Ranking signals and weightings
Read this when choosing where a post goes, how hard to work the window there, and which numbers to watch.
Treat every figure below as an order of magnitude, not a constant. Platforms retune continuously and none of them publish their weights; these come from operator testing and platform disclosure. The relative ordering is far more stable than any individual multiplier, and the ordering is what the decisions rest on. Re-verify before committing budget to a specific number.
Conversation-led text feeds
Short-form text and threads. Ranked primarily on interaction velocity in the first thirty to sixty minutes.
Relative weighting, with a like as the unit:
| Signal | Approximate weight | Operational consequence |
|---|---|---|
| Author replying to a comment on their own post | ~150x | Overwhelmingly the highest-leverage action. The author sits in the comments for the first hour and answers each one individually |
| Reshare with added commentary | ~20x | Worth more than a plain reshare. Ask people for an opinion, not a repost |
| Reply from another account | ~13.5–27x | Engineer the post so replying is necessary. Questions and stated positions outperform summaries |
| Save or bookmark | ~10–12x | The silent signal. Lists, data, and frameworks collect these |
| Profile click | ~12x | Follows from a distinct point of view rather than a distinct format |
| Like | 1x | The floor. A post with likes and no replies reads as low quality |
| Hide, mute, report | Strongly negative | Suppresses distribution outright. This is the yield on bought engagement |
Two documented penalties: an outbound link in the post body suppresses reach by roughly thirty to fifty percent, and stacking three or more hashtags costs around forty percent. Both are avoidable at zero cost. Verified accounts see distribution several times wider than unverified ones on the same content.
Best suited to opinion hooks, threads, live demos, and founder accounts. This surface has the tightest window of any, which makes the T+0 to T+15 block decisive rather than merely helpful.
Short-form video
Ranked first on completion rate, then on saves and shares. The first forty-eight hours determine the outcome, a longer window than text but a harder one to influence after the fact.
Save rate is the underrated accelerator: above roughly four percent is strong, above eight percent is exceptional. Completion is the gate before any of that matters, which puts the weight on the first two seconds rather than on the payoff.
Suits native meme formats, template challenges, and before-and-after demonstrations.
Saves-and-search feeds
Image-plus-text surfaces where the save-to-like ratio and comment interaction drive ranking, and search long-tail keeps delivering traffic for months afterward.
The only family where a post compounds after the launch window closes. That changes what is worth optimising: keyword coverage in the caption matters more than first-hour velocity, and a post that underperforms in week one can still be the best performer in month three. Do not judge these on window metrics.
Suits tutorials, checklists, and honest first-person notes.
Long-form video
Ranked on watch time, retention curve shape, click-through rate, and search. Suits full workflow walkthroughs, deep reviews, and tutorials.
Typically the strongest converting surface and the slowest to move. The launch window matters least here; title and thumbnail matter most, and a video can find its audience weeks after publication. Amplification decisions should use a longer read than six to eight hours.
Professional networks
Ranked on early first-degree engagement, comment depth, and dwell time. Outbound links in the body are demoted here as elsewhere.
Comment depth is doing more work than comment volume: a handful of long exchanges outperforms many short ones, which makes the author's replies the primary lever again. This is also the surface where the audience is legible enough to judge a post by who replied rather than how many did, which is the whole basis of the composition read in the parent skill.
Suits build-in-public milestones, customer cases, and founder narrative.
Cross-surface rules that hold everywhere
Link out of the body, into the first comment. Keep hashtags to two or fewer. Get the publishing account verified. Never edit the original after publish, which resets or penalises distribution on several surfaces. And never buy engagement: the specific pattern it produces, high likes with no replies and no saves, is the pattern every ranker on this list treats as evidence of low quality.