How demand is measured
Superdemands measures demand as Google search interest: how often people search for a topic, compared with that topic's own busiest point. Search is the closest public record of what people want, because people search for what they intend to buy, watch, learn or do.
Which topics are tracked
Every topic that appears on Google's daily trending searches is added the day it appears, with the category Google files it under. A topic stays on the site for a year after it last trended; each time it trends again, its whole history is refreshed. Right now 43 topics are tracked across 10 categories.
Three histories per topic
Each topic carries five years of monthly interest, one year of weekly interest and three months of daily interest. Every series is scaled from 0 to 100, where 100 is the busiest point inside that series' own period, the same convention Google Trends uses. Monthly values are the average of the weeks in each month, so a single busy week does not dominate five years.
Because each series is scaled to itself, the numbers compare a topic with its own past, not one topic with another. A topic at 100 is at its own peak; it is not necessarily searched more than a topic at 40. The compare page lines topics up by timing and shape for the same reason.
Growth
Growth over a range compares the start of the range with its end. To keep one odd day from deciding the answer, each end is the average of a sixth of the range's points. One month and three months use the daily series, one year the weekly series and five years the monthly series. A start below 1 counts as 1, so a topic rising from nothing shows a large but finite number.
Momentum labels
The label comes from the weekly year: the last quarter is compared with the nine months before it, and the last two weeks are checked separately so a fresh spike cannot pass as a lasting rise. Values below 5 count as noise.
| Label | Rule |
|---|---|
| 🚀 Exploding | The last three months average at least twice the year before. |
| 📈 Growing | The last three months average at least 20% above the year before. |
| âš¡ Spike | The last two weeks reached at least three times the year before and twice the quarter, without a lasting rise. |
| 📉 Peaked | A peak of at least three times the year before inside the last quarter, now below half of it. |
| ✨ New | Fewer than four weeks of measurable interest in the last year. Growth numbers are held back until a history builds up. |
| âž– Steady | None of the above: interest within its usual range. |
Seasonality
A topic is called seasonal only with at least three full years of monthly data, when the same month sits at least 35% above its own year's average, the quietest month at least 15% below, and the high month beats its year in most years. Each month is measured against its own year so a long rise or decline does not look like a season.
Categories
Categories are Google's own. A topic takes the category Google listed it under on the most recent day it trended. The category line is the average weekly interest of every tracked topic in the category, counted only in weeks where enough of them have data.
What the site leaves out
Topics about deaths, shootings, disasters and similar harm trend too. Their pages stay reachable by search, but they are kept out of rankings, highlights and search engine indexes. Pages for brand-new topics without a history are also kept out of indexes until their history fills in.
Limits worth knowing
Search interest measures attention, not sales or sentiment: a product recall and a launch both drive searches. Very small topics can be rounded to zero by the source. Trending lists cover the United States. Every page shows the date its data is from.
Getting the data
Every topic page links to a JSON copy of its series. A free key returns the same five-year search histories for any keyword, not only trending ones, through the API or as a connector in Claude, Cursor and ChatGPT.