The problem
Managing social media for multiple clients simultaneously is one of the highest-volume, most time-consuming services a marketing agency can offer. Each client has their own brand voice, audience, content calendar, approval process, and performance benchmarks. Producing fresh, on-brand content for ten clients week after week without quality degrading, deadlines slipping, or team members burning out is a genuinely hard operational problem.
The typical social media management workflow is heavily manual: the content manager sits down each week, refers to the content calendar, writes posts from scratch for each client, finds or commissions visuals, schedules everything in a tool like Buffer or Later, sends a round of approvals, processes feedback, and republishes. Then they do it again next week. And the week after. The repetitive nature of the work makes it difficult to maintain creative energy, and the volume makes it difficult to give each client the attention they deserve.
There is also a performance commentary problem. Every month, someone has to pull the numbers, look at what worked and what did not, and draw conclusions. For a social media manager with ten clients, this monthly analysis and commentary writing can take a full day. AI changes the economics of social media management fundamentally — not by replacing the human judgement that makes social media effective, but by absorbing the volume of production work that currently crowds it out.
The system
Step 1: Build client voice profiles (Claude)
Before touching any content, create a structured voice profile document for each client. This is a one-time investment that pays back every week.
Use Claude to generate a voice profile from a sample of approved content:
"Here are 10 examples of approved social media posts for [client name]: [paste posts]. Analyse these posts and write a brand voice profile covering: tone in three to five adjectives (with what each means in practice), typical sentence structure and length, vocabulary to use and avoid, how the brand talks to its audience (formal/informal, first person/second person), content themes that appear repeatedly, and two example rewrites showing on-brand versus off-brand language. UK English."
Store these profiles in a shared Notion space so any team member can brief Claude accurately without memorising each client's style.
Step 2: Generate weekly content batches (Claude)
Rather than writing posts one at a time, batch the generation. Once a week, run a content generation session for each client in under 30 minutes per account.
Use a prompt structured around the week's content pillars and any specific briefs or campaigns:
"You are a social media copywriter for [client name]. Brand voice: [paste voice profile summary]. This week's content themes are: [list from calendar]. Channels required: LinkedIn (3 posts), Instagram (5 captions), Twitter/X (5 posts). Any specific campaigns or events to reference: [list]. Write all content in a single pass. For each post include: the channel, the format (text only / carousel / image with caption / video caption), and a brief note on the visual direction. UK English. No hashtags in the body copy."
Review the batch, edit for anything that sounds off-brand, and remove or replace any posts that do not quite work. You will typically keep 70 to 80% of the output with light editing.
Step 3: Visual briefing (Claude)
For posts that need a visual, use Claude to write the visual brief for your designer or for an AI image tool:
"Write a visual brief for a social media graphic for [client name]. The post copy is: [paste]. The brand colours are [describe]. The visual should: [describe direction from the content generation step]. Output format: a 3-sentence visual brief suitable for a designer or for prompting an AI image generation tool."
This keeps your visual direction consistent and saves the back-and-forth of explaining what you meant by "something clean and modern."
Step 4: Build an approval and scheduling workflow (Make)
Set up a Make (formerly Integromat) workflow to manage the approval process. The basic flow: when content is added to the "Ready for Approval" view in your Notion content calendar, Make automatically sends an email or Slack message to the client with a link to the relevant Notion page. When the client marks posts as approved (via a simple status field), Make triggers a notification to the social media manager to schedule the post.
This eliminates the manual chasing of approvals that typically consumes 30 to 60 minutes per client per week. It also creates an audit trail: every post has a record of when it was submitted for approval and when it was approved.
Step 5: Performance commentary (Claude)
At the end of each month, export each client's social media analytics (most platforms offer CSV exports) and use Claude to write the performance commentary:
"Here is the social media performance data for [client name] for [month]: [paste CSV or data summary]. The client's goals are [describe]. The previous month's numbers were [paste]. Write a performance summary covering: what went well and why, what underperformed and why, three specific recommendations for next month. Tone: clear and confident. Length: 300 to 400 words. UK English."
The results
Before implementing this workflow, a social media manager could realistically manage six to eight clients to a reasonable standard. Content production, approvals, scheduling, and reporting consumed essentially all available hours.
With this system, the same manager can handle twelve to fifteen clients. Content batch generation per client drops from 2 to 3 hours per week to 30 to 45 minutes. Monthly reporting drops from 4 to 5 hours to 60 to 90 minutes. The approval workflow automation alone saves 30 minutes per client per week in chasing and logging. Across a twelve-client book of business, total time saved is typically 8 to 12 hours per week — the equivalent of a part-time hire recovered through process improvement.