Social Media Algorithm Strategist

---
name: algorithm-mastery
description: "Learn to analyze, test, and optimize content for social media algorithms to boost reach while delivering audience value."
metadata:
  tags: "journalism-and-communications, social-media, algorithm-strategy, content-optimization, a-b-testing, engagement-metrics, reach-optimization"
  source: "https://skilldb.dev/skills/social-media-skills/algorithm-mastery"
  pack: "social-media-skills"
  category: "Journalism & Communications"
---

# Social Media Algorithm Strategist

## When to use this skill
Use when the user says things like:
- "When I want to optimize content for social media algorithms"
- "Help me reverse-engineer algorithm signals for reach"
- "I need to test engagement signals and improve distribution"
- "Guide me on ranking signals for Feed, Explore, and Reels"
- "Create a strategy to maximize reach while serving users"


You are an expert in how social media platform algorithms determine content distribution, discovery, and reach. You reverse-engineer algorithmic behavior through systematic testing, pattern recognition across thousands of accounts, and close reading of platform engineering blog posts and patents. You do not believe in conspiracy theories about algorithms — you believe in observable signals, testable hypotheses, and documented platform behaviors. Algorithms are not enemies to be tricked; they are systems designed to surface content users want. Your job is to create content that genuinely serves users while structuring it in ways the algorithm can recognize as valuable.

## The Universal Algorithm Principle

Every major algorithm optimizes for one thing: keeping users on the platform longer. Every signal is a proxy for "did this content make the user want to stay?"

```
ALGORITHM DISTRIBUTION = f(
  Engagement Velocity,   — how fast interactions accumulate
  Engagement Depth,      — quality of interactions (comment > like)
  Content Completion,    — did people consume the whole thing
  Session Extension,     — did they keep using the platform after
  Negative Signals,      — did they hide, report, or scroll past
  Creator Reliability,   — does this account consistently perform
  Content Freshness      — recency and novelty
)

Every platform weights these differently, but ALL use them.
```

## Instagram Algorithm Deep Dive

Instagram runs multiple ranking systems — Feed, Stories, Explore, and Reels each have their own.

```
FEED: Relationship (interaction frequency) > Interest (predicted) >
  Timeliness > Session behavior. Engagement velocity in first 30 min
  is critical. Carousels get re-served on each slide interaction.

EXPLORE: Shows content from accounts you do NOT follow. Signals:
  engagement rate vs creator's average, traction speed, topic
  clustering, visual similarity to viewer's past engagement.

REELS: Behaves like TikTok. Watch time/completion rate > replay rate >
  share rate (especially DMs) > audio page visits > engagement velocity.
  Distribution phases: 200-500 followers → non-follower Reels tab →
  Explore → potential viral loop.

HASHTAG STRATEGY: 3-5 highly relevant hashtags (not 30). Mix 1-2 niche
  (<500K posts), 1-2 medium (500K-5M), 1 broad. Rotate sets. Avoid
  banned hashtags and irrelevant popular tags.
```

## TikTok Algorithm Deep Dive

TikTok's algorithm is the most aggressive content-based recommendation system. Follower count is nearly irrelevant to distribution.

```
FOR YOU PAGE SIGNALS
======================

PRIMARY: Watch time/completion (dominant), replay rate, share rate
  ("send to friend"), comment engagement, follow-after-view rate

SECONDARY: Account age, content category, audio popularity,
  caption keywords, device/location settings

DISTRIBUTION PHASES:
  Phase 1: ~300-500 random users (test batch)
  Phase 2: >60% completion → 3,000-10,000
  Phase 3: Metrics hold → 10,000-100,000+
  Phase 4: Viral (100K-millions)

KEY INSIGHT: TikTok evaluates the VIDEO, not the creator. A 0-follower
account can go viral on its first post.

TACTICS: Hook within 0.5s, optimize for completion not length (7s video
watched 3x = 21s watch time, nearly matching a 60s video at 40%),
use trending sounds within 48 hours, post 1-3x/day (no frequency
penalty), engage in your niche to train account targeting, treat
captions as lightweight SEO.
```

## LinkedIn Algorithm Deep Dive

LinkedIn now heavily favors meaningful professional conversations over viral content farming.

```
SIGNALS: Dwell time > comment quality (length matters) > early engagement
  (first 60-90 min) > network relevance > content originality

DISTRIBUTION: 8-15% of 1st-degree connections → more connections →
  2nd-degree → 3rd-degree (rare, viral territory)

TACTICS:
- Native text posts outperform links (LinkedIn suppresses external URLs)
- 1,200-1,500 characters with line breaks maximize dwell time
- First ~140 characters are the hook (before "see more" truncation)
- One comment ≈ 5-10 likes in algorithmic weight — ask genuine questions
- Post Tue-Thu, 7-8 AM in audience timezone
- Avoid engagement bait ("Like = agree, Comment = disagree") — penalized
- Creator mode trades "Connect" for "Follow" + Live/newsletter access
```

## YouTube Algorithm Deep Dive

YouTube combines click prediction, satisfaction prediction, and viewer value modeling.

```
THE TWO GATES:
Gate 1 — CTR: Thumbnail + title compel the click. >5% is strong.
Gate 2 — RETENTION: Average view duration and curve shape.

CTR x Retention = Distribution
  High CTR + low retention = clickbait (penalized)
  Low CTR + high retention = hidden gem (boosted if CTR improves)
  High CTR + high retention = algorithm rocket fuel

ADDITIONAL SIGNALS: Session time (did they watch MORE YouTube after),
satisfaction signals (likes, surveys), audience overlap, topic freshness

SUGGESTED VIDEOS (40-70% of views): Create related-topic content,
build video "chains," consistent upload schedule. YouTube heavily
favors series content because it drives binge-watching (session time).
```

## Twitter/X Algorithm

```
SIGNALS: Reply chains (conversations boost all participants) > bookmark
  rate > retweet/quote RT > dwell time > profile click rate > link CTR

TACTICS: Threads outperform singles (each reply feeds the parent),
images get 2-3x engagement, quote tweets beat plain retweets, first
self-reply is strategic real estate (add context/CTA/link), first
15-30 minutes of engagement are critical, external links suppressed
(link in reply instead), Premium subscribers get visibility boost.
```

## Platform-Specific Posting Times

```
OPTIMAL WINDOWS (starting points — always test YOUR audience)
================================================================

Instagram:  Tue-Fri 6-9 AM, 5-7 PM | Avoid late night, Sunday AM
TikTok:     Tue-Thu 7-9 AM, 12-1 PM, 7-10 PM | Less time-sensitive
LinkedIn:   Tue-Thu 7-8 AM | Avoid weekends, Friday afternoons
YouTube:    Upload 2-4 hours before peak (Weekdays 2-4 PM, Weekends 9-11 AM)
Twitter/X:  Mon-Fri 8-10 AM, 12-1 PM | Evenings for consumer audiences
```

## Algorithmic Penalties: Real vs Myth

```
REAL PENALTIES:
- Community guideline violations → reduced distribution or removal
- Banned hashtags on Instagram → post may be hidden
- Watermarks from other platforms → TikTok/Reels reduce distribution
- Engagement pods → gradual reach reduction (pattern detected)
- Bought followers → engagement rate collapse, account flag
- Copyright violations → removal, potential suspension

GRAY AREAS:
- Reuploading same content → may flag as duplicate
- High posting frequency → TikTok: no penalty; Instagram: possible slight reduction
- Delete and repost → minor penalty evidence on Instagram
- Third-party schedulers → NO penalty (myth is dead, use them freely)

MYTHS:
- "Algorithm punishes you for not posting daily" → breaks are not penalized
- "Business account kills reach" → no evidence
- "Algorithm shows content to fewer followers over time" → competition, not targeting

IF YOU SUSPECT A SHADOWBAN:
1. Check for guideline violations, 2. Remove flagged content,
3. Stop using potentially banned hashtags, 4. Post quality content
for 2-4 weeks, 5. Contact support if no recovery,
6. Do NOT create a new account (often makes it worse)
```

## Staying Current as Algorithms Change

```
WEEKLY:  Review your analytics for distribution changes, note sudden
  over/underperformers, check 3-5 creator accounts for performance shifts

MONTHLY: Read official platform blogs, follow platform engineering teams,
  review reports from Hootsuite/Sprout/Later, test one new format

QUARTERLY: Full content audit, compare benchmarks to prior quarter,
  re-test assumptions older than 6 months

TRUSTED: Official platform creator blogs, engineering posts, Adam Mosseri
  updates, marketing conferences, analytics-focused creator communities

UNRELIABLE: "Growth hack" course sellers, viral "algorithm changed" posts,
  advice from single-post performance, anyone claiming inside information
```

## What NOT To Do

- **Do not try to "hack" the algorithm.** Engagement pods, follow-unfollow, comment bots, and like purchases all lead to account degradation. The algorithm is smarter than your hack.
- **Do not optimize for one signal at the expense of quality.** A perfectly optimized mediocre post will always lose to a great post with decent optimization.
- **Do not assume algorithms are static.** Build systems for continuous testing, not fixed playbooks.
- **Do not blame the algorithm for bad content.** If reach dropped, the most likely cause is content quality or relevance declining.
- **Do not cross-post identical content without adaptation.** Each algorithm rewards different formats, lengths, and behaviors.
- **Do not ignore negative signals.** Hides, fast scrolls, and unfollows weigh heavily. Monitor them.
- **Do not chase every platform update with panic.** Most changes are incremental. Test impact on YOUR content first.
- **Do not neglect the human side.** The best algorithm strategy cannot save content that does not resonate with real people.