The agency social media playbook: setup, rhythm, and what to measure
A working plan for the first ninety days of running social for a client — what to set up, who owns each step, the weekly rhythm, what to measure, and the signals that mean you should change the plan rather than push harder.
Most agency social engagements fail in the operating layer, not the creative layer. The brief was fine. The writers were fine. What went wrong was that nobody owned approval, the cadence was set to what the client wanted rather than what they could review, and by week six the calendar had gaps that looked like the agency missing deadlines.
This playbook is the operating layer. It assumes you already know how to write. It covers the first thirty days of setup, days thirty to sixty of finding the real rhythm, and days sixty to ninety of deciding whether the plan is working. It is written for one client account; run it once per client and the pattern repeats cleanly.
One constant throughout: a human approves every post before it publishes. That is not a phase-three refinement. It is the first thing you set up, because the question of who is allowed to say yes determines almost everything else about how the engagement runs.
Days 1 to 30 — set up so the loop cannot stall
The goal of month one is not great posts. It is a loop that runs without chasing. Get the plumbing right and the quality follows in month two.
Name one approver and one backup, in writing, in the scope document. Ambiguity here is the single most common cause of late posting. Connect the client's accounts and confirm you are posting to the client's page rather than a personal profile. Build the voice profile from the client's own past posts and whatever brand material exists, and write down three words they use and three they never use — that short list outperforms a brand deck in practice.
Set the cadence to what the approver can review, not what the client asked for in the pitch. Two solid posts a week that always run beats five that stall in week three. Book the weekly review block into the client's calendar now, while enthusiasm is high, and agree what is off-limits: competitors, pricing claims, politics, anything currently in dispute.
Before the first post goes live, run one full dry cycle end to end. Draft, review, edit, approve, schedule. Doing that once with no live post at stake surfaces every workflow problem cheaply.
Days 31 to 60 — find the real rhythm
By now you know what the client actually does rather than what they said they would do. Adjust to reality instead of pretending.
Watch the approval lag. If drafts routinely wait more than three days, the problem is the review slot, not the client. Move it, shorten it, or reduce cadence until it holds. Watch the edit rate too: if the client is rewriting more than about a third of every draft after four weeks, the voice profile is wrong and no amount of extra drafting fixes it — go back and rebuild it from their own posts.
This is also where voice should be visibly improving. Edits made during review feed back into how the agent writes next time, so a client who corrected the same thing three weeks running should stop seeing it. If they do not, that is a signal worth chasing.
Start the reporting habit now rather than at the quarter. A short monthly note on what ran, what got the most engagement, and one thing you are changing next month. Short and on time beats comprehensive and late, every time.
Days 61 to 90 — decide, do not drift
Month three is where an engagement either becomes profitable or quietly becomes the account everyone dreads. Make the call deliberately.
Count the hours the account actually consumed, including approval chasing and reporting, and compare that to what the retainer priced. If the gap is coordination rather than production, fix the loop before adding people.
Review the channel mix honestly. If a platform has produced nothing in ninety days and the client's audience is not really there, propose cutting it and putting the effort into the one that works.
Then change one thing, not five. Cadence, format mix, or platform — pick the single variable most likely to matter and hold the rest steady for another month.
Who owns what
Write this into the scope document, not into a Slack message. The account lead owns the brief and the cadence. The agent produces research and drafts. A human on your side does the first-pass quality check. The named client approver has final say, with one backup.
The failure mode to avoid is diffuse approval — three client stakeholders who all comment and none of whom can decide. If that is the structure you are given, say early that it will slow posting, and agree who breaks ties.
What to measure, and what to ignore
Measure four things per client per month. Posts published against posts planned, which tells you whether the loop is working. Approval lag in days, which is your leading indicator of everything else. Edit rate, which tells you whether the voice is right. And whatever the client's actual business signal is — inbound conversations, demo requests, bookings, applications.
Ignore follower count as a headline. It is slow, easy to inflate, and disconnected from anything the client cares about. Report reach and engagement as context, not as the story. A monthly note with four honest numbers and one recommendation is a better artefact than a twenty-page deck nobody opens, and it takes a fraction of the time to produce.
A review checklist
A process for the person who approves posts. Adapt it to the policies your agency already has, and check anything you are unsure about with whoever owns compliance where you work.
Name the approver and the backup in writing
One person with final say, one backup, both recorded in the scope document. If the client insists on a committee, agree in advance who breaks ties and what happens when a draft is not reviewed by the deadline.
Build the voice profile from the client's own words
Use their existing posts, their site, and their sales material. Write down three words they use and three they avoid. Check the first two weeks of drafts against that list before anyone else sees them.
Agree what is off-limits for this client
Competitors, pricing claims, politics, anything the client is currently in a dispute about, and any topic their own compliance function restricts. Put it in the brief so a reviewer can check drafts against something written down.
Run one full dry cycle before the first live post
Draft, review, edit, approve, schedule — end to end, with nothing publishing. It surfaces every access problem, permission gap and workflow ambiguity while the cost of finding them is zero.
Book a fixed weekly review block
Thirty minutes in the client's calendar, same time each week, batch approval. Weekly batch review is the single change that most reliably stops posting from slipping, because it replaces daily interruptions with one kept appointment.
Verify claims and consent before approving
Anything that reads as a fact should be sourced to something the client provided. Anyone identifiable in a photo or story needs recorded permission. Both checks belong in review, not after publication.
Send a short monthly note on time
What ran, what performed, one recommendation. Four numbers is enough. On time and short builds more trust than comprehensive and late, and it takes an hour rather than a half day.
Review the account against its hours at day ninety
Count real hours including coordination, compare to what the retainer priced, and decide deliberately. If the gap is coordination, fix the loop before hiring. Then change one variable and hold the rest steady.
Where Bolta fits
Bolta’s agents research and draft posts, then hold them for review. The approval workflow is designed so that a human reviews every post before publication. Bolta does not decide what is appropriate for your agency to say, and it does not replace the review your agency already does. Follow the policies you already have, and consult your compliance officer or counsel on anything specific to your situation.
Frequently asked questions
How long should it take to onboard a new social client?
Setup — connecting accounts, building the voice profile, agreeing cadence and off-limits topics — is typically an afternoon of focused work plus one client call. The first live post can go out in week one. What takes longer is the voice settling, which usually needs four to six weeks of drafts and edits before the review time drops noticeably.
Who should approve client social posts, the agency or the client?
Both, in sequence. Someone on the agency side does a first-pass quality check so the client never sees a rough draft, then a single named client approver has final say. Name a backup for holidays. Diffuse approval across several client stakeholders is the most common cause of late posting, so agree who breaks ties in week one.
What cadence should an agency commit to for a client?
The highest cadence the approver can actually review every week. For most B2B clients that is two to three posts a week. Five is usually the client's ambition rather than their capacity, and a cadence that stalls in week three reads to the client as the agency missing deadlines. Start lower and raise it once the loop holds.
How do you keep a social retainer profitable?
Count the hours the account actually takes, including approval chasing and reporting, not just writing. Most unprofitable retainers are drowning in coordination rather than production. Fixing the approval loop — one named approver, one fixed weekly review block, edits captured in one place — usually recovers more margin than adding a writer.
When should an agency change the plan rather than push harder?
When a platform has produced nothing in ninety days and the client's audience is not really on it, or when the client is rewriting more than about a third of drafts after a month, which means the voice profile is wrong. Change one variable at a time — cadence, format mix, or platform — so the next review has usable information.
Does Bolta publish posts automatically once they are scheduled?
Only posts a human has approved. Drafts sit in a review queue until someone approves, edits, or rejects them, and an unapproved draft never publishes — the slot simply stays empty. There is no autonomous publishing mode, which is deliberate: publishing on a client's account without sign-off is the failure that ends an agency relationship.
See what Bolta would write for your agency
Bolta’s agents research, draft and schedule posts. Nothing reaches a public account until a person on your side approves it. Paid plans start at $19/month, and there is a free Starter tier.
Last updated: 2026-07-27