Which story will go viral tomorrow?
This is the question that keeps editors awake at night. A story that goes viral can drive massive traffic, define the news cycle, and create industry-leading coverage. But predicting virality is notoriously difficult—stories blow up unexpectedly, and favorites fizzle out.
Tagtaly's virality score attempts to solve this problem. By analyzing hundreds of articles daily across multiple dimensions, we identify stories with the highest potential to explode on social media and dominate news cycles.
But how does it work? What is a "virality score," and can you actually predict viral potential? Let's dive in.
What is a Virality Score?
A virality score is a numerical rating (0-20) that predicts how likely a story is to go viral. High scores = higher viral potential. Low scores = stories that likely won't trend.
Think of it like a weather forecast for news. Just as meteorologists can't predict weather with 100% accuracy but provide useful guidance, virality scores predict potential, not certainty. A score of 18/20 means "this story has strong viral signals" not "this will definitely go viral."
Understanding the Virality Scale
The Five Virality Detection Algorithms
Tagtaly's virality score combines five independent algorithms. Each detects a different viral signal. The more algorithms that trigger, the higher the overall virality score.
Algorithm #1: Topic Surge Detection
The Surge Algorithm
Question: Is this topic suddenly getting way more coverage than normal?
How it works: Compares today's article volume for a topic against last week's average. A dramatic increase signals a breaking story or emerging trend.
Calculation:
- Today's Politics articles: 85
- Last week's daily Politics average: 32
- Week-over-week change: 165% surge ← VERY HIGH
Real example: Election announcement causes Politics surge from 30 articles/day to 120 articles/day. That's a 300% spike—extremely high virality signal.
What it detects: Breaking news, emerging crises, major announcements, scandals
Algorithm #2: Political Mention Tracking
The Political Algorithm
Question: Are major political figures getting unusual mention spikes?
How it works: Tracks mentions of key politicians (Prime Minister, President, major party leaders). Sudden spikes in mentions correlate with high-interest stories.
Example signals:
- PM mentioned in 45 articles today (normally 8-12) ← Major announcement or scandal
- Opposition leader mentions spike 10x ← Major criticism or challenge
What it detects: Political scandals, major policy announcements, political conflicts
Algorithm #3: Record Number Detection
The Record Algorithm
Question: Does the story contain record-breaking claims?
How it works: Scans headlines for keywords like "highest," "lowest," "record," "unprecedented," "for the first time." These phrases often indicate notable stories.
Examples that trigger this algorithm:
- "Unemployment hits highest level in 20 years"
- "Record-breaking heat wave sweeps across UK"
- "Lowest interest rates since records began"
- "Unprecedented number of asylum applications"
Why it works: Superlatives ("record," "highest," "lowest") indicate novelty and significance, which drives social media sharing.
Algorithm #4: Sentiment Shift Detection
The Emotion Algorithm
Question: Has the emotional tone of coverage changed dramatically?
How it works: Tracks sentiment swings in a topic. Rapid mood shifts indicate significant developments.
Example scenario:
- Yesterday: Politics sentiment +0.2 (neutral)
- Today: Politics sentiment -0.7 (very negative)
- Change: -0.9 point drop ← MAJOR SHIFT
This kind of sentiment collapse signals a major crisis or scandal breaking.
What it detects: Crises, scandals, surprising developments, major policy failures
Algorithm #5: Media Bias Tracking
The Outlet Algorithm
Question: Are different outlets covering the story very differently?
How it works: Compares coverage patterns across outlets (BBC, Guardian, Sky, Independent, Washington Post). When outlets diverge significantly on a story, it often indicates controversy or conflicting viewpoints—both of which drive engagement.
Example scenario:
- BBC: 2 articles on topic (neutral coverage)
- Sky: 12 articles on topic (intensive coverage)
- Guardian: 8 articles on topic (critical coverage)
- Disparity: 6x coverage difference
High disparity suggests the story is controversial or newsworthy enough for varied interpretation.
What it detects: Polarizing topics, media narratives, stories with multiple angles, controversial policies
How the Five Algorithms Combine: A Real Example
Let's analyze a real story and watch the virality score develop:
Scenario: A major political scandal breaks involving the PM.
Algorithm signals:
- Surge: Politics articles jump from 35/day to 120/day (+242%). Score: 4/4
- Political Mentions: PM mentioned in 68 articles (normal: 10). Score: 4/4
- Records: Headline says "Biggest scandal in 30 years." Score: 4/4
- Sentiment Shift: Sentiment drops from +0.3 to -0.8 overnight. Score: 4/4
- Media Bias: BBC: 3 articles, Guardian: 15 articles, Sky: 12 articles. Huge disparity. Score: 4/4
Combined Virality Score: 20/20 (MAXIMUM)
Editorial Decision: This story is almost guaranteed to dominate news cycles. Assign multiple reporters, prepare for massive traffic, update coverage throughout the day.
Limitations of Virality Scoring
1. Prediction ≠ Guarantee
A high virality score means "strong signals," not "will definitely go viral." External factors (competing stories, celebrity news, natural disasters) can overshadow even high-scoring stories.
2. Viral ≠ Important
Viral stories aren't always the most important stories. A celebrity scandal might have a virality score of 18/20 but be less significant than a quiet policy change with a score of 6/20.
3. Social Media ≠ Real World
What goes viral on social media isn't always what matters in the real world. Tagtaly predicts social media virality, not real-world significance.
4. Algorithm Blindness
Tagtaly can't detect novel situations it hasn't seen before. Completely unprecedented events might not trigger all algorithms.
Best Practices for Using Virality Scores
Practice #1: Monitor Score Trends, Not Individual Scores
A story with 8/20 virality today but trending upward (8 → 12 → 16 over 3 hours) is more interesting than a story that hits 15/20 and drops immediately.
Practice #2: Cross-Reference with Other Metrics
Combine virality score with sentiment and volume data:
- High virality + negative sentiment = crisis or scandal (assign crisis coverage)
- High virality + positive sentiment = celebration or milestone (opportunity for feature angle)
- High virality + high volume + low sentiment shift = stable crisis (routine coverage)
Practice #3: Look for Algorithmic Consensus
If all 5 algorithms trigger simultaneously, confidence is very high. If only 1-2 algorithms trigger, treat with more caution.
Practice #4: Verify Manually
For stories with virality scores 15+, read 3-5 actual articles before committing significant resources. Make sure the algorithmic signals match reality.
When Virality Scores Fail
Example 1: The False Alarm
A topic surge hits 250% (massive signal), but only because it's a holiday week with unusual posting patterns. Manual review reveals the surge is noise, not news.
Lesson: Always verify spikes with manual reading.
Example 2: The Sleeper Story
A story about a policy change has low virality score (6/20) but turns out to have massive real-world impact. Social media ignored it, but experts took it seriously.
Lesson: Don't neglect low-virality stories that align with your audience's interests.
Example 3: The Predictability Trap
If every outlet covers the same high-virality story, you're just following the crowd. The best coverage comes from finding angles others missed—often in lower-virality stories.
Lesson: Use virality scores for efficiency, not cookie-cutter journalism.
Advanced: Virality Over Time
The most powerful use of virality scoring is tracking how potential develops over hours and days:
- Hour 1: 6/20 virality (early development)
- Hour 4: 12/20 virality (gathering momentum)
- Hour 8: 18/20 virality (explosion imminent)
- Hour 24: 10/20 virality (past peak, declining interest)
This progression tells you when to commit resources and when to pivot coverage.
Key Takeaways
- Virality score (0-20) predicts social media viral potential
- Combines 5 independent algorithms: Surge, Politics, Records, Sentiment Shift, Media Bias
- High scores = strong signals, but not guaranteed virality
- Use to guide prioritization, not editorial judgment
- High virality ≠ high importance; always verify manually
- Monitor trends, not individual scores
Next Steps
Ready to apply this knowledge? Read:
- How to Read the Dashboard — See virality scores in action
- Using Tagtaly in Your Newsroom — Practical workflow integration
- Understanding Sentiment Analysis — How to combine sentiment with virality