How to Choose an Email Service to Replace Mandrill
When I think about what’s been going on with Mandrill recently, one word comes to mind - “Clusterfuck”. If you’re reading this post you probably already know the story, in which case you can skip down to the alternatives, but in case you’d like recap, here it is:
Mandrill is a transactional email provider from Mailchimp, whom I’ve always liked. At the end of February this year, with very little warning, Mandrill decided to pull the rug out from under many of their users, in a three step attack, that almost seems like a deliberate and well executed attempt to breed contempt within their userbase:
- Eliminate their free plan
- Force existing users on the free plan to start paying (ie: not grandfathering them)
- Force all users to also have a paid Mailchimp account, costing a minimum of $25 / month
The whole thing has been handled very poorly by Mandrill: they only gave two months warning about this change, they didn’t allow commenting on the blog post announcing the change, and worse still, they give an inconsistent message on their own site, as their pricing page STILL (at time of writing) doesn’t mention the need for paid Mailchimp account.
All in all, a really poor show from a company for whom I used to have a lot of respect. In a word, a clusterfuck.
I feel like they could have easily dealt with this so much better. I don’t begrudge Mandrill for eliminating their free plan, the freemium business model has proven too costly for many companies. They should however have grandfathered the existing free plans, maybe not indefinitely, but the two months notice given really isn’t enough. I’d argue 6 months at least, but then for all I know Mandrill could have been haemorrhaging money thorugh the free plan, so a short notice period was necessary.
The part where they really let customers down, the part I really cannot fathom, is the requirement to have a paid Mailchimp account. Requiring all users to have a Mailchimp account would have been fair, even requiring some extra identity validation for users wishing to use Mandrill would have been fine, but requiring Mandrill users to pay an additional $25 / month for a service they don’t want and wont use is what has really left a bad taste in so many people’s mouths.
Unsurprisingly, when faced with a potential price increase of up to $420 / year, many of Mandrill’s customers are predictably looking for an alternative transactional email service.
The Alternatives
There are a few players in the transactional email game, and in this post I’m going to be looking at Mailgun, Postmark, SendGrid, SparkPost, Mailjet, JangoSMTP and Amazon SES. I’ll be evaluating them based on price, deliverability, and ease of migration from Mandrill.
Pricing
Many exiled Mandrill customers were on their free plan, which included 12,000 emails / month. Of the services I’ve looked at, free plans are offered by:
- SparkPost - 100,000 free emails each month
- Mailgun - 10,000 free emails each month
- Mailjet - 6,000 free emails each month
- Amazon SES - 62,000 potentially free emails each month (see below)
- Postmark - Not a free plan, but free for your first 25,000 emails.
One thing I look for in any service is simplicity of pricing, unfortunately both Amazon SES and Jango fall down here:
Amazon SES’s pricing page is one of the most confusing I’ve ever seen. It seems free emails are only free when the SES service is called from an Amazon EC2 or AWS instance, other email sending is charged at $0.10 per 1,000 messages. Data usage is charged separately, at $0.12 / GB of attachment, plus EC2’s data transfer costs for email data itself, including headers, but the first 15 GB is free, plus 1 GB / month. Wait, what?! I have literally no idea how much it would cost to send email through SES.
Jango SMTP’s pricing is a little less confusing, but still isn’t clear, as their pricing of $36 / 18,000 emails is based on an average of 100kb per email. They say their pricing is based on the number of emails you send, and the data usage, which is charged at $10 / GB, but never says what the per message cost is. So once again, I’m not sure what it would actually cost.
The other services have simple pricing based solely on how many emails you send, but the actual prices vary a lot. Excluding the ambiguously priced Amazon SES and Jango, Mailjet is the most expensive, and SparkPost is almost suspiciously cheap, offering up to 100,000 emails / month for free.
Update: I just had an email from someone at Mandrill informing me they have a new pricing structure that will be coming into force with these changes, I’ve updated the table below to reflect this. The funny thing is, the new prices are actually quite a lot more expensive than Mandrill’s old prices (which I’ve left in the table with a strike-through for comparison).
| Service | Emails per month | ||||
|---|---|---|---|---|---|
| 1,000 | 10,000 | 100,000 | 500,000 | 1,000,000 | |
| Mandrill* | |||||
| Mailgun | Free | Free | $45.00 | $245.00 | $421.50 |
| Postmark | $1.50 | $15.00 | $150.00 | $500.00 | $750.00 |
| SendGrid | $9.95 | $9.95 | $19.95 or $79.95** | $399.95 | Contact them |
| SparkPost | Free | Free | Free | $24.99 | $199.99 |
| Mailjet | Free | $7.49 | $74.99 | $399.99 | $799.99 |
| Jango | $2 ish | $20 ish | About $200 | Around $1,000 | $2,000 or so |
| Amazon SES | No idea | Some dollars | More dollars | ??? | Much Gold |
* Mandrill prices include the now required $25 / month Mailchimp paid subscription.
** SendGrid has two different plans for 100,000 emails per month, the higher price includes more features such as dedicated IPs
Deliverability
The website inboxtrail tracks and analyses the delivery rates of many email services, based on what percentage ends up in spam, or simply goes missing entirely. Here’s how the competitors stack up:
| Service | Delivered | Spam | Missing |
|---|---|---|---|
| Mandrill | 96.25% | 0% | 3.75% |
| Mailgun | 61.25% | 28.25% | 10.5% |
| Postmark | 90% | 0% | 10% |
| SendGrid | 96.25% | 0% | 3.75% |
| Jango | 96.25% | 2.5% | 1.25% |
| Amazon SES | 98.75% | 0% | 1.25% |
Inboxtrail doesn't have any data for SparkPost or Mailjet, and I was unable to find any deliverabilty statistics for them.
Most services seem to be on a par here, with the exception of Mailgun, which really fails. This was a real eye opener for me, as I currently use Mailgun for BugMuncher. I’ve been using Mailgun for 4 years, and I’ve never had anyone complain about missing emails, but it’s still made me consider switching service providers.
Inboxtrail allows you to see a breakdown of deliverability by inbox, and it seems Mailgun only has problems with delivering to Hotmail where ~70% ends up in spam, and AOL where nearly 100% of messages end up in spam. Maybe I’ve just been lucky and none of BugMuncher’s customers use Hotmail or AOL. Does anyone use AOL any more?
Ease of Migration
Migrating from a Transactional Email Service Provider should be straight forward, as they don’t store any mailing lists, they just send what you tell them to. In fact, if you’re currently using SMTP to send emails through Mandrill, it should be as simple as changing the SMTP details in your integration.
However, if you’re using Mandrill’s API, and especially if you make use of their templates or other features, things become more complicated, so here I’m going compare the features of each provider to those of Mandrill, and look into any migration guides the other services have.
Features
| Mandrill | Mailgun | Postmark | SendGrid | SparkPost | Mailjet | Jango | Amazon | |
|---|---|---|---|---|---|---|---|---|
| SMTP Support | ||||||||
| Sending API | ||||||||
| Click / Open tracking | Open only | |||||||
| Dedicated IPs | $29.95 / month | $59 / month | $19 / month* | $30 / month** | Free*** | |||
| Templates |
* Only on SendGrid’s pro plans, which come with one dedicated IP.
** $20 per IP per month on SparkPost’s Pro plan or above.
*** Dedicated IPs are only available on Mailjet’s Silver plan and above.
Predictably, some email services have jumped on this opportunity to nab soon to be ex-Mandrill customers, by publishing migration guides.
Mailgun
Mailgun has a blog post aimed at helping Mandrill customers migrate, but it’s pretty weak, and is mostly just a guide to getting started with Mailgun. Eg: it doesn’t address the fact that if you were using Mandrill’s templates, Mailgun doesn’t have that functionality. 5/10
Postmark
Postmark don’t seem to have capitalised on this opportunity at all. 0/10
SendGrid
SendGrid have gone all out, they have a banner at the top of their homepage saying “Coming from Mandrill? We can help.” with a learn more link. The learn more link just takes you to a slightly customised pricing page, which shows SendGrid are currently offering a 10% discount to Mandrill customers. Further down that page is a link to a full migration guide, including a translation table to help Mandrill customers easily understand SendGrids documentation. Like Mailgun, the SendGrid page includes a guide to getting started, but theirs is much better tailored to Mandrill customers, and they even have a Q and A section. 9/10
Jango
Jango, like Postmark, have not created any kind of migration guide. 0/10
Amazon SES
Predictably, Amazon are not about to make anyone’s life easier by publishing a migration guide. But if you can figure out their pricing, I’m sure you can figure out migration. 0/10
SparkPost
SparkPost are going to a lot of effort here, if you search on Google for ‘Mailgun migration’, the first result is an advert for them. Like SendGrid they also have a homepage banner linking to a marketing page, which contains a link to their comprehensive migration guide, which is even better than SendGrid’s, even including guides for migrating templates. 10/10
Mailjet
Mailjet also have an advert that shows when you Google ‘Mailgun migration’, although they haven’t gone as far as adding a homepage banner. Mailjet’s actual migration guide consists of a feature and price comparison page, and their API docs with an extra table to help you convert Mandrill terms into those used in Mailjet’s documentation. Not bad, but quite on the same level as SendGrid and SparkPost. 7/10
The Verdict
Below are my choices for a transactional email service provider, your needs may be different, but hopefully the information above will help you choose the best one for you, particularly if you can decipher Amazon’s pricing.
Winner - SendGrid
Overall, SendGrid looks to be the best alternative to Mandrill. They have a great migration guide, kick ass on deliverablity & features, all while still being one of the best value services.
Runner up - SparkPost
If you’re on a tight budget, have lots of emails to send, and aren’t too worried about the lack of deliverability statistics, then SparkPost would be my recommendation.
I hope this helps you choose your next transactional email service provider, and if you’d like to suggest any other services be sure to leave a comment below.
Matt