Best Practices on reply-by-email


(Jay Pfaffman) #1

I’m working with a site where I’d enabled reply-by-mail, but I’m thinking now that it’s not a great idea. Here are a few reasons that I have for disabling reply-by-email. Am I missing something?

  • if users can reply by email and can interact with the site without visiting it, then we’re losing ad revenue, although small.
  • I was concerned that disabling reply-by-mail would also disable being able to email in to a group, but in my test that seems not to be the case. (Unless there was some caching problem and I did my test before the setting was really off).
  • Things like “your message is too short, try again” are even more annoying in email than they are on the site. (Of course they’re “annoying” only in the rare cases where someone really did have some reason for a short message and just a :heart: wouldn’t do.)

Is there some reason not to disable reply by mail? It’s a convenience that I use very rarely here; I think that for our community it’ll be OK to turn it off. Right?

And on a tangentially related note . . .

Another problem that we had was a rookie mistake . . . I had a publicly advertised address getting delivered to a group. . . and that address got picked up by a 100X/day spammer. If you’re going to have an address get sent to a group (or category) by un-confirmed users, you want to make sure that you’ve got a spam filter!


#2

The main reason it’s enabled in my community is to retain the old-school life-long listserv members who otherwise would not participate at all.

It would be great to have a clear picture of how much exactly they contribute to the conversations. I’ve asked for better post by email reporting in the dashboard.


(Jay Pfaffman) #3

Here’s the beginning of a data-explorer query. My intent is to do something like percent posts via email per user for each of the last 4 weeks, but my SQL-fu is a bit short of that query just rolling off my fingers.

SELECT   user_id,
count(1) as post_count
FROM posts
WHERE via_email=true
GROUP BY user_id
ORDER by post_count desc

(Richard - DiscourseHosting.com) #4
select username, sum(case when (posts.via_email='t') then 100 else 0 end)/count(posts.id) as perc_by_mail
from posts
left join users on users.id = posts.user_id
where posts.created_at >= now() - interval '4 weeks'
group by username

That sounds very short sighted. Having more content will bring you a multitude of visitors.


(Jay Pfaffman) #5

I guess that’s why I’m asking! With this particular community, it seems that people who reply via email (as opposed to on the site) are those most disenchanted from the move from Ning.


(Mittineague) #6

I have never made a post via an email, so I’m admittedly clueless. But are there any situations where someone might be able to email but not have a browser? eg. at home I have my desktop browser, but on a long commute I can only send emails. Not a matter of preference, but a matter of ability.

In other words, not that I am participating only with email, but that I participate both ways.


(Jeff Atwood) #7

It’s quite rare for any site to have a ton of email-only participation. 25 years later, people are pretty darn comfortable with the concept of a web browser :wink:

The reality is that email replies capture a small percent of the time people are “on the go” and want to reply but all they have access to is email. It is a convenience feature.

If you turn it off, you turn off convenience – but it’s not essential to survival in my experience. I’d still recommend it though.


#8

In my community - yes. Our topics revolve around aircraft construction and flying, so a post will often be seen by a member while they’re at the hangar, away from the desktop browser.

Email is also pushed, so they are engaged on a daily basis. There are those who prefer email to stay engaged even though most have a smart phone and the Discourse mobile interface is good.


#9

Richard, do you have a query to find the count and percentage of all posts made by email vs in browser?


(Richard - DiscourseHosting.com) #10

Per user:

select 
  username, 
  sum(case when (posts.via_email='t') then 100 else 0 end)/count(posts.id) as perc_by_mail, 
  sum(case when (posts.via_email='t') then 1 else 0 end) cnt_by_mail,
  sum(case when (posts.via_email<>'t') then 100 else 0 end)/count(posts.id) as perc_not_by_mail, 
  sum(case when (posts.via_email<>'t') then 1 else 0 end) cnt_not_by_mail
from posts
left join users on users.id = posts.user_id
where posts.created_at >= now() - interval '4 weeks'
group by username

All:

select 
  sum(case when (posts.via_email='t') then 100 else 0 end)/count(posts.id) as perc_by_mail, 
  sum(case when (posts.via_email='t') then 1 else 0 end) cnt_by_mail,
  sum(case when (posts.via_email<>'t') then 100 else 0 end)/count(posts.id) as perc_not_by_mail, 
  sum(case when (posts.via_email<>'t') then 1 else 0 end) cnt_not_by_mail
from posts
left join users on users.id = posts.user_id
where posts.created_at >= now() - interval '4 weeks'

#11

Superb - thanks, Richard!

The result is 19% of our posts were made via email in the last 4 weeks.


(Peter N Lewis) #12

We have some long standing (and long suffering!) users who swear by reply-by-email which is why I would never turn it off. I doubt it makes up a large percentage though.

I’d be curious to know what the stats are, how does one run the SQL query on the Discourse database? Thanks!


#13

Very easy, just install the Data Explorer plugin: