Death to the 'noreply' mailbox | Hacker News Reader

Death to the 'noreply' mailbox

josscrowcroft.com

244 pointsjosscrowcroft87 comments

"So I’ve given you my email address. I’ve given you the permission to contact me directly, via the medium I check more than anything (nasty habit, I know.) And this is how you repay me?"

Hear, hear. We've been preaching the same at Mailgun. A lot of companies pay good money for affiliate lists, scrape web sites; basically do whatever it takes to get emails for marketing/spam. Yet, here are companies with permission to engage their users through email and they totally whiff. EVERY email you send should be considered an opportunity to increase engagement with your users. Tell them them that they can respond with questions, comments, whatever. Grubwithus does a good job of this. They specifically say in the body of the emails that you can provide feedback just by replying.

Email should be considered another interface into your application.

I know this is totally not the right place to ask this, but what's happened to http://bouncr.com/ ?

It just plain doesn't work now, you can't register new emails and the MX for it (mailgun) doesn't accept mail for it.

But the website is still up.

> I know this is totally not the right place to ask this, ...

You might want to ask your question at a right place instead.

I've tried. I get crickets. I should have posted that in my comment.

I like the moo.com mails:

> Hello > > I'm Little MOO - the bit of software that will be managing your order > with us. It will shortly be sent to Big MOO, our print machine who will > print it for you in the next few days. I'll let you know when it's done > and on its way to you.

But the from mail it's donotreply@moo.com

The no-reply email is like sending your error logs to /dev/null

THAT BEING SAID

I used to be responsible for 600 active websites, and when we switched to using a help-desk system for auto-sent emails we broke things and had to go back to no-reply.

Lesson learned for us: it's worse to give the appearance of attention if you don't have the means to deliver on that promise.

>> it's worse to give the appearance of attention if you don't have the means to deliver on that promise.

Good call actually. In which case I think better to go with something like what Wufoo does, I believe theirs is something like "friendlyrobot" and if you reply to it, it sends you a friendly message explaining that nobody checks that mailbox, but there is a support desk they can contact. FTW.

Would be better if they didn't go to that trouble and just forwarded it on.

It's acting as a captcha — otherwise they'd have to deal with a shitton of automated replies and bounces from the customers and their crappy mailservers

Hate to subject customers to a captcha when they need support!

Surely it's the job description of customer service to deal with hassles in order to make users' lives easier?

Often times customer service == the development team == the marketing team == the executives == you

When you're a small shop, dealing with this kind of stuff day in, day out can get very taxing.

OK, so set up a filtration system. If you're good enough to build a system capable of running at that level you should be able to set up some basic mail filters or even a more complex bayesian filter to knock out at least bounces and probably a good percentage of autoreplies too.

Removing noise from a system isn't a job where it's perfect or useless; a substantial reduction is often quite good enough.

I don't see why it's necessary to make this complicated: "friendlyrobot" is the filtration system. As you said, it doesn't need to be perfect to be useful.

Bounces and auto-responders can be parsed and separated from human responses.

What was the problem? Junk? Autoresponders? Somthing else? Junk is easy to weed out if you run it through Gmail. Autoresponders can be managed for the most part with filtering (although there's room for dedicated tools here).

We were a six person department managing 600 websites, for musicians. The worst were for sites that catered to older clienteles.

The issue was staffing mostly. Unsolicited demo issues secondly and "faster horse" syndrome thirdly.

So what's the difference between using noreply, and using somnething like replybot and ... just ignoring those e-mails 99% of the time?

The implication of what the change in the "from" affordance meant to the user.

When we switched from "no-reply" to an actual address (help@artistdomain.com), and we couldn't respond, people got angry.

What is 'faster horse' syndrome, out of interest? Never heard that but it sounds good.

Given the ability to suggest things, users will always demand a faster horse, more sleep, more sex and to lose weight.

Basically its saying, don't solicit opinions because it all normalizes to "I want things to be better"

I thought sending notifications from help@atomiccontacts.com and ending them with 'Any problems, just reply to this email!' was a good idea too.

However, I've had to introduce noreply for a lot of transactional emails for a reason I didn't expect.

If you send an email like:

    Subject: X has requested you become friends/posted a photo/tagged you/sent you a message

...even if you mention that replying to the email will come to you, not X, many users will instinctively reply. E.g.,

    Hey X, So glad you got in touch. Love the picture. Want to grab a coffee soon?

Then you feel bad that you're intercepting personal emails.

I switched to noreply and put a help email link in the signature (which is ridiculed as the height of too muchery in the article).

If anyone's got a better solution, I'd love to hear it.

I can't recall which service, it's right on the tip of my tongue, but I've seen it done before that such a reply would end up being routed to the correct user. In the simplest case you get "X Commented on your Post!" sent from your service's account but with reply-to set to X's address. If you wanted to obscure X's address, you could process replies to your service's address and route them accordingly.

I think this is the case on Facebook. You can reply to emails about, for example, someone posting on your wall, and it will show up as a comment.

Do as Facebook (or GitHub) does - make replying to the e-mail reply to the person.

This works great when the purpose of the site is to interact with other people that you know. It doesn't work for eCommerce or content/newsletters.

That's what we do, using the Reply-to header, in case people hit reply. But the From header (which is what users see) still has to be one of ours. This is a requirement of SendGrid, AWS SES, etc., and it'd look pretty spammy otherwise.

Also, I just checked the latest notification I received from Facebook (someone posted on my wall) and its headers were

    From: "Facebook" <notifications+{code}@facebookmail.com>
    Reply-to: noreply <noreply@facebookmail.com>

It's a requirement of US law, in fact.

So then maybe you could give second thought to sending these mails in the first place. For me, they are the first thing I filter to /dev/null when I sign up for social-like service. I have in the ballpark of 300 people in my circles - so I just have to filter the updates like these in order to be able to actually use the mail account :-)

Maybe I am a super-extravert with a ton of similarly hyperactive friends, so take this with a grain of salt :)

I consider the no-reply address a courtesy to me to know that I won't end up writing a reply to an address that isn't monitored. Things like bank notifications and alerts are designed to be one-way communications, so the no-reply address makes sense. Usually the body or footer of the message contains the proper contact methods/addresses. It does not hurt my feelings to have to click on an email address within the message vs. clicking the 'reply' button in my client.

That being said, I agree that a no-reply address sends the wrong message for start-ups, welcome messages, signup confirmations, etc. I think they are good for recurring message that are inherently one-way, and when the user knows the preferred contact info for the company.

But any email they send out could be set up to reply right to somebody appropriate. Customers don't want a face of the company that only speaks and doesn't listen. Seems like they should be nearly never needed.

True, if it is a relatively small support operation. This would be harder to do with larger companies that had separate addresses for billing questions, service problems, troubleshooting, etc. If every outbound message had a singular live reply-to email address, then you'd need someone to sift through all those and route to the appropriate department.

If it's a large company, they should be more easily able to find somebody that can sort through the handful[1] of replies to that address a time or two a day.

[1] if there are more than a handful, all the more reason to do something with them - lots of people are replying when they "shouldn't" be!

Large companies should dedicate the thought and resources to get their customer human interface right. I hate that many large companies that expect customers to navigate through a morass of the companies half-baked organizational scheme. It's a terrible sign if a company expects a customer to coordinate or route an issue among different internal company departments.

Why can't companies be humane and communicate with customers via email? Every company parrots the falsehood that they are customer-centric and yet they can't figure out how to communicate with customers via email? Why is that acceptable?

I agree with the OP that all his examples were a poor use of no-reply addresses. My point is that they do serve a purpose sometimes - my example being notifications, reminders, and other inherently one-way messages.

If it is a message I am not expecting nor am I immediately aware of why it is being sent, then I think the reply-to address should be a real address capable of handing my response.

I was head of support for a good size startup and we set all of our reply-tos to go into our support system. There was not much junk (thanks Gmail) but lots of autoresponders. But it's totally manageable. I set up some hairy Gmail filters to try to weed out the autoresponders (would be nice if Gmail could help out more here).

My email account was also the "catch all" and even that wasn't too bad. And some good email comes in.

The thing that is astounding is that these companies otherwise spend millions of dollars to try to talk to these very same customers. Unbelievable.

This is my strategy. I almost missed some important client email by not checking noreply often enough, so our noreply now gets forwarded to those responsible for customer service and following up with customers. So far, very few delivery failures, and a few autoresponders. Autoresponders are important messages too however: if we send a booking notification to someone, and they're on vacation, we know we need to contact someone else in the company.

Nope.

noreply is for all the auto responders to not litter your inbox. At the end or beginning of the email is the contact address. A human responder clicks it and replies.

I agree 1080% about letting your customers easily contact you. That said, I think you completely missed why this terrible practice came about:

1) When dealing with a lot of automated emails, you get a lot of automated responses like out-of-office messages and failure notices (I know others have mentioned this). 2) Many companies have a support process that doesn't take place in email, because email doesn't scale very well.

So rather than just open up your inbox to thousands of junk messages and customer inquiries that you lose in the shuffle, I'd amend your call to action:

Get rid of the no-reply. Set a reply-to that goes into your support ticket system (you have one, right?). Set up rules/filters to automatically get rid of the auto-cruft and route responses to different mailings to the right place.

With that addition, I say: hear hear! Commit to listening to your customers and banish the reply-to! I know our team is going to after reading this.

Good points, and I like the amended CTA. I've added your comment to the end of the post.

Couldn't agree with this more. A big part of putting a service out there is getting in touch with your users and potential users.

If a user wants to engage/communicate with you, then you should be making it as easy as possible (Generally true, but particularly for an early-stage startup). The noreply seems brain-dead in this light.

Can't you solve the autoresponder problem by setting the correct Envelope headers? Bounces and automatic responders would go to noreply, actual people pressing reply would each you. Our am I too naive in assuming reasonably well-written autoresponders that know the RfC..?

Unfortunately, lots of people have vacation responders or autoresponders saying you have to click on a URL to get delivered...

Most autoresponders send highly predictable responses that can be reliably filtered on.

And when that filtering fails, causing an email to be filtered out as an auto-response even when it was written by a "real person"? What happens then?

1) This is rare but unfortunate. 2) Autoresponders are almost totally useless these days. 3) If this was handled like spam (at last by Google), it probably would not be a problem.

Well, I guess it depends what level of false positives or false negatives you can live with. I think you might be surprised at the diversity in email autoresponder formats, though.

I think the worst example I have come across of this is my employers e-invoice operator's support email address they use to dispatch error messages from. The error messages (which tend to be fairly huge) have a small disclaimer saying that they wont answer any emails and that you must call a support number that costs almost two Euros a minute. The worst bit however is that the address itself is support@einvoiceoperatorsname.com, how much do they hate their customers?

I used to monitor the reply address for a newsletter list with ~250,000 people on it. At this scale, a simple filter on "out of office" in the subject line really doesn't cut it.

To do it right, it's easily a full time job (and not a very fun one). "Noreply" addresses are bad for customers, but that doesn't mean they're always the wrong choice. As others have mentioned, it's better to bounce a reply than to accept it but not dedicate the resources to deal with it efficiently.

I used to work for a company that fed all replies to a custom CRM system which filtered out of office and bounces (unsubscribing on hard bounces automatically) and it worked great!

Those are enquiries from your (potential) customers. That's what you have support & sales people for!

It's strange that there are companies paying fortune to funnel people from AdWords to their sales enquiry form, but refuse to look at enquiries via e-mail.

> Some people have pointed out (fairly) that by not using noreply, you open yourself up to autoresponders and delivery failure notices. My response to that would be to go with an email address like friendlyrobot, which, when somebody replies to it, replies with a friendly helpful message explaining that nobody checks this address, and offering other addresses to contact.

Does this not run the risk of ending up in a loop?

I generally read 'noreply' as, "Here is something useful. And, we're not asking you for anything in return."

'noreply's are usually something you chose to receive. They're a service.

That's not to say that feedback should be discouraged. Just that, unlike the great majority of startup email, something is being given instead of requested.

I disagree with the logical premise of this article, at least for our specific case (social networking app sending notifications about activity). We most certainly don't want replies. The sender of the emails is not us, our company, but the application itself. When a calendar app on your smartphone reminds you of a meeting, there's no "Reply" button because having one simply doesn't make sense.

The fact that email is a system where it's implied that anything is replyable is unfortunate. Email is currently the best option for sending out notifications on the Internet, since most people don't know about feeds or how to use them. If there were a better, equally established technology for one-way notifications, we would use it. (We use SMS messaging, too, but only when it makes sense, since it's pretty invasive.)

Anyway, we did try to kill the no-reply behaviour at one point. Sounded good on paper, but it was impossible to execute in practice. Some issues, off the top of my head:

* We got a ton of vacation autoresponses, in all sorts of languages and all sorts of phrasings. It would be too time-consuming to set up rules to deal with all of the variations.

* We got an onslaught of replies accusing us of spamming them. These were users of our site, remember; a lot of sign up to an app and then promptly forget that they were now members. They never read the actual email, which has a very visible footer explaining how to disable notifications.

* A lot of confused users replied to the emails, either to thank us for the notification (not useful), or intended as a reply to a specific user, such as in the case of "new comment" notifications.

Sure, in some cases (like comment follow-ups) we could automatically process the rely. However, I'm really skeptical about our ability to handle email formatting well, though. Things like quoting, HTML, "On June 12th, John Doe wrote:" headers (which often are not in English) and signatures must be stripped, and I could not find an existing library that would do this reliably.

Maybe if you are Facebook or Google you can spend the necessary manpower to deal with the technical challenges, but we can't; our company is small enough that we can't even afford the resources necessary to screen replies manually.

I hate noreply as well, at least as it relates to paying customers. At Hover, we switched to a policy of using help@hover.com for the default reply-to on all mailers and it works quite well. Zendesk and our mail system do a great job of filtering out the automated messages and bounce backs and a human being gets the rest. The remainder are usually full of opportunity for us. Consider this Startups - every time you use no reply, you are throwing away an opportunity to solidify a relationship with a customer or potential customer.

> Hey asshole, what if I want to reply?

> What’s that? You don’t want crazies emailing you? You’re worried about spam?

Based on these lines, it seems like you're misunderstanding what "noreply@" means... They're not saying it's against the "rules" to reply, they're just saying nobody is monitoring the mailbox so it would be a waste of your time to do so. It's almost a courtesy, really...

You missed the point. The article is arguing that companies should put a monitored mailbox as the default reply-to, and not use unmonitored mailboxes, as they are missing out on valuable feedback and interactions.

You beat me too it.

I setup clients up with help@example.com when permitting. Turn those noreplies upsidedown ;)

I think for things where feedback is warranted, sure. But for automated email sending, alerts, notifications, things which may have very high bounce rates, using a real mailbox is basically spamming yourself or your support team.

I tend to use 'automated@' or 'a.robot@' though, to reflect that it was sent on behave of a process.

Your reply-to address doesn't have to be the same as the envelope-from on the message, so intentional replies can go to a different address than SMTP bounces. You don't have to spam yourself.

But you won't know whether feedback is warranted. You don't get to decide - the user does.

As an example, yesterday I got an e-mail where a template variable hadn't been expanded. noreply means they don't get to find out. It's worth my time replying to an e-mail. It's not worth finding a contact form and filling it out.

I understand that point, and it's valid. I was only saying that the lines I quoted, from the way they are phrased, make the author seem like he doesn't understand the reason why businesses use "no-reply" addresses to begin with. Which I know wasn't the purpose of the article, but I thought it was worth bringing up.

Sorry if I misinterpreted.

If you don't want to deal with e-mails as such and prefer that people contact you via a web form (hooked to some CRM), then add this to /etc/aliases:

     yes-reply "|/your/script/that/reads/email"

It will pipe incoming emails straight to a script that can decode and file them appropriately.

Over at http://halftoneapp.com we use speak@halftoneapp.com

When it works ;)

I hate seeing that evil no-reply email address. Completely agree that if companies want to better understand their customers, they need to leave the communication door open and allow candid responses to freely flow.

How many legit businesses would not include a return postal address on their snail mail? Why do people expect customers to go to their home page and look for a contact form? Joss has a point.

How many postal mailboxes do you know that trigger automated out-of-office responses and all the other noise?

Bounces do happen in postal mail, but 1) the PO usually has the option to discard marketing mail instead of bouncing it, and 2) the sender has a much higher incentive in wasted postage cost to avoid bounces in postal mail, as compared to an email bounce of a few packets.

I have a tiny company, and any automated emails always have written at the bottom "If you have any questions, simply reply to this email"

People are always slightly surprised by it, in a good way.

What's funny is that many of these same companies will later spam users with surveys to illicit feedback.

I hope you're volunteering for Facebook to use your email as their return address on everything.

Facebook already maps replies to meaningful responses in many cases.

Even if it were just the emails where they don't already map to a meaningful response...

With N million users, odds are high that at least N thousand of them: are irrational, just need someone to talk to, can't express themselves in writing, think you're their new pen pal, find you attractive (even if they've never seen you) and just want to meet you, etc.

I know that as users and customers we all want to be treated as beautiful and unique snowflakes. But a service of any scale will have many thousands (or more) of users that are just completely batshit crazy.

I don't understand why this has become a trendy pet peeve to have now, at least 5 years after email ceased to be a relevant form of communication. If you care about email, you've already lost.

Overuse of automated out-of-office respondes does far more damage to the medium.

How do you suggest companies communicate with their users?

i suggest that companies communicate with their users where their users are. which sounds like a useless aphorism, but i think it is actually useful to think about.

how many sites have added authentication based on your facebook/twitter/etc. account instead of your email address/password combo? it's happening.

I get and send dozens of important emails a day. Why is email irrelevant?

"A May 2011 Pew Internet survey finds that 92% of online adults use search engines to find information on the Web, including 59% who do so on a typical day. This places search at the top of the list of most popular online activities among U.S. adults. But it is not alone at the top. Among online adults, 92% use email, with 61% using it on an average day."

Source: Search and email still top the list of most popular online activities (http://pewinternet.org/Reports/2011/Search-and-email/Report....)

...But me and all my (hipster|geeky|20-something) friends only communicate using a combination of Facebook, Twitter, and SMS! If it's not relevant within my inner circle of friends, then 'anyone who's anyone' doesn't care about it!

i hear what you're saying. however, i think it goes beyond that - the phenomenon of people's parents using facebook is widespread.

email, in a non-work context, has become a communications channel for low-value broadcast messages.

  > If you care about email, you've already lost.

[citation needed]

http://www.allfacebook.com/email-use-declines-59-among-teens...

it's dying. it might be a slow death, but the patient has a terminal disease.

Beware of that line of though. Just because we don't like email anymore does not mean that this is the case for everybody.

jesus. get a grip. its a stupid bot email. if thats the only way you can communicate with your users is through their hitting reply to an obvious no-reply email then you've got other problems.

Hey. Genius. Some of us send over a million emails a week as a free service to users who are happy to hear from us. We just can't afford to check the damn inbox for that many people, so noreply makes it clear. Don't fucking reply because if we had to pay an agony auntie on the other end to feel your pain, the service wouldn't be free.

> Some of us send over a million emails a week as a free service to users who are happy to hear from us.

I find that very hard to believe, but probably I'm just missing the point. What kind of service are you talking about?

Facebook? Reddit? StackOverflow? There are more than enough free high-volume-email services.

Reply on news.ycombinator.com