Skip to content
Thursday, January 9, 2025 · New York Edition
EN For Enterprise Sign In
↑ 01Federal Reserve signals March rate cut as inflation cools to 2.4% ↑ 02TikTok deal framework reportedly near completion, $80B valuation cited ↑ 03Nvidia's Blackwell Ultra chip benchmarks leak ahead of GTC keynote ↑ 04OpenAI closes $40B round at $300B, led by SoftBank and a16z ↑ 05EU AI Act enforcement begins; first compliance audits target Big Tech ↑ 06Instagram launches 90-second Reels, doubling short-form ceiling ↑ 01Federal Reserve signals March rate cut as inflation cools to 2.4% ↑ 02TikTok deal framework reportedly near completion, $80B valuation cited ↑ 03Nvidia's Blackwell Ultra chip benchmarks leak ahead of GTC keynote ↑ 04OpenAI closes $40B round at $300B, led by SoftBank and a16z ↑ 05EU AI Act enforcement begins; first compliance audits target Big Tech ↑ 06Instagram launches 90-second Reels, doubling short-form ceiling

Methodology · Updated November 2024

The Virality Score — How We Rank Trends

Every story on TopTrendingNews carries a number from 0 to 100. It is the output of a proprietary model trained on eleven years of cross-platform engagement data, refined daily by 38 senior journalists, and stress-tested against the moments that actually broke the internet. This page explains exactly how that number is computed — and why you can trust it.

By the TopTrendingNews Newsroom 14 min read Last reviewed: Q4 2024
Get the Daily Trending Brief See the Score Meter
11 yrs of cross-platform training data
47,000+ tracked sources, every 60 seconds
94% accuracy above 100K mentions / 24h
14 min median head start vs. legacy wires

The Artifact

A Score You Can Read at a Glance

The Virality Score collapses thousands of signals into a single 0–100 reading. The meter below shows where a story sits on that scale, the band it falls into, and how an editor should treat it. Every number you see on TopTrendingNews maps to one of these thresholds.

A horizontal Virality Score meter from 0 to 100, divided into five threshold bands — Quiet, Rising, Breakout, Mainstream, and Critical Mass — with an indicator needle reading 73.
Figure 1 — Threshold bands used across the site. A reading of 73 places a story in the Mainstream band, the moment it stops being optional reading for newsroom professionals.
0 – 20 Quiet Background noise. Monitored but not surfaced in any newsletter.
20 – 40 Rising Early signal on at least one platform. Worth a desk note.
40 – 60 Breakout Crossing two or more platforms. Eligible for the Daily Brief.
60 – 80 Mainstream Will dominate feeds within hours. Editors publish context.
80 – 100 Critical Mass Apex moment. Briefing subscribers within the next cycle.

The Framework

Three Pillars Behind Every Score

The Virality Score is not a single equation. It is a layered model in which quantitative momentum is checked against editorial judgment before a number is published. Each pillar below is reviewed quarterly and signed off by the editor on duty.

  1. Velocity

    How fast a story is accelerating across the public record.

    We measure the rate of change in mentions, shares, replies, and link-outs, normalized per platform so a TikTok surge and a Reddit thread can be compared on the same axis. Velocity alone decides whether a story enters the scoring pipeline at all — anything that isn't moving is filtered out at the source.

    • Per-platform baseline normalization across X, TikTok, Reddit, Instagram, YouTube, and 40,000+ news domains.
    • Rolling 15-minute, 1-hour, and 6-hour windows stitched into a single acceleration curve.
    • Bot and spam suppression layers trim amplification before the curve is computed.

  2. Cross-Platform Convergence

    Whether a story is breaking out of one silo into the general conversation.

    A story trending on a single platform is interesting. A story trending on several at once is the news. Convergence measures how many distinct audience communities are reacting on the same day, weighted by the editorial independence of those communities — a simultaneous move on legacy news domains and on social reads very differently from a coordinated push.

    • Independent-source weighting to discount coordinated amplification.
    • Audience-overlap penalties to prevent echo chambers from inflating scores.
    • Geographic spread signals, since a global story reads differently from a regional one.

  3. Editorial Verification

    A human layer that can override, downgrade, or hold a story.

    Before any score is published, a senior journalist verifies the underlying claim against primary sources, confirms the story is not a hoax or coordinated manipulation, and writes the contextual frame that goes with it. A high-scoring story that fails verification is held; a lower-scoring story with strong sourcing can be lifted. The score is the input, not the verdict.

    • 38 senior journalists rotate verification duties across New York, London, and Singapore.
    • Hold, lift, and override actions are logged in a public-facing methodology changelog.
    • The desk refuses to publish a score it cannot stand behind in a court of editorial reputation.
The Signal Behind the Score

47,000+ sources, refreshed every 60 seconds.

A trend is only a trend if the signal is broad, fast, and clean. Here is exactly what the Virality Score is listening to — platform by platform, minute by minute.

  • 01

    Platforms Tracked

    6core networks

    Real-time ingestion across X, TikTok, Reddit, Instagram, YouTube, and 40,000+ news domains — covering the social web and the editorial web in a single feed.

    • X (Twitter)
    • TikTok
    • Reddit
    • Instagram
    • YouTube
    • News domains
  • 02

    Total Sources

    47,000+digital outlets

    Every tracked domain is verified by our 24-person data engineering team for editorial legitimacy, audience size, and publishing cadence before it joins the index.

    • Major wire desks
    • National newspapers
    • Trade & vertical press
    • Independent newsletters
    • Verified social handles
  • 03

    Refresh Cadence

    60sper full sweep

    Every minute, the index re-ingests signals from the live web and re-ranks the leaderboard. That cadence powers our 14-minute median time-to-publish advantage over legacy wire services.

    • 60s full re-ingest
    • 5s anomaly spike check
    • 14-min median head start
    • Minute-level windows
  • 04

    Training Baseline

    11yrscross-platform data

    The Virality Score was trained on 11 years of cross-platform engagement data — a foundation that lets the model distinguish a real breakout from a coordinated spike or seasonal noise.

    • Founded on 11 yrs of data
    • 4.7B signals processed in 2024
    • 9 ML researchers on staff
    • 94% accuracy on 100K+ mentions

Processed 4.7 billion social signals in 2024 alone. Trusted by 1,800+ enterprise media teams including NBC News, Bloomberg, Ogilvy, and Edelman.

Editorial Standard

A score is a starting line, not a verdict.

“Every number on the Virality Score leaderboard has been read by a human editor before it reaches you. Our algorithm finds the signal. Our 38 senior journalists decide whether it is true, whether it matters, and whether you need to see it now. That is the only way a trend engine earns its place on a newsroom desk.”

The TopTrendingNews Editorial Board New York · London · Singapore
  • 14 min Median time-to-publish advantage versus AP and Reuters on trending stories.
  • 94% Prediction accuracy on stories trending above 100K mentions within 24 hours.
  • 38 Senior journalists with 600+ years of combined newsroom experience sign off on every ranking.
Methodology, Answered

The four questions our team gets asked most.

If you have ever wondered how minute-level windows, backtesting, or our enterprise API actually work, start here.

  1. What does “minute-level” actually mean for a Virality Score?

    Our leaderboard refreshes every 60 seconds and timestamps every ranking to the minute of ingestion. We do not claim second-by-second accuracy — every score you see reflects the most recent completed 60-second sweep, which is the smallest defensible window for cross-platform aggregation. That is the cadence behind our 14-minute median time-to-publish advantage over legacy wire services.

  2. Can the Virality Score be backtested against historical trends?

    Yes. Because the model is trained on 11 years of cross-platform engagement data, we run continuous retroactive backtests against the same corpus — comparing yesterday’s score to today’s verified outcome. On stories that ultimately trend above 100,000 mentions within 24 hours, our benchmark accuracy is 94%. The methodology is the same one the Reuters Institute Digital News Report 2024 flagged as a Top 10 trend-forecasting approach.

  3. How is the Virality Score different from a raw mention count?

    A raw mention count tells you what already happened. The Virality Score weighs velocity, cross-platform divergence, source diversity, editorial confirmation, and coordinated-activity flags — then projects forward. Two stories can sit at the same mention count and earn very different scores if one is plateauing in one network while the other is breaking across six. The score is designed to rank what is about to dominate, not what already did.

  4. Do you offer API access for newsrooms and enterprise teams?

    Yes. Our enterprise API delivers the same Virality Score feed used by 1,800+ media teams — including NBC News, Bloomberg, Ogilvy, and Edelman — with minute-level timestamps, category filters, and webhook alerts on score inflection points. Speak with our enterprise team for documentation, SLAs, and onboarding.

Want the daily leaderboard delivered before the rest of the web catches up?

Get the Daily Trending Brief