I pulled the numbers on eight videos we packaged on our own channels last quarter. Four were built for the moment, loud, reactive, tied to something happening that week. Four were built to be found in eighteen months. The trending four beat the evergreen four by roughly 40 percent on click through rate in the first seven days. By day ninety the evergreen four had more total views than the trending four combined.
That is the whole argument in two sentences, and it is still the thing most creators get wrong, because week one is the number you feel and month six is the number you only see if you go looking.
Two different jobs wearing the same costume
A thumbnail is not one object doing one job. It is a promise made to a specific person at a specific moment. Change the moment and you have to change the promise.
Trending packaging is aimed at somebody who already has context. They know the news, the drama, the release, the argument. Your thumbnail does not have to explain anything. It has to signal that you are talking about the thing they are already thinking about, and that your angle is worth the click over the eleven other people covering it. Recognition is the whole mechanic. Big face, big reaction, the recognizable object or logo, minimal text because the viewer supplies the setup themselves.
Evergreen packaging is aimed at somebody with no context at all. They found you through search, through a suggested rail nine months from now, through a browse feed on a Tuesday. They do not know what happened that week. They do not care. Your thumbnail has to carry the entire premise on its own, which means it has to be legible, self contained, and dated by nothing.
Same canvas. Completely different design brief. When a channel uses one visual system for both, one of the two jobs is quietly failing.
What actually kills an evergreen thumbnail
Date stamps. Not literal ones, though those are the obvious sin. The subtle ones do more damage.
A specific person's face who is only famous this month. A product photo of hardware that gets replaced next year. A price that inflation makes look silly. A UI screenshot from a piece of software that redesigns every eight months. A meme format. A phrase that reads as native in March and as cringe in November.
I have watched a good video die in the suggested feed purely because the packaging looked old. Nothing about the content was old. The advice inside was as true as the day it shipped. But the thumbnail read stale, and a viewer decides stale in about a fifth of a second, well before they read the title. That is the tax you pay for a design that only makes sense in one calendar window.
The content ages at whatever pace the topic ages. The packaging ages at whatever pace its visual references age. Those are two separate clocks, and most channels only watch one of them.
The fix is not to make evergreen thumbnails boring. Boring is a different failure mode and it is worse. The fix is to build the tension out of things that do not expire. A face expressing a feeling. A clear before and after. A comparison with two visible sides. A number that stands on its own. Those work on day one and they work in year two.
What actually kills a trending thumbnail
Hesitation. Trending packaging fails when it hedges.
If you are covering the thing everyone is covering, subtlety is a losing bet. You are competing in a rail where nine other thumbnails are shouting, and the polite one is invisible. This is the one situation where I will push contrast further than I normally like, crop tighter than I normally like, and let the text get bigger than my taste says it should be. Taste is not the metric. Being chosen out of a row of ten is the metric.
The other killer is being late. A trending package that lands three days after peak interest is a beautifully made object that nobody sees. Trending work has a shelf life measured in hours, which is exactly why I am honest with people about our turnaround. We work in a 48 to 72 hour window per thumbnail, and that is a real constraint. It is fine for planned uploads and it is a poor fit for chasing a news cycle that breaks on a Thursday night. If your channel lives on same day reactions, you need an in house solution, not a studio, and I would rather say that up front than take the money. There are situations where you genuinely do not need to hire anybody, and I have written about when hiring a designer is the wrong move rather than pretending every channel needs us.
The ratio I actually recommend
Most channels I talk to are running something like 80 percent trending and 20 percent evergreen, usually without having decided that on purpose. It just happens, because trending topics are easier to think of and the week one dopamine is real.
I think the healthy split for a channel that wants to still exist in three years is closer to the reverse. Maybe 60 evergreen, 40 trending, adjusted for niche. A news channel is obviously going to skew trending. A tutorial channel that skews trending is setting money on fire.
Here is how I think about it per video:
- Will this be searched next year? If yes, evergreen packaging, no exceptions, even if a trending angle would pop harder this week.
- Is the topic time bound but the lesson is not? Package evergreen, and put the time bound hook in the title where you can edit it later. Titles are cheap to change. Thumbnails are not.
- Is the whole value of the video that it is happening right now? Trending, and go loud.
- Are you entering a crowded rail? Trending rules apply regardless of topic. Legibility beats elegance.
That second point is underrated. The title and the thumbnail have different lifespans and different edit costs, which is why I treat them as one unit with two dials rather than two separate deliverables. I go deeper on that relationship in how the title and thumbnail split the work, but the short version is that the thumbnail should carry the durable promise and the title can carry the timely one.
The cold audience problem
Evergreen packaging has a specific difficulty that trending packaging does not: almost everybody who sees it has never seen your channel.
On a trending video, a decent share of your impressions come from subscribers who already know your face, your tone, your visual signature. That is a huge amount of free context. Your thumbnail is doing about half the work because their relationship with you is doing the other half.
On an evergreen video that is still earning impressions two years later, that ratio inverts. Most of those people are strangers. Your face means nothing. Your color palette means nothing. Every bit of persuasion has to happen inside the frame. This is the same design problem as packaging for a channel with no audience yet, and the same rules apply, which is why designing for people who have never heard of you is the piece I send to anyone whose evergreen back catalogue is underperforming.
Practically, that means evergreen work gets more text than trending work, not less. It means the subject of the image needs to be identifiable without any brand knowledge. And it means clean, deliberate subject isolation rather than a busy frame, which is a lot of what the background removal tooling comparison was actually about. A messy cutout on a trending thumbnail survives because the viewer already wants the topic. On an evergreen thumbnail, it reads as low effort and the stranger scrolls on.
How this changes what I build for a pack
When someone buys a pack from us, they get 8 custom thumbnails, one per video, at $250. Every one is drawn for its specific video. We do not run templates and we never will, partly for taste reasons and mostly because a template forces every video into the same packaging strategy whether it fits or not.
So the first conversation is almost always a sorting exercise. Which of these eight are evergreen and which are trending. That decision changes the composition, the text length, the crop, and whether I use any reference that will look dated. Title consultation is included in the pack precisely because you cannot make that call about the image without also making it about the words.
When a channel is genuinely unsure which way a video leans, that is the case for the A/B variant add on at $15. Two versions, two hypotheses, real data instead of my opinion. I would rather be corrected by your audience than be right in a meeting.
Some niches make the call for you. Photography channels sit very heavily on the evergreen side, because gear reviews and technique videos get searched forever, which is why the approach on our photography channel thumbnails work leans durable by default. And any packaging system is easier to hold together when the channel art around it is consistent, which is a smaller job than people expect and covered in the channel banner work.
Where I have been wrong about this
I used to think evergreen packaging meant restrained packaging. Softer, more tasteful, less shouty, because it has to age well and loud things age badly.
That was wrong and I tested it into the ground. On a mid size finance channel around 80k subs, we ran a quieter, more editorial evergreen style across a run of videos and the click through rate dropped meaningfully against their previous loud style. Not catastrophically, but consistently. The videos that were meant to earn for years earned less every single month, which compounds in the wrong direction.
Aging well and being quiet are not the same thing. A thumbnail can be extremely loud and still be timeless, as long as the loudness comes from composition and contrast rather than from references that expire. That distinction took me an embarrassingly long time to internalise, and you can see where it landed across our work.
The same logic shows up in long form audio and video too. When we cut podcasts on the Pixel Samy side of the business, the episodes that keep earning are the ones packaged around a durable question rather than a guest's news cycle. Different medium, identical clock problem.
The closing thought
Decide the packaging strategy before you design anything, and decide it per video rather than per channel. Most of the packaging failures I see are not craft failures. The image is fine. It is just answering a question nobody in that particular audience was asking.
If you want a second opinion on which of your last ten uploads were packaged for the wrong clock, send them over. I will tell you honestly, and I will tell you if the answer is that your packaging is already fine.