32

This question relates to the following user activity on Stack Exchange and its network:

  • Best time to ask questions to get fastest answers
  • Best time to ask questions to be more likely to get answers
  • Time of posting (posts include both questions and answers)
  • Number of views for questions created at particular time

I've done a simple research with the great https://data.stackexchange.com/ and have added the results as the answer below.

Knowing these facts won't change everyone's habits, but a possible 15-25% improvement in getting answers faster and better may still be interesting.

When not mentioned, the data comes from Stack Overflow, since it's the biggest subsite. You can analyze other subsites with tools mentioned in the first answer.

Notation

X-axis in the charts below denotes time: 1.0 is Sunday, 12 am GMT. For example, 3.5 means Tuesday, 12 pm GMT.

Improve this question
  • 1
    @AndrewC 1.0 is Sunday, 12am GMT. Thanks for the note; I added notation to the post. – Anton Tarasenko Jul 1 '14 at 13:20
  • 1
    Agree with @blah238, voted to reopen. :) – Shadow Wizard is Vaccinating Jul 17 '14 at 7:16
25
+150

To answer the above questions, I've prepared a few queries.

Queries

To get the data, press Run query and then choose "Graph" tab. Graphs look like this:

enter image description here

Features

  • You can get data for your account alone, so the stats become relevant exactly for your type of questions.
  • You can use this across different StackExchange websites. See "Switch sites" on the query's page.
  • You can change the frequency of data (within a day, week, or combined).
  • Different time frames for the data: for data range and posting timeframes.

Facts from the data

  • You are 25% more likely to get an answer if you ask a question in the right time.
  • You can increase the number of views of your question by about 15% if you post it at the best moment.
  • 62% of accepted answers are given within one hour after asking. 39% are answered within 15 minutes.
  • In most cases, data shows same metrics even if taken from the 5-year activity.

Interesting moments

The probability of getting an answer increases with the average time-to-answer

enter image description here

That is, the time good for getting a fast answer is also the time when the probability of getting any answer is lower. Unlike an intuitive expectation that at some times community is active enough to answer questions and go this fast.

The average number of views gives a more explicit time-pattern for longer periods

enter image description here

The problem with that is: in the long-run these pattern of dependence should not arise because posts less and less depend on their posting time, as they are reached through search engines and cross-links.

And the one-year graph shows a very bleak dependence of views from posting time. Perhaps, that's one difference in behavior patterns that appeared over the years on SE.

Questions and comments

Your questions and corrections are welcome.

There're more interesting questions about the community's behavior. Hard to cover any significant fraction of them, but I hope to post more later.

Improve this answer
8

Three different stories

There's one thing to share about the most important number: the likelihood of getting an answer at all.

The likelihood depends on posting time, but in an interesting fashion. Stack Exchange websites exhibit different patterns. Just three graphs to illustrate (Axis X denotes time, 1.0 = Sun, 12 am GMT).

Stack Overflow

An expected pattern with smooth changes over the time: a 25% increase in the likelihood of getting an answer if posted in time.

Stack Overflow

Mathematics

Mathematics

Small jumps over a stable average probability of getting an answer. The probability is slightly higher on the weekends.

Ask Different

A twofold change in the probability in just 3 hours (between 5.0 and 5.5). These jumps seem to be data artifacts, but there's a noticeable trend of the probability increasing from 12 am to 12 pm each day.

Ask Different

Improve this answer
0

I was asking myself the same question but wasn't able to find this topic. This lead me to create these two queries:

Since the queries above have parameters, I have decided to post these queries here as a no brainers.

Note: Unfortunately, my queries doesn't seemed to work with stack overflow (too many data, I guess). So, if you want to build a query who works, feel free to post it here.

Note 2: Depending on the site, resultats aren't the same. Don't forget to switch to the site who interest you before running the query.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .