and safe hashtag changes from a new round, and requires re-approval when the claim, the price, the call to action or the asset changes materially. Montr freezes approved posts so the published media is the approved snapshot, and in Hootsuite's structure a change after approval sends the post back through the chain. The rule keeps edits fast and swaps loud.

Then agree the silence rule before silence happens. Montr's written default publishes only the posts marked evergreen and holds the rest when 48 hours pass without a reply, and Kontentino's example SLA moves unapproved content to the next slot 24 hours before its time. Both defaults are kinder than the alternative: Kontentino names silently publishing unapproved content as the failure that costs more trust than a hundred delayed posts, and Montr keeps overrides so rare that using one weekly is treated as evidence the chain itself is too slow.

A decision fork from an unanswered review into publish evergreen, next slot, or a recorded override, with quiet publishing crossed out.

An unanswered review forks three ways, and the quiet publish beneath the rules is crossed out. Original diagram, drawn from the re-approval section's cited approval guides, November 15, 2026.

What Superdirector publishes while posting is on

The product facts belong to this matrix, not to a pricing FAQ. Growth, at 199 dollars a month with two accounts and extra accounts at 99 dollars each, schedules research and routines and supports Instagram and X publishing: when posting is turned on, an account's eligible Instagram carousels and single X text posts publish by themselves, and the monthly credit grant on the plan is 2,000. No separate approval sits on each post while that permission runs, and the switch, not a queue, is where the customer can stop it, turn it off any time with edits and corrections happening in the drafts so the next drafts come out closer to the brand.

That is the division this page keeps from front to back: the machine publishes the classes the customer's account-level decision put inside it, and the customer keeps the four claim classes as scope decisions made at turn-on time. Where a class is later moved, the matrix row and the account decision change together, and nothing else in the permission moves.

Frequently asked questions

Which posts should require customer review before they go live?

The post classes with a claim attached. Regulated or specialist claims come first, because FINRA Rule 2210 requires a broker-dealer's registered principal to approve each retail communication before use, and vendor guides route health and compliance content to a specialist reviewer. Money and price claims come next, since a pricing claim publishes unverified at the reader's cost if nobody checks it. Third-party commitments follow: anything naming a sponsor, partner or client binds a promise that person never saw. And synthetic imagery presented as proof, such as a generated product shot or a person standing in as a real customer, needs its accuracy decision before it reaches an audience. Everything else that carries no claim can live inside the standing permission the account already has.

Should the customer approve every single post?

No, and treating every post the same way is the failure mode three guides name separately. Kontentino's client-approval guide answers directly: campaigns, announcements and anything with claims should always get sign-off, while recurring low-risk formats can run on pre-approved templates or themes reviewed monthly. Emplifi's enterprise guide makes the same move from the other side: routine posts stay out of the specialist queue so the queue holds only posts that need a legal read. The customer keeps the decisions that carry risk, and loses nothing by not being consulted on the evergreen tip that goes out on Tuesday.

Does an AI-run account remove the review question?

It moves the decision earlier rather than removing it. An account that publishes by itself needs the class split agreed before anything runs, because after the switch is on there is no queue holding each post. Hootsuite writes that AI-assisted content feeds the same approval pipeline as everything else, and Tareno holds that AI output is not approved content. On a plan like Superdirector's Growth tier, publishing runs once posting is turned on for the account, so the four risk classes are the customer's scope decision at turn-on time, and a class the customer wants to keep close should be named as an exception before the switch flips.

What happens when the customer never replies to a review request?

Decide the default before it happens and write it down. Montr's agency guide gives the full sentence: if the agency hears nothing within 48 hours, it publishes only posts marked as evergreen and holds the rest. Kontentino's example SLA moves content unapproved 24 hours before its slot to the next slot. What no guide endorses is the quiet move: Kontentino names silently publishing unapproved content as the one mistake that costs more trust than a hundred delayed posts. The override that pushes something out anyway should be rare and recorded.

Do AI-generated images need review before they go out?

Yes, and the review is about accuracy rather than detectors. Soar's approval guide sorts AI imagery into publish, revise or replace: publish an accurate concept illustration with a caption that says so, revise a fixable label, and replace a generated product screen that shows an unavailable feature or a synthetic person presented as a real customer. The context rule travels with the image: a harmless illustration can become misleading beside a headline that presents it as a real installation. This class returns to the customer wherever the image makes a claim about the product.

What does approving a post actually bind?

A version, not a topic. Montr freezes approved posts so the published media is the approved snapshot, and in Hootsuite's structure a change after approval sends the post back through the approval chain. Emplifi's audit trail pairs the reviewer's name with the version they saw, which is what turns an approval into evidence instead of a memory. The mechanics of an exact approval, the record fields and the destination check, are treated separately in our approve-exact guide, and this matrix decides only which posts travel through it.

What does Superdirector publish by itself, and what stays mine?

On Growth, 199 dollars a month, Superdirector plans eligible Instagram carousels and single X text posts and publishes them once posting is turned on for an account, after one warning that names the account, the public effect and the cost. No post asks for approval after that; the switch can be turned off any time, edits and corrections happen in the drafts, and Starter, 29 dollars a month, covers live data, ideas and scripts from Claude or ChatGPT without publishing. The account-level rule is why the matrix matters here: the customer owns the classes at turn-on time, so the four claim classes in this article are the ones to name before the switch flips.

Related practical guides

Agree the classes, then turn posting on once

Walk the account through this matrix with the customer, write the four claim classes and their reviewers into the account record, and make the standing permission carry everything else.