What actually counts as a trend
The word trend is asked to cover too much. A hashtag with two hundred million views is called a trend. So is a slow, two-year shift in how people dress for the office. A brand that treats those two things the same will lose money on both. Since I'm building a company on telling them apart, I should write down where I draw the line.
The test I keep coming back to is this: a trend, for a brand, is a change in what people want that you can see coming, confirm, and sell into while it still matters. Every word of that is doing work, so let me take them in turn.
The tag is not the trend
The easiest mistake is reading a view count as demand. A hashtag is a container, and view counts tell you the container is being passed around, they don't tell you what's inside it. I've sat in enough trade meetings to know how seductive a big number is. Two hundred million views feels like it must mean something. Often what it means is that one video was funny.
What's inside the container is people talking, and that's the part worth reading. Under a tag that's actually a trend, you find people asking where to buy the thing, complaining that everything in the shops is the wrong version of it, describing the exact occasion they need it for. That's consumer voice: frustration and desire in the customer's own words. Under a tag that's just noise, you find reactions to a video. The counts can be identical. The conversations never are.
This is why I care more about a thousand comments than a million views. Views measure the reach of content. Comments tell you whether a want is forming, and wants are what a brand can sell into.
Growing beats big
The second mistake is confusing volume with velocity. A big flat signal is a category, not a trend. Denim is enormous and going nowhere in particular; nobody needs an alert about denim. The interesting signal is smaller and moving: a term that was nowhere two months ago and has doubled twice since.
Merchandisers know this instinctively about their own numbers. Nobody celebrates a bestseller for being big; they watch what's accelerating, because acceleration is where next season's range comes from. Trend detection is the same discipline pointed outward. The question that matters is how fast is this growing, and from what base, rather than how big is it today. A small base growing fast deserves attention that a large base drifting sideways does not.
One loud source is a rumour
Platforms have different jobs in a customer's life, and that changes what a signal from each one means. One is where people perform, another is where they ask embarrassing questions, another is where they search when they're minutes from buying. A spike on a single platform can be an algorithm having a moment. The same idea surfacing on platforms with different jobs, entertainment in one place, questions in another, purchase-minded searches in a third, is a different animal, because it means the want has escaped its container and is travelling.
So confirmation across sources beats loudness on one source, every time. The trends that ended up mattering commercially, the ones that filled racks for two years rather than two weeks, all made that journey from performance to conversation to search. The ones that stayed on one platform mostly stayed entertainment.
Trends have a shape
A trend has a life. It emerges, it peaks, it fades, and the correct action is different at each point. Early, the buy is small and the upside is being first. At the peak, the margin is in having stock while everyone else is reordering. Late, the only sensible conversation is about markdown risk. Acting on a real trend at the wrong point in its life loses money just as reliably as acting on a fake one.
The uncomfortable part is that the window for the best decisions sits before the moment of maximum evidence. When every signal agrees, you're at the peak, and the peak is when it's too late to be early. Anyone can call a trend at its loudest. The commercial value is in calling it while calling it still takes some nerve, which is exactly why the confirmation discipline above matters so much: it's what lets you act early without acting on everything.
So what counts
Put together, the bar looks like this. Real people wanting something in their own words, not just watching content about it. The want growing, not merely large. The same want confirmed on sources with different jobs, not loud in one place. A known position in its life, so you act at the right point. And one more, which sounds obvious and is skipped constantly: you can name the products it sells. A trend you can't connect to anything anyone stocks is culture, and culture is lovely, and it isn't a merchandising decision.
That last step, from a named trend to the products in a real catalogue that fit it, turns out to be its own hard problem, and it deserves its own post.