Each author worries about the life of his publication, after publication he looks at the statistics, waits and worries about the comments, wants the publication to gain at least an average number of views. With Habr, these tools are cumulative and therefore it is quite difficult to imagine how the author's publication begins its life against the background of other publications.
As you know, most publications gain views in the first three days. To give an idea of how the publication fared, I tracked the statistics and introduced a monitoring and comparison mechanism. This mechanism will be applied to this publication and everyone will be able to see how it works.
The first step was to collect statistics on the dynamics of publications for the first three days of the life of the post. To do this, I analyzed the flows of readers, according to publications for September 28 during their life from September 28 to October 1, 2019 by fixing the number of views at various intervals during this period. The first diagram is shown in the figure below, it was obtained as a result of matching the dynamics of views over time.
As can be calculated from the chart, the average number of views of a publication after 72 hours with a power-law approximation will be approximately 8380 views.
Rice. 1. Distribution of views by time, for all publications.
Since the "stars" are clearly visible, we will present these data without them for standard publication. We will cut off for those publications that received more than the average number of views for 3 days - 10225 pieces, Figure 2.
Rice. 2. Distribution of views by time, for average publications, without "stars".
As can be calculated from the diagram, the average number of views of a medium-demand publication after 72 hours is predicted by a power-law approximation to be approximately 5670 views.
The numbers are interesting, but there is a tool with more practical value. This is the average share for each time period. Let's define them and present them in Figure 3.
Rice. Fig. 3. Actual time distribution of the share of views from the total number of views for three days and theoretical approximation lines, thin Excel polynomial and bold own solution.
I don’t see much point in conducting a separate analysis on clusters of “stars” and ordinary publications, since in this solution everything was considered in a standardized coordinate system, by shares.
Thus, it is possible to build a table of values with fractions of time and, accordingly, predict the total volume of views for three days.
Build the specified table and predict the flow for this publication
Since I will publish a post around 0 hours on October 3, everyone can compare the flow with the predicted value. If it is less, then I'm out of luck, if more, then readers are interested.
I will try to represent the real flow in the graph below as I observe it.
Rice. 4. The actual flow of readers of this publication in comparison with the theoretical forecast.
In conclusion, I can say that each author can take as a guideline the calculation table, which is presented above. And by dividing the actual flow of his publication at a certain moment by the value in the share column for that moment, he can predict the number of readers at the end of the 3rd day. And during this period, the authors have the opportunity in one way or another to influence the readability of his material, for example, to respond more actively and in more detail in the comments. You can also compare your publication with others, understand how external publications affect readers' priorities. The only advice, please understand that these figures were obtained from the analysis of the flow of readers of publications of just one day, September 28, 2019.
Source: habr.com