Twitter, or how a community of 400.000 is reduced to 20 poor quality replies | Hacker News Reader

Twitter, or how a community of 400.000 is reduced to 20 poor quality replies

onderhond.com

23 pointsonderhond7 comments

Basing the reply rate on the number of followers that an account has is fairly nonsensical. One the one hand, you have bots and inactive accounts, and essentially read-only use cases, and on the other hand the tweet went out to more than those 400,000 followers when you take into account retweets.

For example, I tweeted about the completion of my Toy Story 365 set to my roughly 40 followers last year: http://twitter.com/#!/Alanfalcon/status/9318186977595392

I also @ mentioned some influencers who might care about that sort of thing (Lee Unkrich, Disney/Pixar, the Pixar Blog) and so my tweet got retweeted by Lee and Disney/Pixar, which means the number of followers to whom my message was broadcast was more like 1,000,000 than 40. That day I had 23,874 hits on my flickr page (see the screenshot on my blog here - http://www.syncingdreams.com/2010/12/reward-for-project-comp... ... note that those numbers are not 23,874 uniques–many people clicked through all 365 photos, many more clicked a handful of them). I also had a dozen re-re-tweets and @replies to myself on Twitter, and maybe 6 comments added on my flickr stream itself.

Would you say that my community of 40 (43 including the @mentioned folks) resulted in 6 decent quality replies? 10 replies (including both flickr comments and twitter comments)? Or that my potential audience of more than a million resulted in just 10 replies? How would you determine the potential audience size anyway? By adding all the followers of those who retweeted my link? Do you go through and remove all the duplicates who are following more than one re-tweeter?

Anyway, all of this to come to the conclusion that the numbers presented in this post are pretty much useless without a lot of context, and also to present my own useless numbers as a kind of counter-point.

There is also no way of knowing how many of the 400,000 actually saw the initial call to action.

I'd be very interested to see more detail on this analysis, such as number of click-throughs, number of active Twitter accounts amongst the 400k, etc.

Smashing Mag is a good site, but it also posts very frequently and has a lot of poor-quality content where it's completely fine to ignore a lot of posted links. So I'd naturally expect a much lower amount of responses than, let's say, A List Apart.

Catching someone's eye on Twitter adds to this dilemma. I use TweetDeck, meaning at most I can see 7 tweets in my 'All' field at any time.

Sometimes I scroll down for a while, and sometimes I leave Twitter open and check new tweets as they come in. But following a couple of hundred people means most tweets are part of that magic 7 for less than a minute - that's not a lot of time to catch someone's attention enough for them to click through to an article, reply to your question, even retweet.

Some more stats from Smashing Mag (or another test from someone with a large following) would be useful. Counting hashtags and testing for quality replies misses those who reply without the hashtag, and overlooks other engagement metrics.

Good to see the dilemma highlighted with some data though.

Thank you very much for the very interesting study. The truth is that your findings are actually fairly correct. The problem is, however, that the vast majority of those who reply, do not use the hashtag. They just don't. There are usually more, much more replies that the ones with the hashtag, although relatively speaking it's still not what you probably would expect.

A more accurate average number of replies can be estimated if you consider how many users participate in a poll:

http://twitter.com/#!/smashingmag/status/62126045498839040

This poll was published on a low traffic weekend day which is clearly reflected by the number of participants.

Polls published on weekdays have of coruse a better response ratio:

http://twitter.com/#!/smashingmag/status/55762312174370817

http://twitter.com/#!/smashingmag/status/54897184239398912

http://twitter.com/#!/smashingmag/status/54221240826732544

We never get over 1000 replies to a single tweet, but 300-500 replies is a very reasonable number, especially if it's accumulated over a couple of days.

Just to clarify - it wasn't my study. I was just pointing to some flaws in the conclusion based on methodology, and noting the there was some data only Smashing Mag could provide - thank you so much for doing that!

Seems like the kind of response rate you'd expect for a traditional publication, which isn't surprising given that it doesn't seem like the tweeter has any kind of real relationship with the followers - they are just using it as a twitter version of an RSS feed.

I've seen this on some feeds we run - the response rates on Twitter are very low, especially because most folks don't expect their reply to be read by the other followers of the original account. Contrast that with Facebook where the followers of a page can see each other's responses and our pages with ~25K followers regularly see 150-200 responses to each post - and they post at least 6-12 times a day.

Reply on news.ycombinator.com