How TrendSpry Works
We'd rather you trust the numbers than be impressed by them. Here's exactly how they're built.
TrendSpry measures how much attention a topic is getting every day, compares it with that topic's own recent past, and turns the result into two things: a Trend Signal score (0–100) and a direction (rising, falling, steady, or new). The formula is below — nothing is hidden and nothing is invented.
We collect daily activity
Every day, for every topic: page views of its English Wikipedia article (worldwide), and the number of matching news articles in each country's English Google News edition. If enabled, appearances in YouTube's official trending-video charts are added too.
We compare it with its own past
For each source we take the average of the last 3 complete days (recent) and the average of the 28 days before that (usual). A topic is only ever compared with itself, never with unrelated topics.
We compute the score
Score = 50 + 25 × log2((recent + k) ÷ (usual + k)), kept between 0 and 100. Steady = 50, double = 75, four times = 100, half = 25. The small constant k stops tiny numbers (say 1 article → 3) from looking like an explosion. Worldwide scores blend Wikipedia (55%), news (35%) and YouTube (10%); country scores use news (80%) and YouTube (20%). Above 55 is rising, below 45 is falling; topics with less than a week of history are labelled new and capped at 90.
We show our working
Every topic page lists each source with its recent and usual daily numbers and its own score, links the Wikipedia article we track, and shows the date the data runs through.
Frequently Asked Questions
Does TrendSpry use AI to predict trends?
No. The Trend Signal score is a published calculation that compares a topic's last 3 days of activity with its own previous 4 weeks — not a machine-learning prediction. AI (Claude) is only used to help decide which new topics are worth tracking and to write optional content ideas for Pro users; it never sets or changes a score.
Where does the data come from?
Daily page views of each topic's English Wikipedia article (Wikimedia's official, free pageviews API), the number of matching news articles per day in each country's English Google News edition, and — when enabled — how often the topic appears in YouTube's official trending-videos charts. Every topic page lists the sources behind its current number.
How often is the data updated?
Every day. An automated job refreshes every tracked topic at least once every 24 hours. Scores are calculated on complete days only, so a topic page shows data through yesterday.
Why can't I find a topic?
We only show topics we actually track, so every number is real. When you search for something we don't track yet, it goes into a queue; new topics are reviewed and added every day, and usually show data within about 24 hours.
What does a Trend Signal score actually mean?
It measures momentum, not size. 50 means activity is at its usual level, 75 means roughly double the usual level, 100 means four times or more, and 25 means about half. A small niche topic can score higher than a huge one if it is growing faster.
Why are some countries missing for a topic?
Country figures come from news coverage in that country's English Google News edition, and are only shown when there were at least a few matching articles in the last 3 days. Countries without an English edition, or with too little coverage, are left out rather than shown with a misleading number.