Nobody can reconstruct who approved what
The post is in a scheduler, the sign-off in an email, the change request in a comment thread. When a client disputes a published claim, the record is three tools and a screenshot.
For agencies and in-house teams
A control room per brand, not a group chat per client.
At agency scale the risk moves from writing to governance: who approved what, which claim was allowed, whose voice this draft is in, and what a junior can release without a second pair of eyes. All four live on the work itself.
Role-based releaseDecision trail per postDPA on request
The short version
SelfSM is social media management software for agencies and in-house teams that need per-brand separation and a defensible approval record. Every brand gets an isolated workspace with its own voice profile, confirmed facts and forbidden claims; roles decide who can draft, approve and release; and every post carries the trail of who decided what and when. Clients approve through a link scoped to their brand alone. The Agency plan covers eight brands, 975 posts a month, client approval links and unlimited approvers for $129/month, with a DPA on request.
What scale actually breaks
Writing scales. Accountability does not.
Three failures that only appear once there is a book of clients.
Generated claims are a client risk
A model that fills in a customer number or a compliance-adjacent claim does not know it invented it, and the person pasting it often does not either. That reaches a client feed under your name.
Onboarding a strategist takes weeks
The brand knowledge lives with whoever ran the account. They leave, and the voice quietly resets to whatever the next person writes like.
How the book runs
Same loop per brand, with the release gates you need.
Material, draft, review, schedule and reply happen per brand, with permissions deciding who moves work between those stages.
One workspace per client
Voice, facts, constraints, plan and history are scoped to the brand. A strategist added to one account sees only that account.
Per-brand access
Claims checked before a human sees them
Facts and negative constraints apply during drafting. Missing evidence narrows the claim or stops the draft — it does not get written and caught later.
EvidenceBlocked
Roles decide who can publish
A junior drafts, a lead approves, the client releases. Each step is an explicit decision recorded against the post.
RolesApproval trail
A client link, not a deck
Reports and approvals go out as scoped links. Missing data stays missing rather than being smoothed into a result.
Client linkHonest metrics
Against the usual stack
Where an agency stack leaks
Enterprise schedulerAI writerDocs + spreadsheetsSelfSM
Brand voice per clientNot its jobOne tone settingA doc per clientProfile per brand
Blocks unsupported claimsNoNoHuman review onlyBefore the draft exists
Client sign-offCalendar slotNoneEmail threadScoped approval link
Who approved whatPublish logNoneReconstructedOn the post
Knowledge if staff leaveLostLostPartlyStays in brand memory
What it costs
Eight brands, unlimited approvers, one trail.
Agency
$129 / month
Client approvals, roles and an audit trail across the whole book.
- 8 brands
- 975 posts / month
- Client approval links
- Unlimited approvers
- Permissions and trail
More than eight brands, or need a DPA and custom terms? Write to contact@selfsm.com. Compare every plan.
Before you sign up
Questions this page should answer
Can a client approve without seeing our workspace?
Yes. The approval link is scoped to that brand and shows only the drafts waiting on them. They cannot see other clients, your plan, or anything not sent for approval.
What does the audit trail actually record?
Who drafted, who approved or rejected, what changed between versions, which facts a claim was based on, and when each decision happened — attached to the post rather than a separate log.
How do you prevent a generated claim reaching a client feed?
Facts and negative constraints are applied while drafting, not reviewed afterwards. Without support Chloe narrows the claim or refuses the draft and asks for evidence, so the unsupported version never exists to be pasted by mistake.
Can we restrict who publishes?
Yes. Roles are per brand: drafting, approving and client release are separate permissions, so a junior can prepare everything and release nothing.
Is client data used to train models?
No. Data is stored per workspace and never used for training. A DPA is available on request.
What happens when the account lead leaves?
The brand voice profile, confirmed facts, constraints and decision history stay in the brand workspace. The next strategist inherits the account instead of rebuilding it from published posts.
From the blog
Worth reading first
Who is responsible when a claim turns out wrong
The four places claims break, and who owns each.
Who drafts, who approves, who releases
The permission split that lets juniors do real work.
Handing over an account without losing the voice
What to capture before the strategist leaves.
The approval workflow that doesn't lose the client
Where client work slows down, and the fix.
Not your situation? For foundersFor SMMs