Turn One Article Into a Week of Posts With Make and Buffer

The short answer: if your blog publishes two to four articles a month and each one gets a single social post before the team moves on, hand the rewriting to a Make agent and keep the yes with a person. Every new article becomes seven drafts for LinkedIn and X, and an approver in Buffer releases them into a week of slots. Buffer's Team plan, which holds the approval step, costs $12 per channel a month on monthly billing, so two channels are 2 x $12 = $24, and Make's Free plan includes 1,000 credits a month, more than this job uses (Buffer pricing, Make pricing, checked on 26 October 2026). For example, writing seven posts by hand at 15 minutes each is 7 x 15 = 105 minutes per article, and reading seven drafts at 3 minutes each is 21, so four articles a month drop from 420 minutes to 84.

Disclosure: I may earn a commission through some links on this site. It does not change what this page says.

An article costs a day to write and gets one afternoon on social. The agent does the rewriting, seven angles sized for their networks; a person owns everything said under the brand's name. One Buffer rule makes that split hold: posts created by a teammate with Needs Approval access are always saved as drafts awaiting approval, API posts included, per the approval article. The scenario runs on that teammate's key, so it cannot publish anything.

Before you build:

1. The credit sum, done once. For example, each article costs 10 operations after the feed check: 1 to fetch the page, 1 for the agent, 1 for the iterator and 7 for the drafts, because Make counts a trigger once per check and every other module once per bundle (operations guide). A daily check is 30 operations in a 30-day month, so four articles come to 30 + 4 x 10 = 70, plus the agent's tokens, which Make's AI Provider bills in credits (credits).
2. The channels and voice you already have. A LinkedIn page and an X profile connected in Buffer, a posting schedule with LinkedIn slots Monday to Friday and X slots on Tuesday and Thursday, and five past posts your team liked, to paste into the agent's instructions. Each slot holds one post, per the posting schedule article.
3. A teammate who can only draft. On Team, invite a user for the automation, such as a shared mailbox, and set Publish to Needs Approval on both channels (adding users). Signed in as that user, create the API key in API settings; a key acts for the account that made it (quick start).
4. A number from before. In Google Analytics, open Traffic acquisition and note the sessions each of your last four articles got from the networks you post on in its first 14 days. The same report shows session campaign once links carry UTM tags (campaign URLs).

The scenario, from new article to approved week

1. Trigger. Add RSS > Watch RSS Feed Items with your blog's feed URL and keep Maximum number of returned items low, as Make's RSS app page advises. In Schedule settings, choose Daily at a morning hour (scheduling).
2. Fetch the full page. Add HTTP > Make a request with a GET to the item's link (HTTP app). The RSS documentation does not say whether a feed item carries the whole article, so the agent reads the page itself.
3. The agent writes seven posts. Add Make AI Agent (New) > Run an agent (agent guide). The Instructions hold the voice, the five example posts and the week: the main claim, a figure from the article, the problem it answers, one step a reader can take, a quotable line, a question, a recap. Two rules go in writing: no figure or claim the article does not contain, and X posts under 280 characters, each emoji counting as two, per Buffer's character limits. The Input holds the page body and two tracked links built in the field: the article URL with utm_source set to linkedin or x, utm_medium=social and utm_campaign set to the slug. Set Response format to Data structure, with an array posts of day, channel and text. Give the agent no tools.
4. Test before you switch it on. Open Chat with Agent and feed it your last three articles. Adjust the Instructions until the drafts need light edits rather than rewrites.
5. One bundle per post. Add an Iterator on posts[].
6. Each post goes to Buffer as a draft. Add HTTP > Make a request, POST to https://api.buffer.com, with an API key keychain: Key Bearer plus the teammate's key, placed In the header as Authorization, the pattern in Make's API key guide and Buffer's quick start. The JSON body runs createPost with the text, schedulingType: automatic, mode: addToQueue and saveToDraft: true, as in the draft example. Add three fields from the input reference: needsApproval: true, accepted only where the channel requires approval, aiAssisted: true, and one fixed tag ID in tagIds for counting later. Map channelId with switch(channel; "linkedin"; "<page ID>"; "x"; "<profile ID>"), one of Make's general functions.
7. A person approves. With Collaboration emails on under Settings, Notifications, the approver gets "A draft is ready for your approval" for each post (approval article). Read each against the article: every figure in the source, the right link for the network, a sentence you would sign. Edit, then Approve, which offers Add to Queue; Reject sends a draft back. Approve in day order, because queued posts take the earliest open slot and fill the schedule chronologically (posting schedule).
8. The output. A week of posts in the slots you set, each carrying a tracked link back to the article and the same tag. For the per-channel screens in each tool, see how to use Buffer and how to use Make.

The free version, before any of it

By hand first; for a blog that publishes twice a month it is enough. Buffer's Free plan connects 3 channels, holds 10 scheduled posts per channel and includes the AI Assistant with no usage cap (pricing file), which Buffer describes as able to summarize long content into short posts (AI Assistant page). Paste the article's core section into the composer, ask for the seven angles above, rewrite each draft in your own words, and add them to the queue.

The scenario also runs on free tiers, with one change. Adding users needs Team, so on Free there is no draft-only teammate, needsApproval comes out of the request, and saveToDraft: true is the only guard; drafts exist on every plan, and the owner moves them to the queue from the Drafts tab (approval article). The Free plan's API allowance is 1 key and 3,000 requests a month, per the pricing file; for example, four articles at seven drafts each are 4 x 7 = 28 requests.

The paid tiers add structure, not words. Team turns the guard from a flag anyone editing the scenario can change into a permission on a person. A paid Make plan lets the agent use your own OpenAI or Anthropic connection, where the provider bills the tokens (credits).

What it costs each month

All prices were checked on 26 October 2026. Make's pricing page draws its paid prices in the browser behind a bot check, so they are left out here; the credit sum above shows why this job does not need them.

One default matters more than any tier. Make starts every scenario on a schedule of every 15 minutes (scheduling), and a feed check costs an operation each time it runs (operations guide): 4 x 24 = 96 checks a day, 2,880 in a 30-day month, against the Free plan's 1,000 credits on the pricing page. Left as it is, the schedule nobody opens spends credits on empty checks, and Make sells the credits; one check each morning finds the same articles. Buffer's side has its own version: it bills per connected channel, so every extra network you ask the agent to write for adds $12 a month on Team (pricing file). Start with the two networks that send readers to your articles.

How to know it worked

The measure is sessions from social to each article in its first 14 days, read in Traffic acquisition by session campaign, which the tracked links fill with the article's slug (campaign URLs). Your four-article number from before is the baseline.

Then a holdout, because a good month for the blog flatters any method. Of the next eight articles, send the odd ones through the scenario and give the even ones the old single post, with a tracked link, then compare median sessions per article after 14 days. Inside Buffer, the fixed tag lets an assistant on its MCP server total engagement for the agent's posts with get_aggregated_post_metrics (MCP reference).

Watch the approver too. My rule of thumb: if more than two of seven drafts need a rewrite rather than an edit, fix the Instructions before the next article. Rewriting is the work this setup was meant to remove, so a high rewrite count means the brief is unclear, not that the approver is slow.

FAQ

Can the agent publish without the approval step?

Not on Team with a Needs Approval key: Buffer saves every post that teammate creates as a draft awaiting approval, through the API as well, until an approver moves it (approval article). On Free or Essentials the guard is saveToDraft: true, so check that flag whenever the scenario changes.

Why not let the agent call Buffer itself?

It could. Make agents accept MCP servers as tools (agent guide), and Buffer's create_post offers a share-now mode (MCP reference). A fixed request after the agent keeps publishing outside the model's choices, and Make's guide notes that tools raise token usage.

Does this work for Instagram or TikTok?

Not as written. Buffer requires an image or a video for both (MCP reference), and you host that media yourself, per its media guide. Text-first networks suit an agent working from an article. To keep the job inside one tool, how to use ContentStudio turns a feed into queued posts, and Planable vs Buffer compares approval set-ups.