Social Media Growth skill

Social Media Growth

Operating guide for organic growth across LinkedIn, X (formerly Twitter), Instagram, TikTok, and YouTube Shorts. Bias toward what you can measure and reproduce, not algorithm folklore.

Sibling skills: paid amplification and channel strategy → marketing-analytics; founder-led pipeline and DMs-to-deals → business-development; editorial systems and repurposing → content-strategy; 1:1 outbound (cold DMs/email) → cold-outreach; creator partnerships → influencer-marketing; sponsored-content disclosure law → affiliate-marketing; positioning and voice → brand-strategy.

Read this first: algorithms change, so verify

Platform ranking systems are opaque, A/B-tested per cohort, and changed often. Most "the algorithm rewards X" advice online is inferred from creator anecdotes, not confirmed, and what worked last quarter may be neutralized today. Treat every tactic below as a hypothesis to validate on your own account, not a law.

Two things are genuinely public and worth grounding on:

• X open-sourced a ranking snapshot (the-algorithm on GitHub, 2023) — directionally useful, now years stale; do not quote it as current.
• Platforms publish creator/transparency docs (LinkedIn Engineering blog, TikTok "For You" explainer, Instagram "Ranking" posts by Adam Mosseri, YouTube Creator Insider). These are marketing-flavored but are the closest thing to primary sources. Cite the doc + the date you read it when you make a claim to a client.

Everything else: measure it.

Verification workflow (run this instead of trusting lore)

1. Establish a baseline. Pull 60–90 days of native analytics (LinkedIn → Analytics; X → Premium Analytics / per-post; Instagram → Professional Dashboard + per-post Insights; TikTok → Creator tools → Analytics; YouTube → Studio). Record per-post: impressions, reach, the platform's "watch time"/"dwell" proxy, saves/bookmarks, shares/sends, profile visits, follows-from-post, link clicks. Normalize engagement as a rate (per impression), never raw counts.
2. Change ONE variable per test. Hook style, format, length, posting time, link placement, CTA — one at a time. Mixing variables makes results uninterpretable.
3. Use a meaningful sample. Single-post wins are noise. Compare ≥7–10 posts per variant, ideally over ≥2 weeks, before concluding. Watch the median, not the one viral outlier (which is usually exogenous, e.g. a large account shared you).
4. Separate correlation from cause. "Posts with links got less reach" might mean links suppress reach — or that your link posts are promotional and just less interesting. Hold content type constant when testing a mechanic.
5. Re-test quarterly. A tactic that stops working is signal, not failure. Date your playbook ("times/formats validated 2026-Q2") so stale conclusions get retired.
6. Beware survivorship bias. "Creator X does Y and blew up" ignores thousands who did Y and didn't. Prefer your own A/B data over screenshots from growth gurus.

The benchmarks and "what the algorithm rewards" lists in this skill are rules of thumb gathered from creator reports, not measured constants. They vary enormously by account size, niche, paid amplification, and the platform's current model. Use them as starting hypotheses; replace them with your own measured numbers as soon as you have a baseline.

Platform playbooks

LinkedIn

Signals creators report it rewards (hypotheses — validate per account):

• Dwell time — people stop scrolling to read (favor scannable, line-broken text).
• Substantive comments (a thread of multi-word replies beats a pile of one-word ones).
• Shares/sends to DMs (private redistribution).
• Early engagement velocity (the first ~60–90 min after posting).
• Relevance to your stated topics and to the commenter's network.

Format performance (relative, directional):

Posting practice:

• ~3–5 posts/week. Past that, returns flatten and each post competes with your own previous one.
• Common reported sweet spot: weekday mornings in your audience's timezone — but verify with your own per-post data; B2B audiences and timezones differ.
• Earn the click: lead with a strong first 1–2 lines (before the "…see more" fold). Don't bury the value.
• A genuine question at the end can lift comments — only if it's relevant, not a tacked-on "Agree?".
• Reply to your own comments early to keep the conversation (and dwell) alive.

The "link in the first comment" question. Many creators believe external links in the post body suppress reach and move links to the first comment. This is a widely repeated hypothesis, not a confirmed rule, LinkedIn has publicly pushed back on it, and behavior shifts over time and by account. Don't state it to a client as fact. Test it: post 8–10 link-in-body vs. 8–10 link-in-comment posts with comparable content, compare median reach and actual link clicks (a buried link can tank clicks even if impressions hold). Do what your data says. Either way, an unannotated bare link reads as spammy — give context.

X (formerly Twitter)

Naming: the platform is X; legacy "Twitter/tweet" terms persist in search and habit, so keep them as aliases when helping users find things, but use "X"/"post" as the primary term.

On algorithm signals: X is the one major platform that open-sourced a ranking snapshot (2023). It referenced features like reply/engagement weighting and author reputation — but it's now years out of date and was only a partial view. Treat specific weightings (e.g. "bookmarks count Nx a like") as unverified creator lore, not measured fact. What's safe to say: replies and reposts (especially from established accounts), watch/dwell time on long posts, and profile visits all plausibly correlate with distribution — then confirm with your own analytics.

Growth practice (with guardrails — see Policy section):

• Reply with added value to relevant larger accounts early in their post's life. This is participation, not spam — generic "Great post! 🔥" replies get filtered and hurt your reputation.
• A consistent set of accounts you genuinely converse with compounds over time. Keep it authentic; do NOT run a reciprocal engagement pod (coordinated like/reply rings) — platforms treat that as platform manipulation and demote or restrict participants.
• Test posting times against your own analytics; don't assume "8am/12pm."
• Pin your strongest thread; iterate the hook on weak performers and repost.
• 1–2 hashtags max if any; stuffing reads as spam.

Instagram

Reported ranking priority (directional — Instagram has publicly said there is no single "algorithm" but per-surface ranking):

• For reach/discovery: Reels generally surface widest, then carousels, then static, then Stories (Stories skew to existing followers).
• Sends/shares and saves are reported to carry more weight than likes for discovery.
• Watch time / completion on Reels.
• For home feed among followers: recency, relationship, and predicted interest.

Mosseri has repeatedly stated follower count is not a ranking factor and that they actively surface small accounts — another reason not to chase vanity follows.

Reel practice:

• Earn attention in the first ~1–2 seconds (visual or verbal pattern interrupt).
• Many short Reels land ~7–30s; longer can work if retention holds — let retention graphs decide, not a fixed number.
• Add captions/text overlays (a large share watch muted).
• Original audio and trending audio both have uses; trending audio's reach benefit is overstated and fleeting — don't force an irrelevant sound.
• A clean loop can lift rewatch/watch-time; don't sacrifice clarity for it.

TikTok

Correction to a common overclaim: TikTok's For You ranking weights content engagement heavily, which is why new/small accounts can reach widely — but it is not "pure content quality, followers don't matter." Distribution is also shaped by account standing/health, prior video performance and niche history, your follower graph (a base of fans seeds early signal), viewer location/language, and content-moderation status. A flagged or repeatedly-reported account, or one posting in a saturated niche, will not get the same push regardless of a single video's quality.

Signals it reports weighting (validate per account):

• Early performance on the first audience batch shapes whether it's pushed wider.
• Watch time, and especially rewatch/completion rate on short videos.
• Comment velocity early on; shares/sends off-platform.

Format practice:

• Short videos that fully complete tend to do best; pace for retention, not a fixed runtime.
• Hook in ~1 second.
• Native, authentic production usually outperforms ad-polished video.
• Video replies to comments can extend a video's life.
• 1–3 posts/day in an active growth phase — only if you can hold quality; volume without quality trains the system that your content underperforms.

YouTube Shorts (and the long-form bridge)

Shorts and long-form are ranked differently. Click-through rate (thumbnail/title) × average view duration drives long-form; Shorts lean on swipe-away rate and completion. Use Shorts for reach and to feed subscribers into long-form (which monetizes and ranks in search far better, and compounds over years). Pin a "watch next" long-form video in Short descriptions. Track in YouTube Studio (Reach → impressions CTR; Engagement → average view duration / % viewed).

Viral content mechanics

Hooks that earn the next second:

Avoid clickbait that under-delivers — a hook the payload doesn't honor spikes a click then tanks watch time/dwell (the metric that actually matters) and erodes trust. The hook is a promise; keep it.

Post anatomy: Hook (earn attention) → Setup (why care / stakes) → Payload (the insight, story, or framework — the actual value) → CTA (one clear ask: follow, save, share, reply, or click). One CTA, not five.

Content calendar

Weekly template (B2B SaaS, multi-platform):

Repurpose one core idea across formats (long-form → thread → carousel → Reel/Short) rather than inventing five unrelated posts; see content-strategy for a repurposing pipeline.

Use-case playbooks

Different goals demand different playbooks. Pick one; don't run all five at once.

B2B founder-led (LinkedIn + X). Goal: pipeline, not vanity reach. Post 3–5x/week from the founder's profile (people follow people). Mix: ~40% point-of-view/opinion on your category, ~30% customer problems & outcomes (no names without consent), ~20% build-in-public/lessons, ~10% soft product. CTA is a conversation, not a demo link — move warm commenters to DMs and qualify there (hand off to business-development). North-star metric: qualified DMs and meetings booked, not followers.

Local services (Instagram + Google). Goal: nearby customers who buy. Geo-tag every post/Reel, use neighborhood/city hashtags, show real work (before/after, process, staff, premises). Reels for reach, Stories for trust/availability, a pinned highlight for hours/booking. Funnel to a booking link in bio. Pair tightly with local-seo (Google Business Profile, reviews, NAP consistency) — that often outperforms social for local intent.

Creator commerce (TikTok + Instagram Reels). Goal: sell product/affiliate. Lead with demonstration and transformation, not specs. Use the platform's native commerce surfaces where allowed (Shopping/affiliate tools). Every paid or affiliate post must carry a clear disclosure (#ad / "paid partnership" label) per FTC and the relevant national rules — see affiliate-marketing for the legal specifics. Track to actual sales/clicks, not views.

SaaS community-led (X + a community home). Goal: a moat of engaged users. Treat the social account as the top of funnel into a real community (Discord/Slack/forum) you own. Run a recurring ritual (weekly Space/live/AMA) on ONE topic. Surface and credit community members' wins. North-star: weekly active community members and content they create, not your post count.

Community building

Engagement-first, first 90 days:

1. Identify ~50 accounts in your niche (mix of sizes).
2. Engage genuinely on their content most days — substantive comments, not just likes.
3. Reach out 1:1 only with specific, relevant value and only where it's welcome — never a copy-paste blast (see Policy + cold-outreach for consent-respecting DM/email practice).
4. Create content that credits and amplifies community members — quote/feature their wins (with permission), build a recurring "best of the community" post, turn great replies into their own posts. Reciprocity, done in public and authentically, compounds.
5. Host a weekly Space/live/room on one topic to convert followers into a community.

Flywheel: add value to others → they engage with you → the system reads the engagement → more reach → more community → repeat. The fuel is genuine participation; manufactured engagement (see below) breaks the wheel.

Platform-policy & brand-safety guardrails

Growth tactics that cross policy lines get accounts shadow-limited, restricted, or banned, and create brand/legal risk. Do not do these — and flag them if a client asks:

Regulated industries (finance, health, crypto, legal, alcohol, gambling, supplements): claims, testimonials, and "results" are subject to sector rules (e.g. SEC/FINRA, FDA, FTC, MiCA, and national advertising regulators). Route this content through compliance/legal review before posting. Don't promise returns, cures, or outcomes.

Authenticity & AI disclosure: label AI-generated or significantly altered media where the platform requires it (several now mandate this), and don't impersonate or use undisclosed deepfakes.

Failure modes & when NOT to chase reach

• Reach suddenly flat / "shadow ban." Symptoms: impressions collapse across posts, you don't appear in hashtag/search results, replies hidden. Checks: look for a policy strike or account-status notice in settings; audit recent posts for flagged links/words/banned hashtags or recycled watermarked content; reduce posting frequency and remove anything borderline; on supported platforms request a review. Most "shadow bans" are either a real (often appealable) policy action or a content slump misread as suppression — confirm with native analytics before assuming malice.
• Content fatigue / audience burnout. Engagement decays as you repeat a format or theme. Rotate formats, retire tired hooks, and let your retention/saves data (not gut) tell you what's stale.
• Vanity-metric trap. Followers and impressions don't pay. Tie every channel to a downstream business metric (qualified conversations, signups, bookings, sales) — see marketing-analytics. A 2,000-follower account that drives 20 demos beats a 200k account that drives none.
• Brand & reputation risk. Edgy/contrarian hooks that win reach can also misfire publicly. Have a posting checklist and an escalation path; one viral mistake outlasts a hundred good posts.
• Platform concentration risk. A single algorithm change or ban can erase a channel overnight. Convert social audiences into owned channels (email, community) you control (email-sequence).
• When NOT to chase reach: highly regulated claims, sensitive/crisis moments, tiny-but-high-value B2B niches (depth and DMs beat broad reach), or when production cost per post exceeds its measured business return. Sometimes 20 right people in the DMs is the whole game.

Growth metrics

Track rates, compare to your own rolling baseline, and tie to a business outcome. The "benchmark" column below is a rough, uncited industry rule of thumb that varies wildly by account size, niche, and paid spend — use it only as a sanity check until you have your own baseline (Verification workflow above), then replace it.