How “the survivor’s mistake” makes it difficult to draw conclusions in content marketing and smm

For the first time I heard about the "systematic error of the survivor" with the example of dolphins. Everyone knows that sometimes they ship out shipwrecked. What a great fellow!

In fact, dolphins just love to push objects with their noses, and no one knows how many drowning people they pushed even deeper into the ocean, and not to the shore. We know only examples of miraculous salvation, because the dolphin victims will not tell us anything.

This is the “survivor’s mistake” - because only part of the information is available to us, we make wrong conclusions.
')
During World War II, this phenomenon could cost the lives of thousands of soldiers. The fact is that the US bombers, who returned from a mission, were most often shot through with wings and tail. These units were planned to be reinforced with armor so that as many aircraft as possible would return to the base.

They would have done so if the mathematician and statistician Abraham Wald, who understood the phenomenon of the “survivor's error”, had not intervened. He pointed out that the shot-through parts of the aircraft are not his weak points, as it seems at first glance, but strong. With these damages, the plane can return to the base.

And those planes that were shot down, apparently, suffered other parts - and they need to be strengthened. Later, when it became possible to analyze the downed aircraft, this hypothesis was confirmed. Abraham Wald helped avoid a fatal error based on incomplete data.

What's the marketing

In content marketing and smm, such an error occurs regularly. For example, we recently published an article on “the theory of constraints” on Picaba. Under it, 144 comments, of which about half are different claims to the article.

Many people didn’t like the basic example of pipes (it’s really not the best one). Commentators wrote: "The man does not know what friction loss and hydraulic resistance are." Others complained that this is an inappropriate article for an entertainment site. Still others accused us of superficial reasoning.

If you read the comments, it will be clear that the article met, to put it mildly, cool. But besides comments on Picaba, unlike many other sites, there is a “like-dizlick” scale. Judging by the comments, under this article, the assessment should be divided 50 to 50:



But in fact, it looks like this:



This is not just a successful post, but an incredibly successful one. More than a thousand people rated the article positively. If there were no scale, we would decide that the article “did not enter”. It turned out - exactly the opposite.

Intuitively, it seems that commentators are a good sample of all who read the article. And if half of them are dissatisfied, then half of those who read will put a dizlike.

Actually commentators are those “survivors”. This is not a random sample at all, but specific people who were not too lazy to go in the comments and write something. To put it bluntly, one can imagine that every dislike that put it went into the comments in order to express discontent. And from those who put "like" they went to write comments about ten percent.

We are lucky that there is a scale on this site. Of course, we can get caught in yet another “mistake of the survivor” - likes do not always reflect the real picture of the world. But experience suggests that they are much closer to reality than comments.

Conclusions for the marketer

We often write articles and are confronted with a selection error and a “survival mistake” - a coolly accepted post that we recorded as unsuccessful brings customers.

Another manifestation of this phenomenon is the article that many subscribers requested and waited for, “does not enter.” This happens for the same reason.

For example, we asked subscribers: “Which post are you waiting for?” And half of the commentators wrote that they were waiting for a story about Logmachine employees. But, as we already know, half the commentators are not half the audience. It can be one tenth or even one hundredth. As a result, we make a post for a small part of the audience, who was not too lazy to write comments.

More reliable information can be obtained by methods in which a large part of the audience will be involved. Like-dyslake buttons and polls will do - most are easier to click with the mouse than to write comments. It is only necessary to correlate the number of voters with the general audience to draw any conclusions.

Crib:

1. The systematic error of the survivor - the wrong conclusions that we draw on the group of "survivors", about which we have a lot of data
2. In content marketing, this phenomenon is manifested in the fact that comments and other activities are not a direct reflection of the entire audience.
3. The most common mistake is to draw conclusions about the entire audience by comments. Commentators are not random or representative samples.
4. A more accurate picture can be given by the “like-dislike” scale or a poll - more people are participating in them.
5. You should always evaluate what part of the audience you are drawing.