# When is the best time to post a challenge for optimal views?

I've posted a few challenges before in PPCG, and, judging by the minimal view counts I've been getting (especially recently), I'd say I've been posting them at the wrong time.

Is there an objectively identifiable "best time" to post challenges in order to get the most people to see it (and therefore not let it slip into the void of unanswered challenges)?

• I also thought about this in the past, usually I orient myself on the curves of the chat activity. Perhaps some skilled people could make one of those SQL queries to extract for each challenge in e.g. the past year: UTC timestamp, day of the week, number of votes, number of answers that we could analyze further. Mar 19 '16 at 12:57
• @flawr That'd be really cool - I'm not good enough myself to do that. Mar 19 '16 at 13:09
• I don't know about time but mondays-wednesdays, seem to get the most views on average. Mar 19 '16 at 15:05
• Also, challenges get more upvotes/views when they are easy enough for HNQ vistors to understand and vote, but not easy enough to get downvotes. Mar 28 '16 at 17:56
• In the middle of the day when devs have nothing to do. Mar 30 '16 at 22:59

## When is best for views?

As it turns out, it doesn't matter all that much. Consider this plot:

The mean number of views is slightly higher on Mondays and Wednesdays but things are pretty consistent overall across the whole week.

Also doesn't seem to matter.

## This is good news!

We need quality content everyday, all the time, not just on certain days. The fact that we have fairly consistent views is a good thing.

## When should you post?

Whenever you want, but note that targeting a specific day or time likely won't help.

## For the curious

Stack Exchange Data Explorer query:

select Id,
ViewCount,
datename(weekday, CreationDate) as DayOfWeek,
datepart(hour, CreationDate) as UTCHour
from Posts
where PostTypeId = 1
and DeletionDate is null
and CommunityOwnedDate is null


R code to produce the plots:

library("ggplot2")

posts$DayOfWeek <- factor(posts$DayOfWeek,
levels = c("Sunday", "Monday", "Tuesday",
"Wednesday", "Thursday", "Friday",
"Saturday"))

ggplot(posts, aes(x = DayOfWeek, y = log(ViewCount))) +
geom_boxplot() +
ggtitle("Log Number of Question Views by Day Posted") +
xlab("Day of the Week") +
theme(text = element_text(size = 12))

ggplot(posts, aes(x = factor(UTCHour), y = log(ViewCount))) +
geom_boxplot() +
ggtitle("Log Number of Question Views by Hour Posted") +
xlab("Hour Posted (UTC)") +
theme(text = element_text(size = 12))

• What's with all those resistors connected in parallel? Mar 22 '16 at 12:24
• @DonMuesli That's a good point. You know what they say, resistors are futile. (Or something like that...) Mar 22 '16 at 17:12
• Futile, and useless! Mar 22 '16 at 17:29
• Mmm... view counts are just an int that grows whenever a post is seen, without any indication of the time it was seen. What you seems to measure is that in average, the average of posts always get the same amount of views whenever they are created at one day or at one hour. Mar 27 '16 at 20:07
• @Braiam Yep. The dates/times of the views aren't recorded, otherwise one could match it back to the date/time of posting and look for short-term trends. Mar 31 '16 at 3:50

# Fundamentally, there is none

I don't have specific data but I can say from experience that a question posted on any day at any hour can become popular, can become a Hot Network Question, you just might have to wait a bit.

I've found (or at least it seems) that there are far fewer answers posted when it is nighttime in America, but posting then is fine since within 12 hours most everyone in those time zones who would have wanted to see a new question directly on PPCG (and not HNQ) will have seen it and answered if they wanted. At that point it can still get on HNQ to get more views and answers, so the exact hour of posting is not incredibly important.

There does seem less activity on the weekend, but presumably that goes for all of SE, so your chances at HNQ are not really any worse.

And remember, it's not like questions go anywhere if they have few views or few answers. Their number of views can only go up. Completely unanswered questions are fairly rare, we have around 115 questions with zero answers, of 5127 total.

I'd personally say that a question's title, and of course it's actual content are far more important to worry about.