What this guide does: Helps choose where and how to publish, then evaluate whether distribution serves the customer’s business goal.
Use the account’s audience, available formats and measured results. Treat claims about algorithm weights, ideal slide counts or guaranteed reach as unverified unless a current source supports them. A platform average is context for a test; it does not establish what will work for this account.
Choose the destination
Match the content to a real audience and a useful next step. An explainer might answer a search question, demonstrate a product or help someone recognize a problem. Choose its channel from evidence about that audience and the customer’s existing distribution, rather than making every business start with social posts.
For a connected channel, inspect publishing status and account restrictions before preparing a schedule. A failed publish is different from a published post with low reach. Respect the destination’s current format requirements and the customer’s approved publishing scope.
Build for the actual format
For an Instagram or TikTok carousel, make the opening understandable and let each following slide advance the idea. Preview the full deck at its intended size. Use enough slides to deliver the promise; a fixed slide count is not a quality check or a guarantee of distribution.
For video, inspect the opening, pacing and delivered result using first seconds. A question, demonstration or story can each work when it matches the purpose and the customer’s voice.
For Pinterest, use the Pinterest guide. Prepare a standalone Pin with a relevant destination, board and search wording. Evaluate it with its own metrics and observation window rather than importing a reel’s retention target.
Decide what to measure
Record the intended outcome before publishing. Depending on the task, it can be qualified visits, an order, an account signup or a useful audience response. State how the content connects to that outcome and where attribution stops.
Read only metrics the connected tools actually provide. Keep views, reach, comments, saves and outbound clicks distinct. Platform engagement does not by itself establish visits or sales. A missing metric is unknown, not zero.
Compare content at similar ages and separate paid from organic distribution. Include the sample size and spread, since one unusual post can dominate an average. Avoid comparing different metric definitions as though they measured the same thing.
Choose a workable cadence
Use the customer’s objectives, production process and measured delivery to set a schedule. The saved posting schedule is a preference to follow, not evidence of an algorithmic advantage. Test timing when there is enough comparable activity to learn from it.
Prepare a repeatable content process and keep the queue visible. Requested volume and quality both matter; do not reduce the user’s ambition to a fixed weekly quota. Reply to relevant comments within the authorization the user has given. A research or publishing request does not authorize unrelated likes or outreach.
Learn from a result
When a post meets its intended outcome, identify what could explain it and test that explanation with another example. Reuse a format when the evidence supports it, while writing for the new topic and the account’s voice.
Keep each experiment’s hypothesis, publication references and results with the customer’s plan. If several things change at once, attribute the outcome to the whole post until further evidence separates them. Preserve unsuccessful examples as well as successful ones so later reviews can compare them.
Read the week’s posts once a week, or when the person asks how posts are doing. The thresholds below are editorial defaults for choosing follow-up drafts, not platform rules or tests of statistical significance. Adapt them to the account’s history and declared outcome; a reach outlier is a candidate for another example.
- Read per-post results from
get_marketing_review(social.publishes). A null metrics row is unmeasured, not zero. - Compare at the same age, seven days by default, because views keep arriving. A post’s breakout is its views at that age divided by the account’s median at that age over its last 20 measured posts on the same platform. Before 10 measured posts, treat that baseline as exploratory: report the available results without declaring a repeatable winner. Pins are read on their own clock (Pinterest guide).
- A post at 3x or more is a winner. Draft its part 2 within the week: the same hook template and slide skeleton (slide count, each slide’s role, where the call to action sits) with a new payload, meaning a new item, place, case or example. Duplicate the post and replace its content, or use “Make more like this” where the app offers it. Number the part where the account’s voice allows, and never repeat an earlier part’s payload.
- A part 2 above the median is a candidate for a series. Draft subsequent parts with new material. Two parts below the median trigger a review of topic, audience and delivery before continuing or stopping. Change the schedule only within the user’s authorized scope.
- A post at 10x or more also earns its own video on a video-default
destination (
update_post {outputs}, then the render quote and the person’s yes). 10x is a starting point; move it once the account has a history. - Report each winner with its breakout and a link, the drafts made and what waits for approval. Every draft stays a draft until the person approves it.
Breakout picks what to make next. Whether the channel brings customers is a question for link clicks and the app’s own outcomes, read in a marketing review.
Record a content test in Sprid so the app’s Reviews page tracks it. Change one
thing (the format, the message or hook, or the visual), state what it should
beat, and save it with update_review_hypotheses as an entry whose measure
has source: "posts", a short label and the ids of the posts made for the
test in postIds. Add each new test post to postIds as it is made. A test
has its own baseline, separate from the weekly breakout: Sprid reads each post
at ageDays old (7 by default) against the app’s other posts of the same
platform and kind from the last 90 days. The default metric is
saves and shares per reach, read on Instagram, where reach is reported; a test
with no Instagram posts reads views. Where the usual posts almost never get
saved or shared, the rate cannot tell formats apart: set metric: "views".
Put the number to beat in target, in the metric’s own unit.
When the test is read: if it held up, keep the change in the account’s next
posts, and give its winning posts a part 2 as above. If it did not hold,
propose the next thing to change as a new entry with follows set to this
one’s id, in the order format, then message, then visual, and drop or settle this one with its result. If too few posts were
old enough to read, keep posting under the same entry.
Recheck external claims when they matter
For a platform requirement or a claimed ranking mechanism, consult current first-party documentation and retain the source and date. Separate a platform’s published explanation from third-party speculation and from this customer’s measured results. If the distinction could change the recommendation and the claim cannot be verified, leave it out of the decision.