Social platform algorithms and scheduling — evidence snapshot (2026-10-08)
Status: Researched, NOT proof of a connected account, live publishing capability, guaranteed reach, or a functioning best-time optimizer. Policies and interfaces may change. Recheck provider documentation when using or implementing.
Goal: Let a person produce truthful content in batches and arrange approved, account-specific posts for future delivery when their intended audience is likely to be active. Treat native app scheduling, third-party API access and ReachWeaver execution as three different capabilities. No platform provides a public universal formula for guaranteed reach.
Confidence and evidence taxonomy
Section titled “Confidence and evidence taxonomy”- Official: provider documentation, engineering description, or current product help. An official high-level factor is not a complete model or permanent ranking weight.
- Observed: aggregate timing association in an observational analytics dataset. Not proof that posting at that time causes incremental reach. Do not use as a guarantee or a substitute for an account’s audience insights.
- Hypothesis: operational recommendation or sensible creative principle to test per audience and format. Mark it as such.
Time convention: Timing suggestions below are audience-local, not necessarily the operator’s local time. Never schedule with a bare clock hour; keep IANA audience time zone, explicit date/offset and UTC, paying attention to DST. If the audience is split across Europe and the US, create separate target cohorts or alternate windows. Do not silently assume Europe/Rome is also the reader’s timezone.
Native scheduling / automation comparison
Section titled “Native scheduling / automation comparison”“Native automatic” means that after a user has scheduled a post in the platform’s own UI, the platform is expected to publish at the chosen time without the user being present. “RW execution” describes the existing ReachWeaver v0.2 adapters, NOT theoretical API availability.
| Platform / surface | Native auto scheduling? | Limits, eligibility and workflow | ReachWeaver v0.2 live capability | Suggested starting window, audience local | Research |
|---|---|---|---|---|---|
| LinkedIn member | Yes | Composer clock icon; 10 min–3 months; events, jobs, services excluded | Draft/manual; OAuth not connected | Weekdays 15:00–20:00, start Wed 16:00 | Official native, 2026 benchmark |
| LinkedIn Page | Yes | Admin Page; 1 hour–3 months; excludes multiple photos, reshares, polls, jobs, events/services | Draft/manual; permissions and Page scopes needed | Weekdays 15:00–20:00 (starting prior, not Page-specific evidence) | Official Page |
| Yes, eligible formats/accounts | In-app posts/Reels scheduling and Meta Business Suite for eligible professional profiles and supported media; Stories, interactive stickers, account type and horizon may differ; inspect the actual composer | Draft/manual; media pipeline + approved Meta permissions missing | Weekdays 18:00–21:00; test Thu 09:00 separately | Meta original content, observed timing | |
| Facebook Page | Yes | Meta Business Suite planner for authorized Pages; content type and feature restrictions; do not assume personal profile API scheduling | Draft/manual; Page scopes missing | Weekdays 08:00–11:00, start Thu 09:00 | Meta original content, observed timing |
| Threads | Yes, posts | In-app scheduler; reported horizon up to 75 days; replies cannot be scheduled; reverify current UI | Draft/manual; approved Meta app needed | Weekdays 06:00–11:00, start Thu 09:00 | Scheduling rollout, Meta topic evidence |
| X | Yes | Native web/X Pro or Media Studio/Ads composer where available; account/tool eligibility varies; no assumed free API quota | Draft/manual; paid/authorized API access not connected | Weekdays 06:00–11:00, start Tue 09:00 | X scheduling, ranking source |
| TikTok | Yes, eligible account | Web Video Scheduler for eligible Creator/Business; 15 min–10 days according to TikTok; API direct post has separate approval/audit requirements | Storyboard/manual; no uploaded video or API | Weekends 08:00–11:00, start Sun 09:00 | Official scheduler, ranking help |
| YouTube / Shorts | Yes | YouTube Studio set a video/Short as scheduled, choose date, time, timezone; account restrictions may block publication | Storyboard/manual; no upload/media adapter | Use own audience heatmap; no verified universal slot for long-term results | Studio scheduling, YouTube performance FAQ |
| Pinterest business | Yes | Standard Pins; 30 days ahead, 10 queued maximum, scheduled one at a time, business account | Draft/manual; approved API/media integration needed | No cross-account verified starting window; test analytics by format/search topic | Official Pinterest Help |
| Reddit ordinary community posting | No native general-purpose auto-scheduler | Moderators with Manage Posts & Comments may natively schedule posts in their own subreddits; native scheduler doesn’t support video/image posts. Never confuse mod-only with automatic self-promotion in unrelated communities | Manual-only; per-community policy review | Use subreddit active hours and moderator guidance, not a universal clock | Reddit moderator scheduling |
| Bluesky | No general native scheduler verified | Use a permissioned third-party scheduler or own authorized connector with persistent worker; its Following timeline is chronological, with separate custom feeds | Text-only adapter mock-tested; no live account connected; needs persistent worker | Weekend 16:00–19:00 and weekday evenings; benchmark revised over 2026 | Custom feeds, 2026 observed timing |
| Mastodon | Yes via official server API | POST /api/v1/statuses accepts scheduled_at, at least 5 minutes future; web/app UI support may vary by client/instance |
Text-only adapter mock-tested; current CLI schedules locally rather than using native scheduled_at |
Use instance/audience active hours, not a universal slot | Official API |
| Telegram channels | Yes in native user app | Native scheduled messages; ReachWeaver’s Bot API sendMessage adapter is distinct and must be invoked by a running local worker at due time |
Text-only bot adapter mock-tested; admin permission and persistent worker needed | Subscribers’ local active hours and notification tolerance | Telegram scheduled messages, Bot API |
| Product Hunt launches | Yes, for product launches | Schedule launch day within 30 days; site runs a Pacific Time daily cycle, not arbitrary social post slots; check latest launch scheduling UI | Not in 12-platform publishing catalogue; manual launch planning only | Choose staffed launch day; site’s cycle resets at Pacific midnight | Official scheduling, PH scheduling changes |
| Hacker News | No; do not automate submissions | Guidelines explicitly advise against automating posting and against primarily promotional submissions or generated HN post text | Manual-only, human-written submission | Human chooses genuine community participation; no machine-picked launch guarantee | HN guidelines |
Caveat: Existing ReachWeaver “schedule” means local outbox holds a UTC due timestamp; publish is a dry run until both explicit live flag and authorization. A trusted persistent worker (for example one configured cron on a private host) must execute due work; the GitHub CI runner is not such a worker. No user social account has been connected or real-world tested. LinkedIn, Instagram, Facebook, Threads, X, TikTok, YouTube, Pinterest and Reddit are not currently published by ReachWeaver.
Additional developer and regional distribution channels
Section titled “Additional developer and regional distribution channels”This expands the priority-channel inventory without pretending unknown API capabilities have been verified. The prior list should never classify DEV.to as lacking a scheduler: its own help now confirms one.
| Platform | Native scheduling evidence | ReachWeaver state | Audience/timing recommendation |
|---|---|---|---|
| DEV Community (DEV.to) | Yes; official scheduling help describes the editor hexagon → Schedule Publication, using the user’s local timezone. | Original Markdown article draft; manual/native handoff, no authorized publishing API integration | Developer-language readership: test weekday local business hours; measure article views, saves, meaningful comments and actual repo referrals |
| Hashnode | Yes, with plan caveat; June 2026 Hashnode Pro announcement lists scheduled publishing as a Pro feature. Older 2022 posts described wider scheduling, so verify the account’s current entitlements. | Original Markdown article draft; manual/native handoff, no authorized Pro API connection | Same audience-relevance and evergreen technical search considerations as DEV; no universal optimum hour established |
| Discord communities | No verified general-purpose native scheduled message feature from this research; developer bot tools can be permissioned but are separate integrations | Community-first/manual; no bot connector | Community rules, moderator permission and conversation timing, not bulk posting |
| Indie Hackers | Native post queue not verified; do not claim automatic scheduled publication | Manual/community participation | Helpful founder case studies; test relevant community activity |
| Peerlist | Native post scheduling not verified | Manual/community participation | Developer career/builder audience; measure relevant responses |
| Bilibili | Native scheduling/app eligibility not sufficiently verified from a current official primary source | Manual/localized media upload | Video viewers in China; schedule in China local time only after platform verification |
| Zhihu | Native scheduled post capability not verified from current official primary source | Manual/localized content | Audience in China; Chinese-language Q&A, forum rules and author relevance |
| OSCHINA | Native scheduled article capability not verified from current official primary source | Manual/localized technical content | China-local developer business hours as unverified exploratory hypothesis |
Important: Claims of “yes” above mean native UI functionality, not an available ReachWeaver API. calendar-plan may propose manual slots for DEV.to and Hashnode with --slot; --apply intentionally rejects them. Official sources are linked; unknown status requires in-account verification rather than guessing.
One-by-one algorithm and creative intelligence
Section titled “One-by-one algorithm and creative intelligence”| Platform | Publicly documented discovery/ranking signals | Practical hypothesis to test | Don’t claim |
|---|---|---|---|
| March 2026 feed engineering: topic/profile interest relevance, professional context, historical interaction sequences, freshness and trust | Professional-specific technical lessons, original diagrams, credible outcomes, substantive replies | That a chosen posting minute or a specific number of comments forces distribution | |
| Meta reports higher share of original recommended content; recommendations personalize by viewer interests | Native 9:16 demo with strong opening, clear captions, relevant sends/shares, original footage | “Reels always beat carousels” or secret ranking weights | |
| Original content prioritized over duplicate/minor edits in Feed/Reels | Original case study demo and genuine Page/community conversation | Guaranteed boost just for posting as Page at 9am | |
| Threads | Meta found posts with tagged topics associated with more views | Topic tags, helpful short insights, real back-and-forth | Causal boost from topic tags or auto-replies |
| X | Public 2026 For You code: in/out-of-network retrieval, predicted interaction probabilities, recency, relevance, visibility filtering and diversity | Concise findings, quality threads, genuine discussion, non-duplicative publishing | Raw weight ratios equal like/report exchange rates; public repo is complete immutable production configuration |
| TikTok | TikTok Help: user interactions, video content information, user information; weights vary | Show working result in first 2 seconds, structure for completion/rewatches, authentic voice | Universal completion threshold or timing hack |
| YouTube | YouTube Help: viewer satisfaction and absolute/relative watch time, content relevance; timing not known to affect long-term viewership | Precise searchable title/thumb, retention curve review, publish when subscribers are present for early engagement | Long-term reach is controlled by upload hour |
| Search/discovery context is important; no public precise ranking coefficient established by this research | Descriptive searchable titles, board relevance, original helpful visuals, evergreen queries | A precise day/hour “algorithm requirement” | |
| Community-specific eligibility, participation, votes and time-sensitive conversation; rules and sort choice differ | Carefully targeted useful technical walkthrough with transparent author affiliation, join replies | Universal subreddit “best time”, automated promotion loopholes | |
| Bluesky | User-controlled custom feeds, chronological Following feed and feed-specific ranking | Join relevant software/open-source/custom feed conversations rather than broadcast | Single algorithm on all feeds |
| Mastodon | Federated instance differences, follower timelines and hashtags; no universal opaque For You ranking imposed everywhere | Good tags, clear alt text and authentic replies/boosts | Worldwide optimum hour |
| Telegram | Mainly subscriber delivery, notifications and forwarding, rather than broad recommendation to strangers | Informative channel summaries, considerate frequency and subscriber timezone | A TikTok-style discovery algorithm |
| Product Hunt | Daily launch window and maker/product relevance; on-site and human engagement | Supported maker narrative, live support during launch day | Guaranteed homepage rank from midnight |
| Hacker News | Site submission and voting rules, time sensitivity and human moderation | Human-authored substantive launch story or technical article where genuinely interesting | Safe bot scheduling or guaranteed front-page position |
Best-time selection policy
Section titled “Best-time selection policy”See the standalone Claude skill .claude/skills/reachweaver/references/timing-intelligence.md for an auditable selection procedure. Apply in this order:
- No authorization -> no scheduling/publishing. Drafting can be offline. Actual platform-native scheduling or API publishing needs the human’s verified account and exact-content approval.
- Use first-party per-account active-viewer insights when available (e.g. YouTube Audience). For every channel, distinguish audience geography from account owner’s timezone. Recheck format/account differences.
- Use observed 2026 industry windows only as low-confidence warm-starts if insufficient own data. The Buffer study is based on Buffer customers, aggregated posts, and non-experimental differences. The complete latest public study is Buffer 52M+ posts; don’t present a time as empirically optimized for this specific brand.
- Propose multiple slots, each with audience-local date, IANA timezone, UTC, evidence source/checked date, format, reasoning, confidence, schedule transport and owner approval status. Separate exploratory slots from higher-priority audience activity slots.
- Respect operational rules: no prohibited automation, API capabilities, account post limits, unscheduled community posts or cross-brand mixing; honor holidays, sensitive announcements, review timelines, content embargoes, practical reply coverage, min post spacing and launch-day dependencies.
- Measure at matched ages: 24h, 7d and optionally 28d; collect platform-native denominator and qualified outcomes (profile visits, website sessions, stars/downloads) where accessible. Missing values are unknown. Look for persistent differences across reasonably similar content, not one viral outlier.
- Learn conservatively: update proposed slots after enough comparable account history, keep exploration to avoid optimizing solely around a noisy initial peak, log competing explanations and report uncertainty. Do not describe observational changes as proven causal improvement.
Practical 2-week batching model
Section titled “Practical 2-week batching model”Create 10 approved, platform-specific drafts at a convenient time; let the skill propose local-time target slots and a manual handoff for native schedulers. Use one target audience and one experiment per brand. For LinkedIn, schedule in its native UI; for YouTube and TikTok, complete the video upload through approved native interfaces; for Bluesky/Mastodon/Telegram use the already implemented text adapter only after credentials, user-approved test send, and a reliable persistent local worker are validated. Unsupported accounts stay drafts with a reminder/manual-export checklist, not phantom scheduled posts.
Do not use fake engagement, auto-generated comments, artificial cross-account coordination, repeated near-identical mass posts, scraping login sessions, or API restriction circumvention.
Research maintenance and technical gaps
Section titled “Research maintenance and technical gaps”- Last researched 2026-10-08; validate platform limits and APIs again at execution time and at least monthly. Source dates differ; do not conflate old blog announcements with tested 2026 UI.
- Algorithmic weights are not generally published; the notable exception of X public code still does not expose every serving factor and does not authorize manipulating recommendations.
- ReachWeaver now has a small offline
calendar-planCLI with transparent, conservative candidate windows, IANA timezone/DST validation, blackout and spacing checks and optional approved-text-only local queueing. It does not learn the optimal time from native analytics or guarantee improvement. Per-account heatmaps, automated analytics collection, a durable cloud worker, asset/media upload and official API integration for most networks are still future work; see calendar quickstart. - Related guides: platform matrix, architecture, roadmap, and skill references algorithms and timing.