I bet you have heard that dislikes are a form of engagement. Engagement is always good for our videos.
Unfortunately, this information is outdated. VidCon is the head of the search and discovery team. The people who are creating the algorithm. And the questions he answered gave us new information about dislikes.
The big question is: Are dislikes good or bad for our videos?
If we talk about that then it’s the question about whether the algorithm does more or less promotion for our videos, when they get a lot of dislikes. And the way how the algorithm makes more promotion for our videos or less is by the way of showing our videos to viewers.
When we talk about dislikes. I’m focusing on the dislikes and how they impact the impressions that the recommendation engine does. Have a look at the Studio. You have the impressions shown here in the orange bar and these impressions I’m talking about.
So the big question is how do dislikes influence getting more or less of those impressions?
Practically speaking that means that your video gets shown to the viewer on the home page in the suggested video section.
The goal of YouTube’s recommendation engine is to show those videos to a viewer. That the viewer is likely to enjoy watching and of course not showing videos to the viewer that the viewer will not enjoy.
However, it is always learning. This recommendation engine is always learning for each individual viewer what that viewer likes and what that viewer doesn’t like. Since we are talking about dislikes, let’s look at the negative side.
How can a viewer signal the algorithm that the viewer doesn’t enjoy the video?
That the recommendation engine should show less of those videos to the viewer. There are several different ways. One particular way is that the viewer just doesn’t click a video that gets recommended to that person.
A second signal is that if the viewer clicks the video but then has just a very short watch time. You guessed it already. Here’s also where the dislikes come into play.
If a user dislikes a video, well, then it’s obviously a video that the viewer didn’t enjoy. Of course, there are also other ways how the viewer can explicitly tell the algorithm that he or she is not interested in.
For example, if you click the menu on a video then you can tell YouTube “not interested”.
So, all in all, we can say that all these signals give the algorithm the incentive to recommend less of those videos to the same viewer, less of the same video topic, less of the videos from a given channel and so on.
That was really interesting when I was at VidCon answered a few questions. They looked at how videos got recommended. They were explaining how the recommendation system chooses videos to show to the user and if it does rightly. What they did in order to prove to us how the recommendation system made. Its decision is by looking at the like to dislike ratio.
You know, the likes versus dislikes you also see in the Studio nowadays. This ratio is not the only factor but for them in this situation. It was a usable indication of whether a video is worth recommending or not. And to no surprise after the talk, I witnessed how another youtuber asked about this likes and dislikes ratio.
Whether this is actually a factor in the algorithm and he explicitly confirmed that this is a data point that feeds into the algorithm. However, he also said it’s only a small factor.
So really don’t overstate that. But it shows us how dislikes go into the algorithm.
What do we learn from this?
Well, if a viewer starts to dislike your videos, multiple videos in a row let’s say then for sure the recommendation system will recommend less and fewer videos of your channel.
But don’t worry about that this might affect the success of your video in general. The recommendation system works in a way that it learns the interest of each and every viewer individually.
Each and every viewer gets different recommendations and thus if some viewer doesn’t like your videos. Well, then the recommendation for that viewer gets adopted. But that doesn’t change that your fans will still get to see your videos.
Of course, if your video gets really a lot of dislikes then this information spreads across the recommendations of other viewers as well. But this is just in extreme cases.
Usually, you don’t have to worry about it.
The dislikes only affect the people who do the dislikes and frankly, this is a good thing. If a person doesn’t like your videos, well, then it’s maybe the best that this person moves on to watch videos of other channels.
I don’t want you to worry about dislikes at all. There is actually a scientific paper released about the recommendation system of YouTube. In there it says: “Although explicit feedback mechanisms exist on YouTube.
Dislikes and likes, in product surveys, which would be the “not interested” feature
We use the implicit feedback of watches to train the model, where a user completing a video is a positive example”.
What they mean by that is that watch time, something implicit in a sense that the viewer doesn’t actively do this like clicking the thumbs up or thumbs down button. The watch time really gives YouTube much better information whether a person enjoys a video or not.
So they explicitly tell us that dislikes likes and so on are a factor. But they don’t really use it that much. As confirmed by Todd when he said at the conference that the likes to dislikes ratio is a factor in the algorithm but not an important one.
So the big question is what are the important factors?
What can you do to make sure that your video gets recommended?
Well, this comes down to the most important factors: Do a great video where people see value in it, have great thumbnails that will improve your click-through rate, and create the video in a way that it produces high watch time. All three together will make your video successful in the recommendations.
So Don’t worry about dislikes.