A social media manager's week is mostly assembly. Pulling content from three departments, fitting it into the calendar, drafting captions across formats, scheduling, monitoring, replying, and then doing it all again Monday. The actual creative work — the post that breaks through, the campaign that lands — happens in maybe ten percent of the time. The rest is operational drag, and most of it is a search-and-reassembly problem across too many apps. This guide is about a working setup for the social media manager — calendar, analytics notes, idea inbox, drafts, archive — in one vault where the agent can do the slow middle work. The scheduling tool, the analytics platform, and the visual asset library still do their jobs. The vault is where the thinking and the drafting happen. ## What a social media manager's notes actually have to hold The job has four loads, each of which fails in a particular way when it lives in the wrong place: - **The content calendar** — the running schedule of what's going out across platforms. Usually lives in a spreadsheet that's hard to query and harder to draft from. - **An idea inbox** — observations, headlines, audience questions, and trend captures that might become posts. Usually lives in a Slack channel nobody scrolls back through. - **An analytics log** — the post-mortem on what's worked, with notes about why. Usually never written down because the analytics tool exports a PDF nobody re-reads. - **A drafts surface** — the actual writing happens here. Usually in three different tools depending on which platform. A vault holds all four together. Plain markdown for the text. Inline databases for the calendar and the analytics log. The agent reading across all of it when you draft. The structural shape pairs naturally with [how to build a content calendar from your notes](/guides/creatives-content/content-calendar-from-notes/) — this guide focuses on the social-media-specific operational shape. ## The content calendar as an inline database A spreadsheet for the calendar dies the moment it stops being the place drafts get written. The shape that holds is an inline database in a markdown page, embedded via the `:::database:::` directive, with prose around it for the strategic context. Columns: date, platform, format (post, story, reel, thread, etc.), theme, status (idea / drafted / scheduled / posted), and a link to the draft page. Rows for everything in the next four to six weeks. The database lives inside a planning page that has prose around it — your strategic thinking on the quarter, the campaigns you're running, the themes you're underweight on. The prose holds the strategy. The database holds the schedule. Both update as the quarter unfolds. The agent can answer queries across the calendar alongside the rest of the vault: "Pull every scheduled post for next week. For each, summarize the draft status and flag anything that still needs visual assets." That's a Monday morning briefing instead of an hour of cross-referencing tools. For longer-term planning, the same database supports filtered views. Filter by platform to see whether you're underweight on Instagram. Filter by theme to see whether the customer-stories thread has been quiet for three weeks. Filter by status to see what's pending versus what's scheduled. ## The idea inbox and the weekly triage pass Every social media operation has more ideas than it has time to develop. The challenge is that the ideas usually live in scattered places — Slack threads, email forwards, screenshots in a phone roll, mentions of "we should post about X" that nobody captured. By the time someone sits down to plan content, they're working from a blank page when there were dozens of seeds. A "Future posts" page with a simple inline database — title or hook, source of idea, theme it maps to, format guess, status. Embedded directly via the `:::database:::` directive. Adding a row takes ten seconds. For sources that produce ideas constantly, capture the raw material in sub-pages. Customer support tickets that surface common questions. Conversations with sales about objections. Quotes from internal meetings about the product roadmap. The agent helps you spot which ones are ready to become posts: "Read the last thirty entries in the idea inbox. Group them by theme. Flag the three most likely to become a strong post in the next two weeks." The triage pass is something you do once a week. Open the page, look at the list, decide which two or three to advance into the calendar. Archive the rest with a note. The graveyard problem is mostly about not deciding fast enough; the regular pass fixes it. ## The analytics log — what worked, with the why captured Most social media operations look at analytics and forget what they saw within a week. The dashboard is there, but the synthesis — the running understanding of what's actually working for this account, on this platform, with this audience — gets lost between weekly check-ins. An "Analytics" sub-page captures the synthesis. Once a week, after pulling the platform analytics, type three to five lines about what you saw. The post that overperformed and your guess at why. The format that's quietly compounding. The day of the week that's stopped working. The campaign that landed harder than expected. The agent reads the analytics log when you draft: "Read the analytics log from the past three months. What patterns hold? Which formats are quietly compounding? When you draft this week's posts, weight toward what's been working." You get drafts that are informed by the actual record, not by guesses about what might work. For specific post-mortems on campaigns or major posts, sub-pages hold the full analysis. Useful next quarter when you're planning a similar push. ## The drafting flow — one source, many outputs Different platforms want different formats. Instagram captions versus X threads versus LinkedIn posts versus TikTok hooks. The temptation is to draft each in its own tool. The result is fragmented drafts and lost material. A "Drafts" section in the vault holds everything. One page per piece. The page has a header with the format and the platform, the source material it draws from (the calendar entry, the idea inbox row, the brand guidelines), and the actual draft. The agent helps with format-shifting. You write the long version once — say, a 250-word LinkedIn post about a new product feature. Then ask: "Adapt the LinkedIn post into a four-tweet X thread with a stronger hook. Adapt it into a two-line Instagram caption that drives to the link in bio. Adapt it into a TikTok hook that opens with a question." You get three drafts. You edit each into something that works on its native platform. The compression is real — the work that used to be three separate writing sessions is now one writing session plus three editing passes. For the deeper drafting workflow, see [how to draft emails, proposals, and newsletters inside your notes app](/guides/creatives-content/draft-emails-proposals-in-notes/) — the social-media-multi-format case is a faster variant of the same shape. ## The swipe file — hooks and structures that work on each platform Most social posts run on a small set of structural patterns. The hook structures change by platform. The Instagram caption opener is different from the LinkedIn post opener is different from the TikTok hook. A swipe file captures the structures that have worked. A "Swipe" sub-page with one entry per saved hook or post. Source, platform, structure, what makes it work, your two-line note on why it worked. Over months, the file becomes a real reference. When drafting, the agent reads from the swipe file: "Pull every Instagram caption opener from the swipe file that uses a counterintuitive observation. Adapt one for tomorrow's post about the new feature." You get a starting point grounded in patterns you've already validated. The deeper-dive on swipe-file mechanics is at [how to build a swipe file in your notes app](/guides/creatives-content/build-swipe-file-notes/). ## The published archive — your back catalog as material Once you've been posting for a while, the back catalog becomes its own material source. The post from eight months ago that landed well. The campaign that flopped and taught you something. The content series that quietly compounded into your most-followed thread. Pulling the published archive into the vault lets the agent reference it. "Have I covered the customer-story angle on this product feature before? When? What was the framing? Did it work?" You stop accidentally repeating yourself. You spot the recurring theme you've been developing without naming it. For brand voice consistency over time, the archive is also a teaching asset. When a new social manager joins the team or you onboard an agency partner, the archive shows them what the brand actually sounds like — not what the brand guidelines say it sounds like. The agent can pull representative samples: "Pull five recent posts that exemplify our brand voice on the leadership theme. Note what specifically makes them sound like us." ## Engagement, replies, and the relational work A social media account is a relationship platform, not just a publishing one. The compounding asset isn't the posts — it's the conversations they create. Most social media operations get this wrong by posting and then leaving the comments unattended. A "Comments" sub-page captures the threads worth coming back to. Notable replies, useful pushback, customers who showed up consistently. Three weeks later when you're drafting on a related theme, the agent can pull from the comment threads: "Find every thoughtful comment on posts about the new feature. Pull the substantive points. Are there reply-worthy ones I missed?" For replies under your own posts, the discipline is just to do them. The agent doesn't help with this — it's relational work that has to be human. But the brands that reply consistently are the ones whose social presence compounds past the platform algorithm. ## Stakeholder context and the chronic alignment problem A social media manager often sits between many internal stakeholders — the founder who wants more leadership content, the marketing director who wants more lead gen, the product team who wants every feature announcement to feel "premium," the sales team who wants more case studies. Each one has a strong opinion. None of them have time to read the calendar before saying what they want. A "Stakeholders" sub-vault, with pages per major internal voice, captures the running context. Their priorities. Their preferences. The asks they've made before that you've handled. The things they care about that have shifted over time. When a new ask comes in — "we should be posting more about X" — the social media manager has the receipts. The agent reads across the past three months: "Has the marketing director asked for X before? When? What was the conversation? Did we end up doing it, and how did it perform?" You can respond with grounded context instead of guessing. ## A boundary on what AI should and shouldn't do A practical note: the agent helps with assembly, format-shifting, and surfacing material from your archive. It doesn't replace your taste, your sense of the audience, or the actual creative judgment that distinguishes a good social presence from a generic one. Posts that are AI-generated end-to-end have a flat tone you can hear in three sentences. The brands that succeed on social long-term are the ones with a genuine point of view, not the ones that automate the writing entirely. Use the agent for the chores. Do the actual writing yourself, especially the parts where the brand's voice has to come through. ## A calmer way to run social Running social media doesn't have to feel like a treadmill. The treadmill comes from drafting reactively in three tools, with no idea inbox, no analytics synthesis, and no system for repurposing across formats. The shape that calms it down is fewer pieces — a calendar database, an idea inbox, an analytics log, a drafts surface, a swipe file — and a small weekly habit of moving things forward. Try Docapybara free — [sign up](/accounts/signup/), build a calendar database for the next four weeks, drop in the last ten things you've saved that triggered an idea, and ask the agent for two drafts grounded in that material.