For about a year I ran what I thought was a clever strategy. Watch what was surging, make something about it quickly, capture the search traffic while it lasted, repeat.
It worked in the narrow sense that it produced views. It failed in every sense that mattered and it took me longer than it should have to see why.
The mechanics of a trend spike
Worth understanding, because it explains the failure.
Something becomes briefly interesting. Search volume rises sharply over a few days, plateaus, and then falls off faster than it rose. A month later it is close to zero.
If you publish during the rise, you get traffic. If you publish at the peak, you compete with everyone else who had the same idea. If you publish after, you get almost nothing.
Which means the whole game is speed, and speed is directly at odds with quality. I was making things in a day that should have taken a week, because in a week the window would have closed.
The audience never formed
This is the part I misjudged completely.
Someone who searches for a specific thing, finds your video, watches it, and gets their answer has no particular reason to come back. They were not looking for you. They were looking for information about a thing that no longer interests them.
My subscriber count rose but engagement per subscriber fell continuously through that year. I had accumulated an enormous number of people who had subscribed on a whim and never watched anything again.
Worse, when I published something outside the trend cycle, it performed terribly, because the audience I had built had no interest in me. They had an interest in a topic that had passed.
The channel became incoherent
Twelve months of trend-chasing produced a list of videos with nothing in common. Someone landing on the channel page could not tell what it was about, because it was not about anything.
Recommendation systems appear to use some notion of topical consistency when deciding who to show things to, and a channel with no consistent subject is difficult to place. My videos would perform in isolation and never seemed to help each other.
Channels with a clear subject seem to accumulate momentum, where each video benefits from the ones before it. Mine did not accumulate anything.
What the good version looks like
I do not think reacting to what is happening is inherently wrong. The version that works is different in a specific way.
Cover trends within your subject, not trends generally. If your channel is about photography and something is happening in photography, that is your story and your existing audience wants it.
Bring something the fast coverage cannot. Everyone can report what happened within a day. If you know the field, you can say what it means, which is worth more and lasts longer.
Accept losing the initial spike. Publishing a week later with something substantial gets less traffic on day one and keeps getting traffic for a year.
The evergreen comparison
The numbers made this obvious once I bothered to look properly.
My trend videos front-loaded almost all their views into the first week and then flatlined near zero.
The handful of things I made that answered a persistent question have been accumulating views steadily since. Several have now, over eighteen months, passed the total of trend videos that outperformed them by a wide margin in their first week.
The trend video felt better because the feedback was immediate. The evergreen video was worth more and gave almost no feedback at the time, which is exactly the wrong shape for how motivation works.
The psychological trap
Worth naming because it is the actual mechanism.
Trend chasing is intermittently rewarding. Sometimes you catch one and the numbers are dramatic. That intermittency is extremely reinforcing and it kept me doing it long past the point where the aggregate was obviously poor.
Steady work on a coherent subject produces no such spikes. The growth is gradual and unexciting and it compounds.
I do not think I reasoned my way out of it. I think I got tired, and the tiredness forced a slower approach, and the slower approach turned out to work better.
What I would do differently
Decide what the channel is about first, in a sentence, and check every idea against it.
React to things inside that subject, quickly if I can, but never at the cost of the standard.
Spend most of my effort on things that will still be worth watching in a year, and accept that they will look like failures for the first three months.
And treat a spike as what it is — a lot of strangers passing through, most of whom will not remember the channel by Thursday.
The comments told me before the numbers did
Something I noticed only in retrospect. During the trend year the comments were almost entirely about the topic and almost never about the video.
People correcting a detail, adding information, arguing with each other about the subject. Almost nobody saying they had watched anything else of mine or intended to.
On the slower work the comments changed shape completely. More questions directed at me, more references to earlier videos, more people who clearly knew the channel.
That shift was visible months before the subscriber quality showed up in any metric, and I had been reading the comments the whole time without registering what they were telling me.