Hacker News Reader

Comment by latitude | Hacker News Reader

That's my post there. It'd be very interesting to compare email confirmation rates. I am running on a private mail server that is configured according to all possible spam filtering guidelines and yet I am still at only 82%. I wonder if these 18% is just people overwhelmed with emails in their Inbox or if there's still some issues with being mis-categorized as a spam.

Anyone else cares to share their numbers?

View on HN

No scientific data here, but it sounds to me that 82% is pretty damn good. You'll never reach 100% because there'll always be folks who sign up and then simply don't follow thru. I know I'm guilty of that. It's like handing out Broadway show flyers... just because I take one doesn't mean I'm going to actually buy tickets to see the show.

You should be able to get to 95%. On the form there is no indication that an action is required after entering my email address. I also see a generic "support" as the sender in gmail.

Go, click on that link already. You are skewing my beautiful stats :)

Good point regarding the sender name. I will play with it once HN wave subsides.

I am consistently running around 75%, with subject line "Confirm your email address". I am going to change it to "Please confirm your email address".

I send the confirmation emails through Google's email servers (Google Apps).

update: If I break it down by domain, I see gmail->80%, yahoo->70%, and hotmail->66%.

What's the best wording for email body with confirmation link?

Currently I use:

====

Please open this link to confirm your email:

http://www.postjobfree.com/a.aspx?k=c3d26a0fc5d145cf88898a92...

This link will expire on: 8/19/2011 10:29:11 PM EST.

====

Thanks for the tip about subject for email confirmation!

In the past I was using "Link to confirm your PostJobFree email", but your new version with clear call to action ("Please confirm your email address") should perform better.

Please spell out the month name. In many countries dates are formatted as DD.MM.YYYY rather than MM/DD/YYYY. For example, a native German speaker will have a lot of trouble recognizing that date. Especially if it is something like 8/7/2011 (is that August 7 or July 8?)

If you write the date as "August 8, 2011" it should be trivial to parse for everyone. (Except for people using different calendars)

Not only native Germans, other native English speakers, like people in the UK use the DD/MM/YY format. Unlike people from Germany, people from the UK have no language barrier, and are a potential customer.

Yes, you are totally right. But don't overrate the language barrier: I sell a software product that is English only (localization is too expensive), and the majority of my customers are from non-English-speaking markets.

Surely you don't need expiry to be accurate to the second. Also, most people in the world have no idea where "EST" is (or even what it means).

Why not just say "The link is valid until August 19", and then let it expire some time after all the world has moved on to the 20th? (No MM/DD or DD/MM. Please.)

I personally prefer the old fashioned "Just reply to this mail" method, which many mailing lists use for confirmation, to opening a browser window - in the worst case I have to wait for my browser to launch, too - and I'm going to close that window immediately anyway so I can get back to what I was actually doing, so I just think "I'll get to that later", but then I don't, and when I remember [did I click that link? I'm pretty sure I did, so there's probably a bug in their list somewhere and I shall ignore them henceforth], the link has probably expired anyway [can't be bothered checking what time and day it is in "EST" (Estonia?)] so I'll just browse around for funny kittens instead. And all of that just because somebody wanted me to click on a link. They're sooo cute!

Maybe that's just me, but everytime I see a "just reply to this email", I stop and wonder about the details. Do I have to put something more in the body ? Will my mail go through ? Is it automated, or will a real person on the other side process my answer ?

Maybe it's just that the email way is too rarely used, but it makes me pause. I prefer a link, that's much more of a no-brainer.

Why would you browse for what EST mean if you can simply open confirmation link and see if it works?

But the suggestions overall are valid - thanks!

What's better:

1) "The link will expire 2011 August 19"

2) "The link is valid until 2011 August 19"

3) "The link is valid for one day until 2011 August 19"?

"The link will expire after August 19"

This gives people a sense of urgency, but doesn't discourage them from trying the link if they're a little late. "Through" is (to my ears) an Americanism.

"valid through August 19"

I've always found "through" to be confusing when used like this. It's like the whole "Next Friday" thing - is that the one about to happen (it's the "next" one to occur), or is it the one after that? (in Australia, it's usually the later) I much prefer "valid until August 19"

(I'm a native English speaker, not from the US where "through" seems more common)

Right. "Through" is better than "until" because "until" leaves the reader unsure whether it includes August 19th. And 2011 is unnecessary; if someone does find an old message, they can look at the Date: header.

Personally, I would say "August 19th", not "19".

Agreed, especially with the dropped year.

If the duration is less than three months, the year can be inferred from the date of the email.

You should drop the bit about the expiration. It requires someone to do a timezone calculation to even figure out what it means, and you probably want them to click the link even if it has expired, then you can give a proper support message with what they need to do.

Agreed on the formatting, but I bet that having that expiration date on there increases the CTR.

I think it would be even more user friendly if you included a more readable expiration. Something like "this link will expire in 2 days (8/19/2011 10:29:11 PM EST)". This gives the user a more understandable deadline, without having to check the current date

Exactly, and only, what this is: a subject with a clear call to action instead of a subject that seems to indicate that this email was generated as an auto-response to a previous action and contained a body that did not require user interaction.

If you are looking for confirmations... perhaps a percentage just didn't use valid emails? Possibly accidentally?

Yep, there are dupes and there are mistyped addresses. But plenty of unconfirmed @gmail.com (about 50%) as well as addresses personal servers (on judging by the domain names).

What about:

Next Step: Confirm your email address

Thinking along the lines of how "Follow me on Twitter" tests better than "If you liked this blog entry, please follow me on Twitter." Just give people a direct order. "Next step" might make it clearer that they're not yet finished with the process?

Interesting idea. Having said that, "Next Step" makes it sounds like a longer process. It doesn't seem like I won't be quite done after this.

I think rewarding the user for singing up would be an interesting incentive. "Click the link to receive your first email".

Having said that, unlike registration confirmations, where I would like to get back out of my inbox to your site. For mailing lists, I don't need to go back to your site, really. I'd much rather stay in my inbox if that's ok. I'd quite like to see an A/B test between the usual method and a request for people reply to the email, rather than click a link. "Reply to this email to start receiving updates": simple order, direct resulting action. It also makes it sound like a similar action will be required to stop receiving updates.

True. What about: "Final step: ..." or "Confirm your subscription to receive your first issue"?

Wonder how many people are using web-based mail clients and whether that would impact a preference to reply or click?

"Awaiting email confirmation"

Are you going to email old subscribers (who didn't confirm) a reminder with the new subject line?

Ha. That's a nice idea. Let me see if it works :)

Are you tracking emails that bounce? Or people that have whitelists of who can email them? Then there's the people who always send signups to a special yahoo signup address they rarely check. You can probably do a little better than 82%, but I doubt you'll ever beat 95%.

Does simply seeing 82% vs. 50% enough? Shouldn't there be some significance measurements added to it? If you had only 6 people in this experiment maybe it is just a statistical error...

If there were 6 people, he wouldn't be saying 82%. 83%, maybe.

Reply on news.ycombinator.com