Mailchimp to Sendy: Cutting Email Costs | Hacker News Reader

Mailchimp to Sendy: Cutting Email Costs

victorzhou.com

191 pointsvzhou84296 comments

There is a deep misunderstanding of the intricacies involved in actually delivering an email to the inbox, repeatedly, with the lowest risk. What you are paying for with ESP services is mostly that expertise.

Yes, but you've read Sendy is connected to SES right? So you're still getting the benefits of solid deliverability. Or are you saying SES deliverability is not as good as Mailchimp et. al?

SES is as good as Mailchimp and very inexpensive. Too bad Mailchimp phased out Mandrill, which was reasonably priced and had a nice interface. SES has no interface at all but works great.

This. I will never touch mailchimp again after what they did to us when they shut down (priced out) mandrill.

If you like SES you should check out Email Octopus. It hooks up to SES and gives you a nice GUI for putting the messages together. We moved a good sized client over to them and their marketing people wanted a GUI and it filled the niche nicely.

There are many such interfaces - including free options like phplist...

We used to use Email Octopus and switched to Sendy. We like Sendy's interface a lot more.

Most of the plain ESPs like AWS SES, or mailgun accomplish that same reliability for comparatively peanuts.

The main thing you lose with them are UX for sending newsletters, handling bounces on those lists etc. that's what sendy covers.

There's also GDPR/CAN-SPAM/CASL/etc compliance and abuse handling. Those can be costly in terms of time, money, and access.

These can be costly if you're using Mailchimp et al. Mailchimp's compliance features only handle compliance within the Mailchimp lifecycle, not at the entrypoints. The primary part being subscription management. Using Mailchimp doesn't guarantee compliance.

I think I'm agreeing with you, but I think that 99% of people who end up using AWS SES will find they'll need to write a bunch of handling code on top of it to deal with bounces (i.e. if your bounce rate is high, SES will kill your ability to send emails).

I found it incredibly frustrating that there was functionality that should have been easy for SES to implement, and that pretty much everyone will need at some point, but they don't really make it easy to add on after the fact.

It sounds like a great idea for a potential feature request for Sendy and software like it.

SES uses SNS for bounces, complaints etc. So _really_ you need a http(s) endpoint that receives these notifications and can act on them in a way that you'd like. (Unsubscribing from a mailing list, analytics, etc).

Sendy handles bounces. By default it stops all future transmission to that address, IIRC.

There are free alternatives to Sendy et al.

I've used phplist feeding to AWS SES for many years now without issues.

I came here to say exactly that ... with the addition of the fact that MailChimp is run as an enterprise service while hosting your own on a $5 DO machine is clearly boutique. I'd like to see an up-time comparison as well as the delivery % noted above after 5 years. You're now also in charge of making sure your emails aren't considered SPAM and dealing with the ramifications of people who signed up for your newsletter forgetting and thinking that your emails are unsolicited.

AWS SES does all of that. With the exception on handling unsubscribes. But any cheap or free mailer software worth anything does that as well.

You are paying for glitzy UI, marketing, and some nice features as far as email themeing and then of course, support.

The biggest problem is large companies blocking small servers... otherwise it would not be that hard

Sure.

Except MC doesn't provide this one, mission-critical feature of inbox delivery.

Most of my friends and family that subscribe to the PhotoStructure newsletter found their email in their spam folder.

And yes, I set up all my domain records properly, at least according to MC, and there aren't any warnings or diagnostic failures on their side. Each campaign has said there was 100% delivery success.

Delivery success is usually based on the download of a tracking pixel. Problem is, Gmail downloads and proxies all images (including those tracking pixels). So anything that doesn't get completely filtered out before showing up in the inbox or spam box is going to be seen as -received- by, as far as I know, any software out there.

There are quite a few messages that never even make it to the spam box, which I have never understood because plenty of very obviously sketchy spam messages get through to my spam box. Yet, occasionally we have trouble getting very legitimate transactional emails through to some customers.

"One downside of using Sendy is that you have to setup / host it yourself. "

This is the classic mistake of calculating savings purely on costs and not factoring in time.

Pay $75 per month for hosting and 0 hours on Ops.

-or-

Pay $5 per month for hosting and 5 hours on Ops.

The latter option values your time at $15 per hour. Basically just creating a minimum wage job for yourself.

Or spend 2-5 hours a year on Ops, and suddenly it looks more attractive again. You get to decide where on the range your personal preference flips, where your available skills probably put you on it and choose accordingly.

Is it a one time thing or a weekly/monthly/yearly thing?

you multiply that 75$ figure times 12 (you won't spend 5h/month on a firewalled webapps ops) and you are valuating your net (!) wage at about 100$/h (and from my experience standalone/not-internet connected "it works"-solutions have a half life well beyond any arbitrary fiscal horizon). While that might be in the range of a FB-SWE, for most "normal" private ventures the mailchimp "cloud"-offering just isn't gonna be worth it. Though one might argue, what is so private about a 10000 recipient "private" self-marketing campaign...

The other thing to think about is that with self-hosted applications -- of which I am a big proponent -- typically any extra hours of maintenance work are usually going to hit when you least expect and/or want them to, id est during some kind of failure mode.

We put Sendy on a server that we already have in use and have spent maybe 1 hour on admin work for it per year since. (Occasionally a new version comes out and you have to copy the files to the server). Plus if we had gone with Mailchimp, it's not like anyone is going to get paid 1 hour less.

> The latter option values your time at $15 per hour

Makes the classic mistake of assuming your time is worth anything. If nobody will pay you money for those 5 hours, you're just picking a random number to justify either spending money or not spending it. May as well say it's $200 or $2 million.

I mean, you could just as easily use a real reference for what it would cost.

You can either pay mailchimp, or you can pay your senior developer to set it up for $80/hr.

Because you can pay someone to do something for $80/hr does not mean your time is worth $80/hr.

It's worth noting that there are a few open source email marketing platforms out there too. Sendy is $60 and obfuscates some of the source code.

Mautic and Mailtrain are two that I have tried. In my experience, both have some odd areas, but Mautic is a bit more flexible and has fewer significant bugs.

Mailchimp's pricing really doesn't seem all that outrageous to me - considering the time you state it saves

It’s definitely much more manageable for a business, but as one person running a small hobby project, paying hundreds per month is nowhere near worth it. Mailchimp is probably (understandably) focusing on enterprise customers that can pay much more and for whom convenience is a bigger plus.

If your list is large enough to cost hundreds per month, is it still a hobby project? Shouldn't you be able to recoup the hosting cost with all those subscribers?

This is the real issue. If you have 10k email newsletter subscribers and are completely unable to monetize them such that they pay for your email sending, that's the real problem. What's the point of having the list at this point?

Is it really inconceivable that the author has a primary goal other than maximizing revenue?

There's a big range between maximising revenue and 'this service can't afford its administration costs'.

But yes, some people are happy to personally pay the costs for others to enjoy their services.

Fair point, but not everyone is in a for-profit business. Many non-profits have large email lists and little funds.

Non-profit doesn't mean non-revenue.

Any organization with a list that large should also be able to raise enough funds to cover the cost of hosting.

Sure non-profits can probably put hundreds of dollars a month towards mailchimp. But don't you think that some non-profits would rather put that money towards whatever their non-profit is about? I don't get what's so hard to believe about that.

It’s not hard to believe. I’m sure they’d rather put all money paid for services towards whatever their nonprofit is about. What’s your point?

It depends on your ARPU/month. 5,000 customers at $49.99 is 1c/customer/month just for email capability.

There are some industries like Game Development where only one or two emails are sent to the entire distribution list per year.

If you're collecting emails on game announcement at month 0, and releasing at month 24, for an average Indie game at $14.99, deducting 30% platform commission, Mailchimp costs alone represent 2.1% of your net revenue per customer - pretty extreme.

Mailchimp is reasonably priced for small companies or large businesses that can justify the cost as a business expense. But for small, bootstrapped startups or even personal email lists—which seem to be trendy these days—it can be very expensive.

EDIT: Why is this being downvoted?

TinyLetter, perhaps.

That's a Mailchimp product FWIW.

But it has a much higher subscriber limit than Mailchimp's free tier, in exchange for losing many of the power features. (Tinyletter is where I moved my own mailing list to when I hit the ceiling: https://tinyletter.com/gwern )

Hope you don't mind the plug, but my project is roughly "Tinyletter/Mailchimp for small startups/SaaSes" — https://buttondown.email

Can confirm, Buttondown is awesome!

If price is your only concern, you can do mail merges for free and just have your Outlook send the emails for you. But I think you would be underestimating the value of an ESP:

- Establishing trust is everything for email. Even outside of traditional spam filters, you need to stay outside of clutter or bulk filters as well. Mailchimp does such a good job of cleaning their instance of spammers and spam traps that their IP addresses are pretty trusted by inboxes. When you pay for Mailchimp, you are piggy backing on their reputation.

- Compliance is important and getting more important. It's real nice to just have a service like Mailchimp just handle liability and you don't need to worry about lawyers.

- UX. Consumers are familiar with Mailchimp and the subscription settings, etc. Not a big deal, but a nice way to establish trust.

But to more of a point, if you are taking on the responsibility of handling a database of thousands of people, is $30 a month really too expensive a price?

Sendy is just a front end for Amazon SES, another ESP whose reputation is as good as Mailchimp's at 1/100th the cost. We’ve never had a problem with deliverability.

Hi Victor, since you're submitting your work here I thought it'd be OK to give feedback to you here, too. I suggest you label your affiliate links as being affiliate links. If you try to slip that by me, I wonder what else you're trying to hide.

This info is at the bottom of the article:

> Disclaimer: some links in this post are affiliate links. These help support the blog by allowing me to receive a bonus if you sign up.

There are multiple unmarked affiliate links, and you don't know about this until the very end.

I find it at least mildly scummy. A minimum of some form of (affiliate) next to each one would be nice.

The FTC also does not like this behaviour of putting it very far away from all of your aff links and small: https://www.ftc.gov/tips-advice/business-center/guidance/ftc...

I do see this at the bottom of the blog. Possibly was added after your comment.

“Disclaimer: some links in this post are affiliate links. These help support the blog by allowing me to receive a bonus if you sign up.”

Do you also complain if there are ads on a personal blog and it's not disclosed?

Yes, if they were disguised as ordinary links.

Hey, thanks for the feedback. Like some other people mentioned there was always that disclaimer at the bottom - I didn’t know about this FTC stuff some people are citing. Will look into it more and update appropriately!

This is an interesting business model: Sell code for a one time fee. I'm guessing the code has no DRM or phone-home code. I'd be afraid that the code would just be posted somewhere and I'd lose a big chunk of revenue.

Parts of Sendy are obfuscated. Those parts might contain a license checking routine that phones home from time to time.

Back to the future. Selling perpetual license was the old software game - great for bootstrapping

>This is an interesting business model: Sell code for a one time fee.

Wait are you being sarcastic? Isn't this the original business model for all software?

The original business model for software is selling compiled binaries. This is a self-hosted PHP script with no server side functionality. It's similar to a product shipping with it's source code.

It not new. Xenforo, Vbulletin etc. has been doing this for ages.

Making templates that are nice to look at and compatible is hard. Obviously, you can send emails for much less but I have yet to see a WYSIWYG editor as easy to use as Mailchimp's.

Agree. I’m definitely going to miss that the most.

Have you checked out Mailjet[1]?

They created MJML[2], an email templating language that I've used quite a bit. The WYSIWYG editor for Mailjet leverages it, and creates pretty slick looking and compatible email templates if you don't want to use MJML directly.

[1] https://www.mailjet.com/

[2] https://mjml.io/

Mailjet has always been a bad service for us. Price increases, horrible support and constant bugs.

Oh wow! I've used them off and on, and haven't really had many issues. Albeit I haven't run into any issues that warranted contacting support, so I have no idea what that's like.

That said, you don't have to use Mailjet the service to use MJML the templating language. I've also used their WYSIWYG editor quite a few times to edit templates and export the HTML and MJML files for use elsewhere.

If you really wanted to you could set up a free mailchimp account with one recipient, use their editor to create the template and then send it to yourself, and then put that raw template into your other mail sending program.

I understand why authors do it, but to me the credibility of a post like this is severely undermined when I see affiliate referral links.

I don't even need to see affiliate links, most posts like this I assume are incentivized.

Understandable. I tried to make sure everything I mentioned could be easily verified by the reader - you should certainly do your own research and not take anyone’s word for it.

I really like the popup window for a request to subscribe to the newsletter that says "At least this isn't a full screen popup"

MailChimp has gone super crappy. It's not just the pricing (the whole "we charge for subscribers who aren't active" thing is just pure underhand profiteering) but the damn UI is an absolute clusterf* too. Finding anything (forms, automation, even the last email you sent) is just awful. I get the "groovy interface with scribbled cartoon" thing in some instances but dammit sometimes you just need to do stuff easily and quickly rather than being wowed by how clever and groovy a company is being with their design.

IMO they've gone from "do one thing well" to "do all the things badly". I've gone from being lifelong MC advocate to someone who is looking for an alternative for our clients.

Right now I'm liking MailerLite. It's like MailChimp used to be when it was good.

I agree and we are looking to move. USD$200 per month for ~12k subscribers isn't worth it for us, especially with such poor segmenting and automation features. At least they finally got tags, but you can never add someone to the same automation twice even with years between events that's dead for a lot of use cases.

I was really looking forward to trying Drip but they doubled their prices this year so I'm still looking around.

Perhaps it’s gotten better, but sendy for large email lists and campaigns was pretty terrible a few years ago (some lists were > 1m records).

Not to mention the delivery rates of SES and overhead of setting up dedicated IPs (if you have that scale).

Sendy seemed fine for small workloads, so does SES. The software is decent for the price. But anyone serious about delivery or scale should look beyond Sendy or an SES based tool.

There are a whole host of players that are essentially sendy as a SaaS (moonmail, email octopus). From what I’ve heard these are alright. People considering Sendy may wish to look into them

I’ve built and used a few products/tools that are backed by sendgrid, SES, and mailgun. Overall, I am a fan of sendgrid but understand it’s not a direct replacement for tools like sendy.

> Image, attachment and CSV uploads requires the /uploads/ folder's permission to be set to 777.

- https://sendy.co/get-started > Step 2

Is this really true? Why do uploaded files need to be executable?

The folder needs to be +x. So that the user can actually enter it in order to... upload.

Oh yeah. Haha I feel stupid now.

MoonMail[0] is a similar product, but runs via Serverless framework on Lambda and other services on AWS (or their hosted version for $$). I really love the idea but it was a huge hassle to deploy due to many service configurations and different Serverless framework versions. It's also somewhat expensive running single tenant due to the minimum cost of running DynamoDB tables and Kinesis streams (even with near 0 usage).

0: https://moonmail.io/

DynamoDB's free tier is incredibly generous. Did you exceed it?

I had a lot of problems with Moonmail. Took forever to get our domain verified properly even though we were following all of the steps exactly, and in the meantime their support said that the lack of verification wouldn't be a problem, which wasn't true. Then after it was all worked out, something got reset and we had to go through the process again.

We're still on it because we don't use email a lot and it'd be a big time commitment to start over somewhere else, but if I was starting again I'd stay away.

I’ve looked at a lot of these services built atop SES (including Sendy), but the main problem that you still seem to have to spend a lot of time accounting for is not getting marked as spam by clients like Gmail. There are things you can do to help stop this, but it takes a lot of manual intervention.

In any event, it’s good stuff like this exists. I would actually be curious to know how many email newsletter services that aren’t one of the big players (Mail Chimp, Adobe, Constant Contact, etc) aren’t just built on SES.

I send hundreds of thousands of emails using SES per day and haven't had any trouble. As long as you have DKIM set up correctly (and DMARC if you have a dedicated transactional email domain) the deliverability is pretty high. When you reach a large enough daily volume you will need to provision a pair of dedicated IPs but they are only around $25/mo

I've used Sendy for years and haven't found this to be true at all.

Victor-- you've done the community a service by showing us that There Is More Than One Way To Do It, (TIMTOWTDI) even in the 2019 world of web services.

Don't let the nitpickers who don't like affiliate links get you down. Just keep improving your site and providing useful info like this.

I'm glad to know somebody who went to Princeton is created such a useful site. I live in the area. Maybe I'll see you at a Meetup sometime.

Cutting costs by 100x? What's the math on that?

I switched to MailGun, after MailChimp forced everyone using Mandrill to buy MailChimp. Years afterward, I'm still happy.

I've been using sendinblue with good success. This author could have used sendinblue and still been in the free plan if he spaced out his newsletter (max 300 emails a day for free).

Even the first paid plan at $25 lets you send 40k a month, which is pretty solid. And they have transactional as well as mailing list support.

One of the great features of Mailchimp (for us) is the seamless integration with Mandrill. Our Designer creates templates, pushes them to Mandrill and they are immediately used in production when we send out notifications.

I mean you could also just use any one of the other 500 different ESPs that don't require you to host your own mail server and who have different cost structures than MailChimp, but ok!

We moved to Drip from Mailchimp since their policies are not great. Drip is far more functional and less then half the price but not quite this cheap

Rather than supporting a proprietary product like sendy I would recommend using the far superior civicrm which is fully open source, has a great community around it (not used by marketers and seo-spammer-types but rather by ngo's)and is totally extensible to grow with your needs. It is simple to be GDPR compliant by default and is really easy to get started. It obviously has a far bigger scope but you don't need to take advantage of all that additional functionality.

https://civicrm.org/

Is he totally confusing 100x with 10x? This is 10x cheaper, not 100x cheaper.

I had a very interesting setup for my news letter[0]. I let Mailchimp handle subscription and unsubscription and deal with spam subscriber, GDPR and the like etc.

Then I exported the CSV file and have my script[1] generate email and send it using AWS SES. I paid about $1 per month because my server run outside of AWS. If it were on AWS, then this AWS SES fee will be waived.

---

[0] https://betterdev.link [1] https://github.com/yeo/betterdev.link/blob/master/baja/fanou...

MailChimp is shockingly expensive.

question to those who do email marketing. What's the income of this look like? you guys pulling in 6-digits per year? how long does it take to build that kind of following?

yeah, its like buying shitty instant coffee for 25cents - sure you saved 100x and technically did get coffee, but none of it was enjoyable.

p.s. - sendy relies on having a cronjob running, which makes it very ill suited for docker. there are ways to run a sidecontainer in k8s that can do the cron part, but what a giant hassle and waste of running resources. was not a fan.

Reply on news.ycombinator.com