Where the numbers come from

The source

Wikipedia publishes how many people read each article every day, through the official Wikimedia pageviews API. If a thousand people read about air fryers today and eight hundred did last month, interest is climbing. It is real recorded behaviour from millions of people, not a survey and not a model.

Bot traffic is excluded — the figures count human readers only. Every topic page links to the exact article behind its number so you can pull the same data yourself.

How to read a topic

Readers per day

The 30-day average. Bigger numbers are steadier, so they are the more reliable signal — a topic on 2,000 readers a day tells you more than one on 90.

30-day change

This month against last month. It drives the verdict, and it is deliberately slow to react.

7-day change

The past week against the one before. Quicker to spot a turn, and noisier for it.

The verdict

Surging, rising, steady, cooling or collapsing — set by the 30-day figure. It can differ from the weekly number, and when it does, both are shown.

What this cannot tell you

• Readership is attention, not sales. People read about a thing before they buy it, so interest usually leads demand — but it never promises anyone is spending.
• Seasons look like trends. Tents climb every summer and fall every autumn. With months of history rather than years we can’t always separate a cycle from a real decline, so affected topics carry a warning rather than a confident verdict.
• Quiet topics are noisy. Anything under about ninety readers a day is excluded, because at that volume a swing of a dozen people looks like a major move.
• Some topics share an article. Wikipedia files “air fryer” under “convection oven”, so that reading covers both. Where it happens, the page says so.