"That's the point they [mailgun] lost me as a customer."
I would have been mad about the suggested $1k+ cleaning service too.
But, you do have to consider mailgun is protecting their reputation and assets as well. Like an IPV4 range that could get blacklisted if your diy cleansing wasn't right. And you mentioned your first run with the new provider was over 5% bounces, so it wouldn't have been good enough.
I'm less mad about the cleaning service and more mad about the pace of responses and the fact that I got 3 different responses from 3 different customer support reps who apparently didn't read any of the prior history in the thread.
I think the shutdown is totally valid and a good practice, but my initial ticket started with (paraphrasing): "ooooops, my bad for forgetting to migrate my bounces to the new domain. I've paused that campaign until I can clean it up." It still took almost 2 days to get reinstated after jumping through a bunch of unrelated questions.
Sure, it was on a weekend. I'm also slower at responding on weekends. But I also choose not to operate a mission-critical business specifically because I don't want to be on call on weekends. Maybe don't autoban people if you can't respond in a timely manner?
Follow-up thought: they just got acquired, so maybe there's some consolidation happening in terms of support or business processes that I'm not privy to. This was definitely more of a "straw that broke the camel's back" kind of response than abject outrage after the initial WTF response to the "clean your list before we re-enable you" message.
I started migrating while the service was down and they had given no indication that it would be turned back on any time soon. By the time my emails were reinstated, I was around 80-90% migrated. At that point, it's worth kicking the tires on the new service to see how deliverability compares.
Surprised this isn't at least account wide or maybe even global for all users? Are there really email services that bounce (not send to spam, but bounce) email based on domain? (Assuming you have SPF and DKIM and reverse DNS and whatever else configured correctly, which I assume you must to be able to use Mailjet?)
Yes, your domain can end up on a blacklist such as SORBS or Spamhaus and then you will have deliverability issues across the board. I have also heard from many companies and individuals with personal domains that have difficulty with Gmail and the verizon/aol/yahoo postmaster. They will just bounce you out of the blue.
DMARC and friends only help with authenticity. There's plenty of fraudsters sending authentic spam out there. The cost of a new domain is like $6, and there's plenty of people inputting their credentials on fake webmail signin pages for spammers that have trouble maintaining a credit card.
This is all true, but I don't understand why email providers will bounce personal domains. If you run DMARC and your domain is years old, what's the issue? I imagine that reduces the risk of spam by a lot.
I don't know why they don't just keep a master list of every bounced email across every domain (or maybe that's what their paid verification product is and I just haven't looked at it closely enough?) and let me set a flag when sending an email to reject anything they know is going to bounce.
We gotta give Mailgun some empathy here. It's tough to run an email service and keep reputation of it. If some bad actor got in, they can send large spam email and dilute the reputation.
Of course, this is just a mistake and after explaining Mailgun shouldn't charge $1358 for clean service. But as a customer, you should take some responsibility when doing something wrong as well. For example, if the account is old, and this is the first time this happen, and once we explain the mistake, Mailgun should waive that fee.
On AWS SES if the bounce rate >10%, your account is temporarily suspended.
If the exact samething happen with any mail provider, where a large of emails volume are bounced. They would need to pause/restrict your account in some way.
I would suggest look into AWS SES and started to write code to handle bounce email yourself to get a sense of it.
I would love to use MailChimp, but it was price prohibitive when I first got started. The site was using Mailgun when I acquired it, and I stuck with it after evaluating a BUNCH of options when I got up to around 30-50k users.
I don't recall what MailChimp's pricing was back then, but 50k users would be $640/mo today (vs ~$80 on Mailgun). I'm currently north of 300k users, which is up into the "contact us for pricing" range. It just doesn't make sense for a site with revenue sub-$10k (like, well sub-$10k).
I could've dropped that price by spending a shitload of time deactivating contacts and trying to keep my active users count lower, but email just isn't that profitable for me. I wouldn't say I'm great at email or attribution, but I estimate I drive ~$250-500/mo in sales through email (up from $0 when I started).
The other points are good ones -- even on Mailgun, I should've had different subdomains set up for IntroCave for transactional/campaign/newsletter blasts. I got "good enough" results doing it the lazy way, though. I still wouldn't have remembered to migrate the bounces, and this might have actually increased my bounce rate. Since bounce lists are per-domain and not per-account, a signup bounce wouldn't necessarily have prevented me from sending transactional stuff later (I've since added a bounced flag to my user model which would catch this).
(edit: actually the first point is pretty good. i disagree with the rest. list was mine, domain was warmed with a week of email before the ToS blast, and I manually ramped up volume on the ToS blast for the first 30 hours or so while keeping an eye on deliverability. This was entirely my fault, so I'm not sure what you mean by blame... I'm not happy with the customer support response, but the title of the post should make it pretty clear where the blame lies--me!)
Average ~$3500/mo, but that can go up and down depending on organic traffic, competition (who's running ads), and world events (2-3x'd for a month or two during peak Covid lockdown). It was running ~$1-1.5k for a good chunk of the time when I was building out my own newsletter system, and I wasn't super eager to throw 30-50% of my revenue at something off-the-shelf.
> I don't know yet whether Postmark will be any better, but I like the product so far and they seem to be a lot more customer-focused. Fingers crossed!
Postmark is definitely better in my experience with multiple customers. They actually check the delivery to different big mailproviders like gmail.
The support I had there was also quick and helpful.
It is definitely pricier, but should spare a couple of headaches if email is vital to your business.
Noob here (with his own mail server so I'm not familiar with these kinds of issues). What's the problem with a bounce? You pay for the delivery attempt anyway I assume, why do they care?
Mostly to do with making sure their service isn't being used for spam, to protect reputation. Bad reputation can mean blacklisting which is a bad time when your service is centralized on emails.
Sure, but how would allowing bounces lead to being used for spam? If it bounces, that means the mail was not delivered. Why are these errors bad for a reputation? It doesn't even matter whether the mail was spam or not, since it wasn't delivered anyway. That's what I don't get, of course I know they're trying to keep their reputation good.
Cool, email support@intromaker.com from your account email and I'll delete you.
(Which is the same offer I made to everyone else in lieu of an unsubscribe button.)
I did weigh whether to send a notice or not, and for me it came down to emotion more than economics. My thinking:
It'll end up costing a few hundred dollars to send it out. A lot of users have probably forgotten what IntroCave is or don't need it any more, and hopefully they'll request account deletion (hundreds of people did). I expected to generate approximately $0 in sales from the email, but if I can save even a handful of users from typing in the old domain, getting redirected to the new domain, wondering WTF, and bouncing -- that seemed worth it.
According to @SimianLogic, they had 300,000 users. If they are sending about 1,000 emails a day, that would mean that each user is receiving an average of 1.2 emails a year.
There is also a monthly newsletter that goes to 30,000. That means that in addition to the 1.2 emails a year each user receives, 10% of users receive an additional one (1) email per month. Given that only 10% of users receive the newsletter, I'm assuming it isn't too difficult to opt out.
This doesn't seem like spam. Rather, IntroCave (now IntroMaker) has hundreds of thousands of users and therefore needs to send thousands of emails.
These emails all go to users who have signed up for an account. Traffic is weird right now (domain migration in progress), but I was getting ~250-350 user signups a day.
I send a monthly newsletter to users who have been recently active, a welcome sequence of 2 or 3 emails depending on where you sign up, and have an abandoned cart sequence of 3 emails (here's the link to your video / your preview will expire soon / your preview has expired).
Both of those channels (newsletter / automated sequences) have opt-outs, but the automated ones see pretty good engagement. The max number of automated emails is in the 5/6 range before unsubscribes, so it only takes ~3-4 emails per user to get up to 1k/day -- especially when you toss in another 100 or so for normal password resets, receipts, order emails, etc.
CANSPAM defines it by describing the opt-in and opt-out nature of the messages.
What he did is upload a mailing list manually, rather than curate one with a service. This manual practice makes it almost impossible to guarantee that valid unsub requests have been honored, among other things. In this particular case, he admits accidentally including bounce out addresses.
That's why good email services make it difficult to upload 100k addresses and call it a day.
Additionally, permission to email someone with app notifications is not permission to send them bulk.
CANSPAM doesn't cover account/privacy/ToS notices -- it's not a marketing message (falls into "other content").
I didn't "upload a mailing list manually" -- the emails are all attached to user records in my wholly-owned database. I send them from a queue on my own server, not through Mailgun's mailing list product (which queues a message to every email on the list instantly and doesn't allow you to ramp it up over time).
"Newsletter" is easier to type than "new content and product updates," but again -- these go out to existing customers and account holders. There's implied permission there from the fact that they signed up for an account. Opt-out is easy, and I don't feel guilty about emailing people who signed up for an account.
I would send this to the whole list if deliverability didn't go to hell after about 50k users (or maybe I just need to write better emails to keep people engaged/opening).
My reading was that he actually he HAD an opt-in/curated list just hadn't removed the bounces and other detritus the list had generated over time. I didn't get any impression that he was spamming people, the problems were all bounces gmail/yahoo "mark as spam".
I would have been mad about the suggested $1k+ cleaning service too.
But, you do have to consider mailgun is protecting their reputation and assets as well. Like an IPV4 range that could get blacklisted if your diy cleansing wasn't right. And you mentioned your first run with the new provider was over 5% bounces, so it wouldn't have been good enough.
I can see why Mailgun pushes that solution.