DEV Community

Cover image for How I Found the Best Posting Times on dev.to With NodeJS(+ Usable Code)
Code_Jedi
Code_Jedi

Posted on • Updated on

How I Found the Best Posting Times on dev.to With NodeJS(+ Usable Code)

One thing that bothers a lot of creators(including me) is, well, what if I post at the wrong time? What if I post and my post will get barely any traction and all my effort will go to waste?
Sad Pepe

But not to worry, I found a way to get an idea of when the best posting times are for different tags on dev.to!


The solution

I made a NodeJS script that:

  1. Scrapes the dates on which the top 50 posts of the last year for different tags have been posted on. Post date
  2. Converts the dates to weekdays.
  3. Prints the first, second and third best days of the week to post on for a tag.

The code

Here's the code that does the magic:

index.js

const puppeteer = require('puppeteer');
async function bestdays() {
  const browser = await puppeteer.launch({});  
  days = []
  var page = await browser.newPage();
  const args = process.argv.slice(2)
  await page.goto('https://dev.to/t/'+args[0]+'/top/year');    
  for(x = 0;x < 50;x++){
    try{
      let date1 = await page.$eval("#substories > div:nth-child("+x+") > div > div.crayons-story__top > div > div:nth-child(2) > a > time", element=> element.getAttribute("datetime"))
      const d = new Date(date1);
      let day = d.getDay();
      days.push(day)
    }
    catch{}
  }
  function getAllIndexes(arr, val) {
    var indexes = [], i = -1;
    while ((i = arr.indexOf(val, i+1)) != -1){
        indexes.push(i);
    }
    return indexes;
  }
  function max(input) {
    if (toString.call(input) !== "[object Array]")  
      return false;
    return Math.max.apply(null, input);
  }
  var indexes1 = getAllIndexes(days, 0);
  var indexes2 = getAllIndexes(days, 1);
  var indexes3 = getAllIndexes(days, 2);
  var indexes4 = getAllIndexes(days, 3);
  var indexes5 = getAllIndexes(days, 4);
  var indexes6 = getAllIndexes(days, 5);
  var indexes7 = getAllIndexes(days, 6);
  dm = [indexes1.length, indexes2.length, indexes3.length, indexes4.length, indexes5.length, indexes6.length, indexes7.length]
  dm2 = []
  weekdays = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
  console.log("\n")
  console.log("Best day: "+weekdays[dm.indexOf(max(dm))]+" with "+max(dm)+" top posts from last year")
  for (j = 0; j < dm.length;j++){
    if (j != dm.indexOf(max(dm))){
      dm2.push(dm[j])
    }
    else{
      dm2[j] = 0
    }
  }
  console.log("Second best day: "+weekdays[dm2.indexOf(max(dm2))]+" with "+max(dm2)+" top posts from last year")
  dm3 = []
  for (j = 0; j < dm2.length;j++){
    if (j != dm2.indexOf(max(dm2))){
      dm3.push(dm2[j])
    }
    else{
      dm3[j] = 0
    }
  }
  console.log("Third best day: "+weekdays[dm3.indexOf(max(dm3))]+" with "+max(dm3)+" top posts from last year")
  console.log("\n")
  await page.close()
  await browser.close()
}
bestdays();
Enter fullscreen mode Exit fullscreen mode

Make sure to have NodeJS and Puppeteer installed before running the script, then run the code by entering node index.js and the tag you want to get the best posting times for (without the #) in your terminal.
For example: node index.js webdev.

Once you've ran your script you should see that it outputs something similar to the following.
Best Days


Best posting days for 20 of the most popular tags

For people who don't want to setup the script themselves, here are the best posting days for the top 20 most popular tags on dev.to:

Javascript

Best day: Wednesday with 10 top posts from last year
Second best day: Thursday with 10 top posts from last year
Third best day: Saturday with 10 top posts from last year
it's a tie

Webdev

Best day: Wednesday with 14 top posts from last year
Second best day: Thursday with 11 top posts from last year
Third best day: Friday with 7 top posts from last year

Tutorial

Best day: Wednesday with 9 top posts from last year
Second best day: Saturday with 9 top posts from last year
Third best day: Thursday with 8 top posts from last year

React

Best day: Wednesday with 12 top posts from last year
Second best day: Saturday with 10 top posts from last year
Third best day: Thursday with 9 top posts from last year

Python

Best day: Tuesday with 12 top posts from last year
Second best day: Sunday with 12 top posts from last year
Third best day: Monday with 8 top posts from last year

Productivity

Best day: Wednesday with 12 top posts from last year
Second best day: Monday with 11 top posts from last year
Third best day: Friday with 10 top posts from last year

CSS

Best day: Wednesday with 14 top posts from last year
Second best day: Tuesday with 8 top posts from last year
Third best day: Thursday with 8 top posts from last year

Devops

Best day: Wednesday with 12 top posts from last year
Second best day: Friday with 9 top posts from last year
Third best day: Saturday with 7 top posts from last year

Discuss

Best day: Friday with 11 top posts from last year
Second best day: Monday with 10 top posts from last year
Third best day: Wednesday with 7 top posts from last year

Career

Best day: Tuesday with 11 top posts from last year
Second best day: Monday with 9 top posts from last year
Third best day: Wednesday with 8 top posts from last year

Opensource

Best day: Wednesday with 11 top posts from last year
Second best day: Monday with 9 top posts from last year
Third best day: Friday with 8 top posts from last year

Node

Best day: Sunday with 14 top posts from last year
Second best day: Tuesday with 8 top posts from last year
Third best day: Thursday with 8 top posts from last year

HTML

Best day: Wednesday with 11 top posts from last year
Second best day: Tuesday with 9 top posts from last year
Third best day: Thursday with 9 top posts from last year

Codenewbie

Best day: Wednesday with 12 top posts from last year
Second best day: Monday with 9 top posts from last year
Third best day: Tuesday with 8 top posts from last year

AWS

Best day: Monday with 12 top posts from last year
Second best day: Wednesday with 10 top posts from last year
Third best day: Tuesday with 9 top posts from last year

Typescript

Best day: Tuesday with 10 top posts from last year
Second best day: Monday with 8 top posts from last year
Third best day: Saturday with 8 top posts from last year

Android

Best day: Tuesday with 10 top posts from last year
Second best day: Wednesday with 10 top posts from last year
Third best day: Friday with 7 top posts from last year

Java

Best day: Sunday with 14 top posts from last year
Second best day: Wednesday with 8 top posts from last year
Third best day: Monday with 7 top posts from last year

Github

Best day: Monday with 14 top posts from last year
Second best day: Tuesday with 9 top posts from last year
Third best day: Friday with 6 top posts from last year

Blockchain

Best day: Monday with 9 top posts from last year
Second best day: Friday with 9 top posts from last year
Third best day: Saturday with 9 top posts from last year
another tie


That's it for this article, hope it was helpful! Follow for more programming stuff.

Top comments (36)

Collapse
 
grahamthedev profile image
GrahamTheDev • Edited

Or you could...if you were really silly...just:

  • download 250k+ worth of live articles on dev.to using the API over the space of about 15 days so as not to get rate limited,
  • store them all in a database (well over 10GB…and that is just text not images!)
  • and then query that database based on likes (per post) vs time posted, removing the top 5% and bottom 5% to make it slightly more accurate and output it all to a simple table.

I would imagine the results would look something like this if anyone was silly enough to attempt to do it (third table):

inhu.co/dev_to/analyse/timeofday.php

But nobody would be silly enough to do that now would they? 😉

Being serious for a second, I didn't think to break it down by tag...might have to look into that if I ever get around to finishing that project as that is a great idea!

Great post, as you can probably tell I find stuff like this really interesting! ❤🦄

Collapse
 
renanfranca profile image
Renan Franca

I am about to ask you on another article discussion (that you share the tables link) how did you get the final result 🤣

Thanks!

Collapse
 
dmk1111 profile image
dmk1111

Could you please share which time zone do you use in this table? Thanks!

Collapse
 
grahamthedev profile image
GrahamTheDev

It is GMT, hope that helps!

Collapse
 
zippytyro profile image
Shashwat Verma

This is a better approach I think, the person didn't knew about the API

Collapse
 
grahamthedev profile image
GrahamTheDev

Their approach was far more efficient though, I just brute forced it!

The DEV API is not well publicised either to be fair, so it is easy to miss.

Collapse
 
techman09 profile image
TechMan09

Love it!

Collapse
 
z2lai profile image
z2lai

What on earth do these numbers and colors mean? Did I miss the legend?

Collapse
 
grahamthedev profile image
GrahamTheDev • Edited

No it is for personal use so it is not user friendly!, I just happened to share it in case it was useful for someone. The numbers are reactions / article released at a given time as an average. The colours are just a heat map, dark green is best (and yes the colours are awful!) 🤣

Collapse
 
thumbone profile image
Bernd Wechner

While a very impressive effort, and I love it, I should point out that there is a risk here of the classic confusion between correlation and causation. This confounds forecasting all the time and is a central conversation among forecasters I think.

Why? Well this is a forecasting effort, that is the first observation. Essentially we're hoping to predict that if I post my JavasScript article on Wednesday my chances of reaching people, gaining views and reactions, rises. While not explicitly stated, it's implicitly the goal.

One of the main tools forecasters have is history. But all history can show us quickly and easily is correlated measures. In this case we see a correlation between Wednesday posts and appearing in the top 50.

But for this to form a substantial forecast we need to assume or demonstrated or hypothesise if you will a causative relationship. Is it the case that posting on Wednesday raises the likelihood of that. A true forecasting effort pursues some evidence as to that hypothesis.

One hypothesis is that the editors compiling those lists attend to the job on Thursday as a routine ;-). Confirmation would require a word from DEV editors.

Collapse
 
code_jedi profile image
Code_Jedi

I'm pretty sure the top posts lists are compiled automatically by number of reactions and/or reads.
I guess it might also be that there are a couple of really big creators in a tag like Javascript that just routinely post on a Wednesday and get the attention of their followers no matter what and become top posts.
Still, I doubt that this is the case because from what I've seen, the top 50 posts from a tag are dominated by at least twenty different creators.
As for evidence, of corse, if you make a crappy post but post it on a Wednesday, you still won't get good results. But considering that you've made a proper post for a tag, taking into account that the biggest portion of the top 50 posts in your tag have been posted on a Wednesday, wouldn't that, considering the diversity of the posts and creators in the top 50 list, increase your post's likelihood of performing better than it would have if you had posted say on a Monday?

Collapse
 
aneeqakhan profile image
Aneeqa Khan

wow! most common days are Wednesday and Thursday ✌🏼

Collapse
 
code_jedi profile image
Code_Jedi

That's why I posted on a Thursday ;) 👍

Collapse
 
aneeqakhan profile image
Aneeqa Khan

awesome 😄 and I can see this as a top post 🤞🏻

Collapse
 
Collapse
 
code_jedi profile image
Code_Jedi

I've never heard of this API. How did you find it?

Collapse
 
ouzkagan profile image
ouzkagan

Oh I just checked network tab :)

Collapse
 
zippytyro profile image
Shashwat Verma

Dev.to have their own api. Just search devto API on google. You can do ton of stuff with it. Great post btw

Collapse
 
johnrock profile image
John Rock

Love the analysis and great work building the app to gather the data.

Collapse
 
code_jedi profile image
Code_Jedi

Thanks ;)

Collapse
 
lucasandre profile image
Lucas André

Awesome! Thanks! :D

Collapse
 
ronaldgrowe profile image
Ronald Rowe

Great idea for a project! Also, great work on the project! 👊

Collapse
 
code_jedi profile image
Code_Jedi

Thanks👌

Collapse
 
shareef profile image
Mohammed Nadeem Shareef

Now I will post only on Wednesday.

Thanks man!

Collapse
 
theaccordance profile image
Joe Mainwaring

Awesome work creating a solution that outputs an insight!

Collapse
 
trishathecookie profile image
Trisha Lim

Interesting. What about the best time of day?

Collapse
 
renanfranca profile image
Renan Franca

@code_jedi, I am interested in the best time of the day too 🤓