New Insight: Time to reply

[adampash]

One-liner

Time the user takes to reply to others' posts.

Full explainer

More often than not, social media is a conversation, and in most conversations, response time matters. This insight gives the user a sense of her response times on her social media.

Audience for the insight

• Which networks, if any, are excluded from this insight?

This insight should work on Twitter and Facebook. Prob doesn't make sense for Instagram.

• Does this insight serve users with less or more activity?

Serves both, but I suppose more reply activity is better.

How often this insight runs

• [ ] Always (triggered by a data event, check the data and update insight on every crawl)
• [ ] Daily (check the data once a day, don't regenerate every crawl)
• [x] Weekly on a fixed day (specify day of week per network to space these out)
• [x] Monthly on a fixed day with bonus magic day within first 2 weeks of use (specify day of month per network to space these out)
• [x] Annual on a fixed date (specify day of year)
• [ ] Annual every 365 days from first appearance

Weekly, let's do Facebook on Saturday, Twitter on Sunday.

Monthly, let's do Facebook on the 2nd, Twitter on the 3rd.

Annual TBD (end of year).

Headline

• %username's quickest %reply_type this %period

Emphasis: Medium

Body

• On average, %username %reply_type-d [on|to] %post_type-s this %period %avg_time after they were posted. [if baseline]That's [faster|slower] than last %period's average of %old_avg_time.[/if] The quickest time-to-%reply_type, at %quickest_reply_time, was:

  [embed %quickest_reply]

  The longest time-to-%reply_type, at %slowest_reply_time, was:

  [embed $slowest_reply]

E.g.:

• On average, @adampash replied to tweets this week 3 minutes after they were posted. The quickest time-to-reply, at 2 seconds, was:

  [embed quickest tweet]

  The longest time-to-reply, at 30 minutes, was:

  [embed slowest tweet]

  Criteria and logic

Describe the rules for when this insight runs.

• What data does this insight need? Last week's posts? Last month's? Just the user's posts?

This insight needs the previous week/month of the user's replies, then examine the time difference between the reply created_at date and the created_at date of the post it was in reply to. (I believe this comes from twitter via in_reply_to_status_id?)

For Facebook, it would require comments, but it looks like we can only compare comment response times to the parent, which is either the original post or another comment in the thread. That's mostly fine, but it doesn't seem as fine-grained as the Twitter version.

• Are there baseline comparisons? What is the logic around the comparison?

Baseline compares latest response time to previous response time for the same period.

• Are there special copy cases? (for example, if the baseline comparison matches)

If there is no baseline, exclude the "compared to..." clause. Also, if it's monthly, should say "last month", since it runs at the beginning of the month. If weekly, should be "this week", since it runs at the end of the week.

• Is there a minimum threshold for any bit of data before the insight should get generated?

User should have at least 3 replies/comments in this period.

Included elements

• [x] Headline
• [x] Text
• [ ] Header image (image off to the left in side-by-side style insights)
• [ ] Hero image (giant image on top)
• [ ] List of user(s)
• [x] List of post(s)
• [ ] List of link(s)
• [ ] Action button (please specify button label and URL button should link to)
• [ ] Line chart
• [ ] Bar chart
• [ ] Other viz
• [ ] Other graphic treatment

  Priority

• [ ] High
• [ ] Medium
• [x] Low

[ginatrapani]

Question: should this insight supersede our current response time insight? It's much less detailed than this. Example: https://ginatrapani.thinkup.com/?u=Gina+Trapani&n=facebook&d=2014-09-15&s=response_time

[adampash]

Real talk! That response time insight has always confused me!

They have similar names, but as far as I can tell, they measure different things. This one is about your replies/comments on other people's posts. That one is about rates of other people's responses (likes/faves, shares/retweets, comments/replies) to you.

FWIW, my confusion with the other response time insight is the fun-fact body copy, which always feels like a confusing non sequitur. Generally, both the headline and the body provide stats that I have no reaction to.

[ginatrapani]

Fully agree the current response time insight is lame and weirdly written and repetitive. Let us replace that one with this one.

[ginatrapani]

Okay, reading this more closely, I misunderstood the premise of the insight.

(Relatedly, let's try to make that one-liner do a better job explaining what the insight is, ala the existing insights https://thinkup.thinkup.com/account/?p=insightsgenerator )

I thought this insight counted the shortest and longest times it took others to reply to the user's posts, as in, I got a reply to my posts within 2 seconds, and then after 3 weeks. I like this idea because it gives you a sense of the shelf-life of your posts. The value here is I know how long to wait until a post is effectively "dead."

On second read, I realize this is the quickest/longest time YOU take to reply to OTHERS. Trying to figure out what the value is here - is it to see how fast on the trigger you are? What does that tell you?

If we're doing the latter, this is NOT a replacement for the current response times insight. If do the former, it is. I prefer the former. Thoughts @anildash @adampash ?

[adampash]

My thinking was that it gives you a sense of how you use the network. In my mind, it gives you insight into how intensely you're using the network. Similarly, I think that iff your baseline is rising or dropping gives you a good week-over-week insight into the way you're using the network.

I agree that the other direction is valuable (the shelf life of your posts). I'm fine shifting this in that direction, though. In fact, I think one really good argument in favor of changing it to how quickly others reply to you would be that that's way more shareable, specifically when we call out the fastest and slowest responses for that period.

[ginatrapani]

My thinking was that it gives you a sense of how you use the network. In my mind, it gives you insight into how intensely you're using the network.

Yeah, I get this... I wonder if it really does tell you that in practice, though. On Twitter, typical usage might be: load up your timeline and reply to whatever is most provoking at the top, without necessarily scrolling back, so that could look like you're a quick replier, but you're actually just picking items out of the freshest part of the stream. On Facebook, with the news feed algo often showing older items getting a lot of play at the top, you might respond to something kind of old but new-to-you. Anyway, this is a question worth discussing with the team.

I do think the "shelf-life" insight is worth doing as an upgrade to the current response_time insight. Can we file that one separately, and decide what to do about this one here?

[adampash]

Haha, you blockquoted me repeating myself. How dare you!

Re: Twitter and Facebook habits: Agreed, though even seeing the difference in how that works b/w Facebook and Twitter would be interesting to me (maybe this could be the first insight for only people with Facebook and Twitter connected!).

Anyway, I also agree that this insight in the other direction is probably more interesting for more people.

I'll write up the shelf-life insight today.

[ginatrapani]

Ha! I didn't even see that I blockquoted your repetition! That was less passive-aggression and more cluelessness than it seemed.