Hacker News new | past | comments | ask | show | jobs | submit login
Twilio to Acquire Sendgrid (twilio.com)
633 points by coloneltcb on Oct 15, 2018 | hide | past | favorite | 184 comments

Ah, YC's one that got away - https://www.quora.com/Why-was-SendGrid-rejected-from-YCombin...

>Apparently, only one partner, Robert Morris, reviewed the SendGrid application, and he gave it a highly negative score, calling it a "spam company."

Robert Morris[1] is VERY (!VERY!) technical and he commented on my YC application many many many years ago somewhat along the same lines. I am not trying to be negative, but sometimes VERY technical people "poopoo" ultimately successful businesses because they seem trivial or technically not interesting to the engineer inside them (see "Show HN: Dropbox" comments).

In contrasts perhaps they fall in love with deeply technical ideas that pleases the engineer inside them, but ultimately will fail to become a successful business.

[1] https://en.wikipedia.org/wiki/Robert_Tappan_Morris

Slightly off topic but PG was funny talking about him in the recent video https://www.youtube.com/watch?v=4WO5kJChg3w&feature=youtu.be...

Was he wrong? I used SendGrid for a while, and ultimately stopped when I found that I had better deliverability sending email directly from an EC2 instance. And I certainly receive plenty of spam via SendGrid.

I used to receive a lot of spam from them, but recently (in the past year) they seem to have cleaned up their act. My last spam report to them was in July.

Most of the spam that gets through my spam filter now comes from gmail (which is impossible to do anything about, as google doesn't give a fuck about abuse reports).

Yeah, gmail is consistently my biggest source of spam. Unfortunately they're also a very large source of ham...

Just to be clear this is spam from gmail to a non-gmail account? What kind of numbers are we talking? A few a day, tens, hundreds?

I receive about 1-3 spam at most in my old, well published, gmail account. And it's never in any of the sorted columns besides maybe "Promotion", which I rarely check.

I suspect the parent poster is talking about spam sent from gmail to non-gmail accounts - such as my own self-hosted mail-server.

I receive 5-20 spam messages a day from @gmail addresses. Not a huge number, but some days it jumps to 50-100. (These messages really did get routed through google, rather than being faked.)

Yes, email sent from gmail to non-gmail. I typically get ~25 spams/day originating from google (admittedly not all to the same destination address), but from time to time it jumps substantially.

Spam from gmail's smtp servers. There are a number of spam services oriented towards small business. Their users are required to use their own gmail credentials. MailShake and FunnelBake are two services from my inbox this week, though they're deliberately hard to identify - no note in the headers, no footer on the email, customers encouraged to set up domains aliasing "unsubscribe" links, etc.

This is a growing source of spam for me personally and services I've worked on that have public email addresses (eg. support@), especially if near a mailing address. (Other services scrap lists of "local" businesses and sell them to the end users of the spam services.) Blocking gmail smtp is a non-starter due to popularity, graylisting doesn't help, and I haven't been able to find anywhere they take spam reports; search results are overwhelmed by discussion of reporting spam as a user of GMail.

My guess is that if we keep reporting SPAM, Google will eventually have to crack down on this.

A spam company. Not a terrible analysis. But perhaps what he missed is that the company that figures out how to send email well despite having horrible spammers as customers will make millions.

I think what we missed is that there would be two tiers of email senders, differentiated by price. There would be $100/MM email senders that would appeal to spammers, and $500/MM senders that would appeal to companies that manage their communications well. They’d have different delivery rates through spam filters. The bottom tier is bad for the world and a low-margin business. The top tier is OK for the world and a profitable business. SendGrid seems to have found the sweet spot.

The purpose of y-combinator is not to make tons of money, neither should that be the default purpose of a company. "Business ethics" is a thing.

Y Combinator does not strike me as particularly concerned with ethics.

Hmm, really?

In a capitalist society (such as we live in, like it or not), we express value in money. IF your company isn't making money, is it adding value? If people aren't willing to pay for it, is it actually any use to them?

The notable example for this is Facebook. We don't pay for it. Advertisers do. That shows where the value is...

Phillip Morris makes money, where’s its value add?

I smoked for 20 years. I loved driving with a Marlboro Light hanging out of the window. I chose to do that to my body from my own free will, and willingly paid to do it. I'm glad I stopped, but I enjoyed it while I did it.

Things that are bad for you can still be valuable. I still willingly buy beer, even though I know it's bad for me.

Value isn't what you think it is, if you think that smoking provides no value.

Saying that value is expressed in dollars is not the same as saying that dollars are the only thing a company should ever seek. That's not how it works.

If every company sought only money for money's sake and did nothing else, then they would all act like Oracle.

> In a capitalist society (such as we live in, like it or not), we express value in money. IF your company isn't making money, is it adding value?

Yes, but it's a very imperfect way to capture value.

Without legal support, it doesn't account for negative externalities - eg, if your factory saves money by dumping waste into a stream, making people sick (but you don't pay for that). Or your site gets visitors by sending spam, wasting the time of the recipients and everyone who works trying to filter spam (but you don't pay for that).

It doesn't account for things consumers don't have the foresight to care about, or don't have options to choose between. Eg companies that manufacture a fridge that will last 10 years and be unrepairable may make more profit than those who would make one that lasts 50 years and be repairable, because they'll sell more product. Planned obsolescence wastes resources, but is profitable.

It also doesn't account for the value of things like raising children. If I play for my own children for 2 hours, that's not part of GDP, but if I pay someone else to, it is.

And it doesn't account for the fact that people's desires and priorities are warped. Child pimps generate revenue, whereas the social workers rescuing kids from them do not.

obviously you don't like living in a capitalist society, then.

Fair call, good stuff. I hope you manage to change the world to your liking.

Meanwhile, in a capitalist society (such as we live in, which you clearly don't like), we express value in money.

I never said that. It's not perfect, but I far prefer living in a capitalist society to a centrally-planned economy, which I think can never work.

The game of capitalism requires referees. Otherwise, robbery is an excellent business model.

That, I completely agree with. Unfortunately we have a habit of closing the stable door after the horse has bolted... people are much quicker at inventing ways of abusing the money system than the regulators are at stopping them.

I'd like to demonstrate my point of view about money=value with a story, if you don't mind reading a bit:

When I was in my late 20's I was earning a lot of money working on a software project that was doomed to failure. Everyone on the project knew it was doomed, but it was politically infeasible for the current management to pull it or deal with the fact that it was going to fail. So every day, our little team of very expensive contractors wrote a bunch of code that we knew was never going to be tested, let alone run on real server. We all collected our pay and waited for the project to be canned.

During this time, I met a nurse randomly one night. We ended up walking around town talking. She had just got off shift at a local hospital, saving people's lives. She got paid about 10% of what I did. The ridiculousness of me getting paid 10x more than her, when I was producing nothing of any value, while she was saving lives, was stark, and it has always stuck with me. It seems like such an obvious mis-allocation of funds, a failure of the economy.

But in a larger sense, building the Internet (by which I mean our global networked computer-based infrastructure) is the thing that Humanity is engaged in at the moment. When they write the history of our time, it will be all about the Rise of the Internet. Creation of public health services, on the other hand, was the thing we did last century. So in that sense, it's easy to see why our profession is highly paid, even if the small-scale project we're working on right now is doomed. We are doing the most important thing on the planet right now, and everything else is merely background noise to that effort (at the moment, this will change).

All this is a long-winded way of saying that seeing emergent behaviour from inside the thing is much harder than seeing it from the outside. To us, this all looks like random bullshit and injustice. But zoomed-out, from a historical perspective, it makes more sense. Still not perfect, but at least coherent. The individual injustices and weird nonsense get smoothed out, and it looks like everyone had a plan all along.





Are just some of the social sites whose founders were all spammers.

There are more but i forget some.

From one spammer to another then.

What makes you assume YC has no ethical basis for investment?

Their expansion to China.

Are you using Slack, Uber, Docker etc.? Because if yes, then you're not the most ethical either, as these have investors from Saudi Arabia to the CIA. China is easy to pick on, (and am not saying it shouldn't be picked on), because it is a competitor to the U.S, however if it is such an ethical concern for you, there are earlier partnerships you should be concerned about.

China is easy to pick on because it's the greatest evil the world is facing today. There are plenty of other shitholes around the world, but only China is capable of forcing almost every country and company into their submission.

Also, there's a big difference between using a free product and working for a company that supports China/Saudi Arabia/etc. Or in YC's case, bringing billions of dollars worth of investment into a country that sends Muslims to concentration camps, and helping them build a dystopian surveillance state that will force everyone to spread their propaganda and fall in line; while being completely helpless as hundreds of millions of lives are destroyed..

That being said, I do try to avoid using products of unethical companies, and I would resign without a second's hesitation if I found out my company was involved with these evil governments.

> China is easy to pick on because it's the greatest evil the world is facing today.

Not really a big fan of China, but lets be a bit more consistent here.

China is in the middle of an empire-building enterprise and like every such enterprise, (UK, Russia, United States), does many awful things. Yet compared to some places in the Middle East, I don't think it's nearly as evil. Certainly not "the greatest evil the world is facing today", (the only way I can see that is if you think U.S. global supremacy is great).

Personally, I'd argue the greatest evil the world is facing today is people in power ignoring climate change.

> only China is capable of forcing almost every country and company into their submission

You must have missed the war in Yemen, as an example.

> there's a big difference between using a free product and working for a company that supports China/Saudi Arabia/etc.

How so? The products I mentioned are not free software for the most part, by using the product, (even their "free" plan), you're supporting the company China/Saudi Arabia invests in and thus supporting these governments economically.

> bringing billions of dollars worth of investment into a country that sends Muslims to concentration camps

As I said, that's really a case against empire building and a case I agree with, however it's been done by every empire. Similar things have happened at black sites in Iraq etc. so I think the consistent line to take is to be against investing in any empire in that case, which would include YC's home turf.

> I would resign without a second's hesitation if I found out my company was involved with these evil governments

That is a great stance, however the cynic in me thinks they probably already are in some way, even if you don't know it.

I recall Nike was using it.

The term is "mainsleeze spam." Just because their customers aren't pushing v!agr4 and their lists are smaller doesn't mean their emails are any less unsolicited. If you don't take technical measures to require your customers to affirmatively ask for permission before putting someone on their lists, you're a spammer.

Well Twilio is "phone spam" when abused so I guess its a good match.

I believe there's at least one company with a larger private valuation that got rejected from YC (haven't seen a primary source confirmation of their rejection though).

ironically, first time I have ever heard of sendgrid was a few minutes eaerlier, when I saw that an obviously fake email from verizon was actually from spammy@sendgrid.com

downvote a factual annecdote all you want. still a fact :)

Smart move. I used SendGrid and Twilio years ago, they basically solved the email and SMS dev-ops problem for more product focused engineers like myself. Two companies with a solid product consolidating, hopefully that will lead to better integration and more good news for devs moving forward. Congrats to SendGrid team!

I’ve used sendgrid for a while but had issues with deliverability. Then I moved to postmarkapp [1] and have not had any issues so far.

1 - https://postmarkapp.com

Same exact boat, circa mid-2017. Was seeing lots of email go to spam on SendGrid, asked their support about it, and they suggested I purchase a monthly static IP. I moved to Postmark to save money and their deliverability ended up being WAY higher, due to the transaction-email-only policy.

But postmark is only transactional. Not mass mailing.

That’s the point, it helps keep the reputation of their servers high and ensures fast delivery of transactional email.

That might be why their deliverability is better :-)

+1 for postmark. Had deliverability issues with mailgun which were resolved once switching.

is deliverability purely based on the email server, or is it a combination between your domain/etc? I always wondered if it was YMMV for everyone saying different things about the different EMS out there (SendGrid, Mailgun, etc)

deliverability is actually incredibly complicated. Every mailbox provider (gmail, yahoo, hotmail/office365, comcast, etc... being the main ones) has their own algorithm for determining what is or isn't spam and includes things from user interaction to IP sending reputation to domain reputation to various types of authentication techniques (dmarc, dkim, spf, etc...)

I've had some negative experiences working with SendGrid a few years ago, but honestly I think they've improved a great deal in the last couple of years. Some stuff still could use some improvements (API, UI) but it felt a lot more stable and reliable last time I used it.

Still, I miss Mandrill the most.

Mandrill was amazing. Then they jacked up the price and tossed it into MailChimp.

I kinda love that they jacked the price up so the delivery is great. The only thing that i hate is (i use the SMTP endpoint) is that the BCC does not work.

BCC is probably disabled to discourage you from using mandrill for mass/marketing emails.

no, it kinda works, but not like a regular email. like when i send some email with to: contact@abc.com with bcc: we@def.com the receiver of the bcc sees the full email header and sees who the original receiver is. but with mandrill you simply get two emails, one for the original receiver and one for bcc, both send with to fields, the bcc recipient does not see who the original recipient is

They put me on a spam IP 2 months ago & i wasn't able to deliver to outlook, account got disabled later that day for 'abuse'. Same for Mailgun without the account being disabled, although they quickly switched me to a different IP and it was fixed. It has given me anxiety about deliverability, even with domain and everything configured correctly you could just be put on a bad IP, especially for low-priced options.

I thought Mandrill used a range of sending IPs so this wouldn't be an issue? It's seem very lame if they just shard accounts across a IPs.

I'm not too knowledgeable on how these deals work, but figured someone on HN would know:

A quick Google search shows that Twilio's market cap is currently $7.4 billion. Does this $2 billion "all-stock" transaction mean that they are giving away over a quarter of the company to pay for this acquisition? Or how else should I read this?

I haven't read the details but not entirely. They will likely dilute their own outstanding shares by adding the required shares needed to acquire the company at that value. This will lower the total ownership as a percentage that each share represents but now every share owns more stuff.

So yes they are creating new shares making every other share worth less but it isn't a direct transfer of existing shares. You'll notice companies get board approval for setting aside a number of theoretical shares they could create if they wanted to for things like this or secondary offerings, etc.

Most likely this will be done by issuing new equity. The press release states that they will exchange Twilio Class A common stock per share of SendGrid common stock. They defined the exchange ratio (that's why the price is ~$2B and not a specific number) and will issue as many shares as they need to based on that ratio:

> 0.485 shares of Twilio Class A common stock per share of SendGrid common stock

Technically, issuing new shares dilutes every existing shareholders' ownership. But, if the market deems this to be an intelligent combination that was valued correctly, the market cap of the combined company should be ~$9.4B or (ideally) more. So the goal is for existing shareholders to hold, at worst, the same amount of value in dollar terms as before or, at best, hold more value as a result of the acquisition.

SendGrid is up 17% while Twilio is down 3%.

The other commenters gave good answers, but to simplify (in case it helps), the math of what they're saying about issuing new stock, changes the arithmetic from 2 / 7.4 = 27%, to 2 / (7.4 + 2) = 21%.

They will be creating stock to fund the purchase, so in essence, yes. The number of shares outstanding will rise by $2B (price) / $76 (Twilio price as of transaction agreement) = ~26M new shares of Twilio to Sendgrid owners.

They are essentially merging

Approx $26 million of net profit fetched $2 billion in acquisition? A somewhere between 65 and 75x multiple?

That doesn't seem right. Are my numbers wrong?

It seems like they paid paid 13x projected 2018 revenue [0] for the company. Sendgrid's products are a solid complement to their existing product lines and could probably be cross sold to many of Twilio's existing customers. Makes sense to me

[0]: https://sendgrid.com/news/sendgrid-announces-second-quarter-...

SaaS companies trade on revenue or free cash flow multiples for the most part:

https://medium.com/@alexfclayton/how-much-is-your-saas-compa... https://www.bvp.com/strategy/cloud-computing/index

Yeah, btw I think I cited quarterly. Still I think the multiple is too high. I understand 15x for a private company and 10x for a public. But in an inflated market, higher than that? If I had internet other than my phone I would look it up.

So $100M Profit per year? $2B is only 20PE, sounds fair to me, in todays stock market where many are trading at 30 - 40+.

Enterprise value is not the same as synergy value

For those that use email for marketing. Is there anyone cheaper/better than Elasticemail?

1m emails/mo:

elasticemail.com - $90

sparkpost.com - $474

mailgun.com - $515

sendgrid.com - $534.95

smtp.com - $900

Or is there anyone between elasticemail / mailgun worth considering?


* Please don't suggest Amazon SES.

Considering that those prices are for 1 million emails a month, I'd say price is so similar between all of them that you should just pick whichever one gives you and your customer the best experience. The hours saved for your employees will easily save more than they cost if you pick the one that saves them the most time.

Exactly, if you're at the point where you're sending 1 million emails a month you're either spamming people or you don't care about the difference between $100 and $900.

I have a counterexample for you. Start an Internet forum as a side project that lets users subscribe to threads for email notifications. Grow it to xx,xxx users making x,xxx posts per day.

Before you know it, you're sending 20k-50k notification emails every day, and over a million emails a month, without making any money or while barely staying afloat with $100-400/month in ad and donations revenue.

Almost none of the SaaS or IaaS solutions HN loves to praise are within this use case's price range but scrappy not-for-profit communities like this deserve to exist, too. You're basically limited to leasing a bare-metal server or two and making it do almost everything you need.

The difference between $100/mo and $900/mo is the difference between getting to stay online or shutting the forum down. Even the difference between $10/month and $100/month might be enough to close shop over.

Personally, in this scenario, I've found setting up my own mail server with a tool like Postal or Cuttlefish to be worth every minute and delivery rates to even be better (due to getting to manage my own IP address's reputation vs sharing a mail-sending IP with anyone else).

SendGrid (and I assume the other similar services) charges a premium to pay for the work that your emails don't get blacklisted anywhere enroute -- that's worth 1/20th of a penny to most business emails (since it increases open rates), but it may not be worth it for every use case.

Isn't it the case these days that if you aren't sending spam and have a stable sending IP that the system will eventually learn that you are not a spammer and your emails will get delivered?

It can take months for an IP to earn a reputation as a non-spammer, but once it's earned I find it hard to believe that Sendgrid offers any additional ability to get emails delivered outside of the natural earning of reputation.

It doesn't even take months. Reputable hosting companies give you a "clean" IP that isn't on any blacklists; its reputation is yours to make or break from that point onward. Send clean mail from the get-go and the IP's reputation goes from "neutral" to "trusted" very quickly, in a week or two tops.

You end up on blacklists by sending out spam. Don't spam - meaning, don't send unsolicited emails. That's it. That's all. Done. There's no magical secret sauce to this like SES, SendGrid, Sparkpost, and friends want you to believe.

There are two things you have to do to not spam:

- Don't be the one sending spam. Don't send your users any email they don't ask for short of extenuating circumstances that truly or legally warrant it, like a security breach. Just don't. This also includes making it brain-dead simple to unsubscribe from anything they once signed up for.

- Don't let anyone else send spam through your server. Keep it locked down. Lots of criminals are constantly prowling the IPv4 address space for poorly secured mail servers to fire out a million crappy scam mails from before a blacklist notices.

The amount of work that blacklists create is grossly overblown. If you're doing email right, the only time you'll ever interact with one is when your mail server gets owned - in which case, you send the blacklists a quick note to explain what happened, that you fixed it, and they take you off of it within a day.

Sign up for mxtoolbox.com, secure your server, respect your users' inboxes, and you only have the box's uptime to worry about.

That's simpty not true. We never ever do send spam, and i mean not even newsletters for our customers. And i still had problems when delivering invoices (PDF attachments), And thats with paying mxtoolbox 25$ to monitor blacklists. The biggest problem was microsoft365 mail service. We then switched to external local mail provider (europe) that had better delivery but even then some of the invoices would simply get stuck in microsoft365 and i would get no reponse. We are now using mandrill, and it works ok. I did not want to host it outside of europe but i did not have any other option.

One anecdote against another. My suspicion is that you were in a "bad neighbourhood" - ie. using a hosting or email service that didn't keep their IP's squeaky clean. AWS/EC2 is a common host among the HN crowd but is one of the worse places to host a own mail server at for this reason - they'd also rather upsell you SES than fix the EC2 IP ranges' reputation.

Microsoft is the pickiest email provider to get good delivery rates to - they require a gradual volume ramp-up over 1-4 weeks on a fresh IP to see it as good, in my experience, and SNDS is something to keep a close eye on during that time - it clearly indicates your IP's reputation with them and how much mail they see from it. That number being higher than expected plus the sample emails they provide may indicate something going wrong - through this, I once discovered an obscure feature in a software package I used that allowed visitors to send emails to arbitrary addresses.

Mandrill has better delivery than the other email services I've tried but is cripplingly expensive for a hobby project that happens to have a nontrivial userbase.

> My suspicion is that you were in a "bad neighbourhood" - ie. using a hosting or email service that didn't keep their IP's squeaky clean.

No, not really, we are using one the most expensive offerings in Switzerland. Renting dedi servers for spamming would be not only extremely expensive, they would kick you right away if they noticed. As i said we did nothing wrong, used monitoring and our invoice emails would sometimes end in the spamlist for microsoft, after trying to get it right i gave up and hooked mandrill.

if you are runnig some random internet forum, some part of your notifications ending in spam is not a big deal, but for us, invoices ending in spam box is not something you can ignore.

I remember I had to go through a couple of hoops with outlook.com and gmail(?) using some un-/under-documented web sites to get "approved" (sending a mail, getting a reply, typing reply in web form).

And in my experience the major services have unpredictable spam filter behavior.

I even saw Google being cute and assuming mail in Japanese was spam - because my profile is western.

So yeah. If your goal is 100% (not high 90s) deliverability - including the odd big attachement or foreign encoding mail - it's not trivial.

Thankfully I don't care that much if a recepient at Gmail or outlook don't get my mail;it's their fault. If it's a problem the recipient should move to a more reliable mail provider.

It's obviously not as easy for commercial/work mail...

Reputable only matters when small. When large like aws you could be getting anyone's ip.

AWS has a vested interest in upselling you SES for email-sending needs. The quality of their EC2 IP ranges has little to do with their size; running a tighter ship with those IP's is merely bad for their business model.

Don’t respond to spam feedback loops from major mail providers and you’ll quickly get blacklisted no matter how much reputation you built up. This could happen at any point and doesn’t require that many people to mark as spam.

It’s worth paying someone if deliverability is important.

Deal with the spam feedback loops - they exist for a reason. You have to treat an FBL email as someone hitting the "unsubscribe" link (though it might also be an indication your unsubscribe link has to be more visible or your unsubscribe process less obtuse). Some off-the-shelf software can monitor your bounce and FBL messages and automatically process them as such.

Getting blacklisted is a temporary state and I maintain it's easy to recover from if your IP's and domains had a good history. I've experienced the IP's that email hosting services send my mail through being blacklisted and then waiting days for them to deal with it versus being able to contact blacklists on my own behalf to resolve things same-day.

I used to pay one of these services $xxx/month and experienced delivery rates around 60-70% that they couldn't do anything about other than offering me a dedicated IP for a lot more money. Moved to my own mail server with an OSS transactional email system and delivery rates shot up to 99%-ish while eliminating the email bill.

What OSS projects did you use for this?

Postal is my favorite: https://postal.atech.media/

I'm not sure I see how this is a problem (or viable market). No income, volunteer resources - your best bet would be run it yourself (affordable dedicated servers, lots of free volunteer hours) and/or do it for/with others (like neocities.org is doing for web hosting).

Seems to be fewer such services now than a decade ago, but maybe something like: https://www.freelists.org/

The idea behind SaaS is that it's wildly more expensive (higher margins) because it frees up all those hours to making profit. So make a hundred spend fifty, versus make twenty spend ten on hosting. Or something along those lines.

Or you run a popular free service or service that makes so little money this is important. Consider F5bot. It wouldn't take too many subscribers getting several emails per day to hit 1000000 per month.

That's why you don't run a free service with actual value that people would pay for. F5Bot is just being taken advantage of in my opinion. They recently got front page of HN again, and their patreon page went from $41 a month to a whopping $50.

I can come up with examples all day of things people could create where they care about $900 leaving their pocket each month. Dismissing all these projects wholesale (as the parent comment did) is not helpful.

If you are on shoe string , sending your own email is a better solution. It will take few weeks to establish your rep, as along you process the abuse reports , it is not a problem for transactional email.

The original question was for marketing email, if you are sending a million marketing emails a month and not making enough money then the conclusion that you are sending spam is not unreasonable

> If you are on shoe string , sending your own email is a better solution.

Or just use something like Elastic Email or SES.

> The original question was for marketing email, if you are sending a million marketing emails a month and not making enough money then the conclusion that you are sending spam is not unreasonable

That's fair enough.

What's wrong with SES? (Not a rhetorical question. I've played around with it and it seems ok, so looking to hear about downsides.)

Many of their IPs have been blacklisted by spammers.

When I last tried to use them for marketing purposes with paying clients, it was really hit and miss.

See below. They are old links, but serves my point.

[0]: https://forums.aws.amazon.com/thread.jspa?threadID=233001

[1]: https://forums.aws.amazon.com/thread.jspa?messageID=538207

[2]: https://forums.aws.amazon.com/thread.jspa?threadID=220517

Also if your sending that amount of emails, you want your own IP. Maybe even two IPs, One for transactional and one for marketing.

SES allows you to purchase dedicated IPs

Postmark? Been using them for a while for inbound/outbound and they are great. For that volume, give them a call.

Yep I have used Postmark for years and I'd recommend them over anyone else. Fantastic service that does exactly what I need and no farting about with tacked-on features.

What about setting up your own SMTP server? I’m genuinely curious.

Just configure SPF and DKIM, and send email basically for free, no? Is there some back magic that these services perform to avoid getting blacklisted?

What’s the benefit of using one of these services besides a nicer API and a dashboard?

At a previous company I worked at we used to manage our own email infrastructure before finally switching over to a dedicated service. There are a few problems if you are sending over bulk email that is customized per user when running your own SMTP service.

1. IP address reputation - Keeping your IP addresses reputable is not a simple task. It requires balancing your emails for popular destination domains (gmail.com, aol.com, yahoo.com, etc.) across multiple external IP address. It requires you to deal with many different conflict resolution departments, who don't care about email, when a dispute comes up. It's practically a requirement to use a service like ReturnPath to maintain your reputation.

2. Throttling - When doing it yourself you need to throttle yourself. This is problematic on "big" days, especially when your marketing department wants to send many millions of emails for a big product push, promotion, or on days like black friday/cyber monday.

3. Hiring - A lot of people think sending email is easy. When you get up to the multiple million per day mark things start to fall apart. Do you have someone(s) on staff who really know sendmail/postfix/qmail inside and out?

4. Monitoring - sendmail/postfix/qmail are often times hard to monitor. You have to put together all of your stats. You have to put together all of your alerts. If you aren't really experienced with bulk email, you won't know what to look for and that can impact your reputation. Also consider your logging infrastructure. sendmail/postfix/qmail are noisy.

5. Cost - All of the points above play into the cost aspect of it. Is it cheaper to run it yourself, pay for all of the services and salaries, etc. Or is it actually cheaper to just use sendgrid/mailgun/etc. IP address reputation services are not cheap. Infrastructure cost is also something to consider. AWS IPs all have pretty terrible reputations so running this in AWS (and maybe other cloud providers) is a non-starter since no one will accept your email.

If you've got the expertise and you are sending a massive amount of emails then it might be worth it to run your own infrastructure, but at the end of the day, a single developer consuming an API is often easier and less problematic.

The other important one, at least in the US, is unsubscribe/CAN-SPAM compliance [0]. You've got 10 days to comply or you'll be at risk of up to a $41k fine per email!

[0] https://www.ftc.gov/tips-advice/business-center/guidance/can...

My favorite features:

1. SAAS web app to manage the service with logs, and configuration as well as multi-user management with various roles. 2. HTTP Rest-like api for sending mail from your app 3. API keys for authentication, managed through web app 4. They do monitor black-lists and removals from blacklists. They also manage the warm-up time for your IP to make sure it is ready to send the volume of email you are sending, and can queue emails to avoid sending too many at once to the same destination mail service. 5. Spam control, in case you use the key in an app that turns out to have a relay vulnerability. 6. Return-path management to catch blocks and smtp errors.

I've managed my own SMTP services with PostFix and SendMail, and I will never go back to doing that myself again. It would take a lot more than 1M emails/month to make that worthwhile.

The API alone is pretty amazing. SendGrid allows you to send up to 1000 customized non-campaign (transactional) emails in a single call. This is huge if you send a ton of email. Plus the deliverablity stats for transaction emails is great, too.

Sending a high volume of mail through a server with a new IP will raise red flags.

If cost is a concern it does pay dividends in the long run though. Biggest hassle is the initial setup.

It warrants physical infrastructure, which for many companies this is a non starter. Cloud IP addresses tend to have pretty terrible reputations with email clients.

Generally if you run your own server at any reasonable volume, you will be blocked by anti-spam algorithms and once your domain makes a blacklist, it's very difficult to get it reinstated.

You can't know if your IP is already tainted until you realize your mails are landing in people's spam box.

I use one of the service even for my private email server (which is essentially free for the volume being sent), so I don't have to have stupid conversation like check for spam box whenever I send emails.

Having emails land in spam box pretty much makes your email non existent with the volume of spams being sent today. What is the reason not to use one of the services either privately or professionally?

> Is there some back magic that these services perform to avoid getting blacklisted?

Yes. No matter what your project is, unless your secret sauce has to do with email delivery or your business model requires email delivery to be free, do not attempt to roll your own email delivery service. No one has ever looked back and wished they had.

What is the missing sauce? My understanding is that once your sending IP has some history and has developed a reputation as not sending spam, there isn't anything offered by Sendgrid that improves upon that.

If you set up SPF and Domainkeys/DKIM that is an added vote of confidence but Sendgrid doesn't even require it, so I think most of the value of Sendgrid is simply offering a stable IP address and a convenient API. Am I wrong?

Read here: https://news.ycombinator.com/item?id=18225068

That poster gave a very detailed breakdown, but point #1 is enough to make anyone who’s done this before cry in a corner from the abuse they endured.

If you go this route, I would recommend a separate domain for any emails that are are bulk in nature.

Keep the one off transactional ones on a different domain. Then, if your bulk emails cause a false positive, you haven't burned essential things like receipts, password resets, shipping notices, etc.

No personal experience with them, but socketlabs.com is pretty close to mailgun for price.

At a previous job (company sent ~2m emails/day), I was good friends with the engineer in charge of our mail infrastructure and he was really impressed by their platform, particularly the APIs around getting non-deliverability notification.

If this is marketing mail receivers want to receive and anyone in the company has experience with setting up email systems: Use some fresh IPs and run your own.

As long as it isn't spam and bounces are handled appropriately deliverability shouldn't be a problem.

You can't likely know which IP is fresh until you test it with a certain volume and check for deliverability.

The costs for delivery service aren't really high, not sure what the point of sending it all by yourself.

You should at least use a long living domain to increase deliverability.

> anyone in the company has experience with setting up email systems

I could very easily see someone spending $500/mo of their time, or a couple of days at most, fiddling with the mail system every month or dealing with crises.

I've been looking for a low-cost provider so I signed up for Elasticemail. Their new account confirmation email went to Gmail's spam folder. Not a good sign :(

I guess I'll have to keep looking.

I use Pepipost but not at that scale. They quote up to $245 a month for 1 million emails.

Didn't know about these guys. Thanks for suggesting them.

they have a logo which looks very similar to mailchimp..

Which has the best API?

Wow, this is a big deal! Great move by Twilio, bringing together the premier SaaS SMS provider and the premier SaaS transactional email provider under one roof.

It's only too bad that SendGrid has struggled to get its marketing email solution off the ground in a meaningful way. If anyone was going to eat MailChimp's lunch, SendGrid would be my choice as top contender. And yet despite SendGrid releasing many iterations of their marketing email solution over the past 5 years they've never seemed to get a lot of success outside of their bread-and-butter transactional email vertical. Their marketing solution brings in less than 20% of total revenue last I heard.

Because the marketing platform is terrible. Frequent need for a full page reload to fix a caching issue. Test emails don't support email template tags without a default value. Searching for an email address is terrible slow, the upload is glitchy. Email activity is also slow and limited, sometimes I don't even know if I hit enter because I have to wait so long with no UI indicators. The pricing is also horrible with no way to purge bounce or unsubscribed emails, so they just collect and drive up your "Contact Storage" fees. The recent Sender Authentication change created unreadable sub-domains until they started to offer an advanced option again.

Yes, I emailed support about a few issues but received the blanket "What browser are you using?" when I provided full details. So I built a marketing system on top of their API and fixed those issues for our users.

What's next snail mail, LOB?

Next is Dialpad.com.

I don't know if MailChimp is on its way down or what. I recently reset my phone before realizing that 2FA would be affected. Now, I was encouraged to enable 2FA, but I can't get back into my account and MailChimp isn't getting back to me about what I can do if I don't have any backup codes. I know it's a tough situation, security-wise, but I can send my passport if they need it!

From my experience, mailchimp's transactional email story is a bit of a convoluted mess with mandrill.

I remember MailChimp/Mandrill removed their free tier, making me move to Sendgrid.

Only because of that I hope they go under sooner than later.

Ouch, that forced transition was certainly developer-hostile, but I don't want to see Mailchimp go bankrupt just because they pulled the plug on me (and switching to Sendgrid was basically an improvement for me, anyhow).

The developer-evangelist approach from both companies has been top-notch: Supporting a lot of smaller people/events/organizations.

Kind of shocked that SendGrid didn't IPO but still this is a great victory for both teams!

SendGrid IS a public company! "SEND" is their ticker symbol...

SendGrid did IPO, its traded under SEND on the NASDAQ


They ipo'd last year

What are the benefits of using SendGrid over AWS SES? I take it the overall API ease-of-use / UX is a real factor.

I ask, as I believe that SendGrid is now built on top of AWS: https://sendgrid.com/blog/sendgrid-joins-amazon-web-services...

There are people who can confirm or deny this better than I can.

lets say an email bounces so the address gets blacklisted, but its a legit email address from a real customer. customer emails in and asks why they arent getting their emails, and customer support should have the tools to handle.

sendgrid: search email address, find the bounces, click the button that says allow and user gets emails.

ses: set up an sqs queue. Pipe all bounces into a db. manage said db. Make a service with ui to search for emails. include a ui to unblock address.

That, and ses attachment size limit is a joke.

UX is huge. Sendgrid is absurdly easy to use, and at least 3 years ago, SES was an impenetrable mass to me. I got my company up and running on Sendgrid in like 20 minutes.

Maybe it's easier nowadays, but at least a few years ago it was a night and day difference.

Using it today, still complete garbage UX wise.

My least favorite aspect is that the API always claims a success even if that address is in the global blocklist, the only way to learn about bounce backs is to subscribe via SNS.

Also the UI is effectively useless.

> SES was an impenetrable mass to me

AWS UX in general is kind of a mess. Vast in its services, no doubt, but just messy.

I have an anecdote, email me if you care! Your profile doesn't have one.

It's interesting SendGrid is worth so much, while I believe they were one of the first players in the outbound space, competitors such as Mailgun, Amazon SES, and Postmark have very competitive products at lower prices. Seems like a heavily commoditized business. I'm guessing SendGrid just has a lot of "enterprise" clients and contracts that rake in revenue.

Last I checked Postmark is actually a fair bit more expensive, and SES is cheap and very, very, very barebones. After Mandrill exited the space in a fireball of burning good will, the remaining players in what I'd call the "full service but not eye-wateringly expensive" space are Sendgrid, Mailgun, and maybe Sparkpost, and there's not a lot of distance between them in terms of price and features.

One thing Sendgrid is good at it letting people easily send emails (ie, receipts, confirmation emails, etc.) on behalf of their clients, while still passing SPF/DKIM checks, which Mailgun and Sparkpost make a bit harder. If you're working on a SaaS product that needs to send emails on behalf of your clients, Sendgrid is an obvious choice.

Mandrill (Mailchimp) changed their terms of use which resulted in an exodus of high-volume customers who refused to pay Mailchimp's exorbitant bulk email rates. Sendgrid was johnny-on-the-spot helping people migrate.

Sendgrid has always had that “top of the line” experience. You know you can always count on them. I’ve worked with them several times over the past 6 years sending 10 mil / month and about 300k / month with zero issues. Inbound did have a hiccup with their SPF/DKIM checks though.

SES isn’t really comparable (a lot more work on your part). Mailgun has been doing some really good work though.

The word "just" implies that getting enterprise clients is easy. If they're established (i.e. old) enough and their support is good enough to justify the price point and support substantial enterprise revenue, that's a fine differentiator.

Exactly. We use them because they have a robust feature set that works for our use case (highly regulated data, etc). That's not a trivial thing. It's not a "moat" but it's wall.

I wish we could use them. Too bad they're not HIPAA-compliant (Mailgun is the only big provider that is, last time I checked)

That's a smart move by Mailgun, I'm sure Sendgrid will get there. That's the kind of thing people don't think about with enterprise customers.

- Compliance - Support - Security (not keeping things secure - offering custom security options. SAML, user roles, permissions, IP whitelisting, etc) - Customization

Building to support big complex customers is hard and if you get it right from a product and sales perspective, the inertia you can build is massive.

There's at least one big difference between Mailgun and SendGrid that I know of: Latency. I implemented an email-only authentication library based on random tokens for Django[0], and that's very sensitive to latency (you don't want the user to have to wait ten minutes for a login link).

Unfortunately, Mailgun is pretty slow to deliver emails, so I use SendGrid to send emails for that.

[0]: https://pypi.org/project/django-tokenauth/

Really surprised to hear that - we use Mailgun and haven't had any issues with latency at all.

Hmm, that's very odd, I've had problems wherever I deployed it, and multiple users have reported the same. Switching to another provider fixed the problems completely, this is very surprising indeed.

There is something of a moat in that brand new services will end up in the spam box. All of the services you mention did a lot of trial and error to see how many parallel connections they could make to big email domains, dealing with spam-marked ip addresses, etc.

Perhaps not a huge moat, but there is some secret sauce.

A concealed, sauce-filled moat? Yum.

The good thing about competing in the enterprise is that price is just one part of the overall criteria that you are judged on. Clearly, SendGrid has some winning value proposition that resonates if they are indeed more expensive and command their reported margins (I haven't personally checked).

$2b is definitely a "doesn't have a great moat but has a great business" valuation.

I reach for these any time a client asks for SMS and transactional email.

Their UX is miles better than any AWS offering so I am very excited.

Wasn't Sendgrid rejected by Y-Combinator? I guess it can be hard to tell at the early stage.

Is there a curated list anywhere of YC rejects that made it? That'd be neat.

VCs often call this their "anti-portfolio", some publish their lists. Bessemer's is super impressive:


They were pitched & passed on: AirBnb, Apple, Ebay, Facebook, Google, Intel, Intuit, Paypal, Snapchat, Tesla..

This is a phenomenal list, and quite inspiring if you look at it in the right light. Hindsight is 20/20 and VCs make plenty of mistakes.

> BVP had the opportunity to invest in pre-IPO secondary stock in Apple at a $60M valuation. BVP's Neill Brownstein called it "outrageously expensive."

Wow, that has got to be hard to swallow now.

There was a HN thread about this. Might be outdated by now but here it is: https://news.ycombinator.com/item?id=11229700

Yes. I'm surprised this isn't on here with the headline "YC-reject company Sendgrid acquired by Twilio for $2B." Maybe that will be the TechCrunch writeup.

Don't know whether they were rejected by YC, but I do know they are often cited as one of TechStars' standout companies.

"Spam company"

Congratulations to the Sendgrid team. We've been very happy with the service. Sending emails is not our core competency and with Sendgrid we don't have to worry about deliver ability. If I ever get around to hosting my personal email I would definitely use them for SMTP so that I can focus on having my inbox under my control. The free tier is 100 outbound emails a day which is more than enough for me.

Why is it so hard to email from twilio ?

Let me explain ...

The easiest way to do things with twilio is to use a 'twiml bin' because you don't need any external hosting or any third party - you code the twiml bin right in the twilio interface and it resides there. The end.

However, there is no email "verb" or any email functionality of any kind available - not even in the new twilio functions.

Let's say you want to cc: an incoming SMS to an email address in addition to its phone destination, you need to host code somewhere.

Why isn't there an email verb for twiml bins ?

I don't want a sendgrid account. cc'ing an SMS is a very, very simple and obvious use-case and I should be able to do it right in a twiml bin (or, at least, in a function ...)

Probably due to "send" in their name incoming email parsing was terrible for years and my two very specific tickets was ignored. Some crazy mess with encoding parsing, while MailGun does all the dirty job and gives clean UTF8. After moving to MailGun cannot find any difference in sending simple emails for apps (email confirmation /pass recovery)

Makes a lot of sense, and really smart to do it all-stock even if they're paying a premium to do so. This late into a bull market it offers Twilio significant downside protection on the deal.

What potential downsides should SendGrid customers brace themselves for?

Nothing. So far the details of what SendGrid will look like with the acquisition is yet to be determined, but it's pretty obvious that SendGrid will be SendGrid. If anything, you'll probably be able to integrate with SMS and voice send and receive at a discounted price.

Is this worrisome for the future of twilio? Now it sounds like they are entering yet another low margin commoditized business. What could be the synergies with the existing business areas?

My employer is a customer of Twilio and Sendgrid, and for the exact same reasons: Sending notifications to customers via SMS and/or email, depending on circumstances.

I suspect we represent a core customer type for both firms, and I rather suspect they have a pretty extensive overlap in their client list, so some obvious synergies will be around support, APIs, account management, etc.

Fundamentally, their business is about running APIs that let them accept incoming requests from people like me, processing it, and then passing it on to some very ugly external services that will (eventually, hopefully) display a text message to someone at the far end. That's a lot of overlap!

Both sending SMSes and emails require relationship management with the other companies involved - the mobile networks and the large email providers - and, related, large-scale sender-side spam prevention.

Aside from that, Twilio can now act as almost a one-stop shop for your outbound communication needs - all that’s left is physical letters.

I bet twilio buys lob.com next :)

Move over LinkedIn, here comes Sendlio+!

It seems like a perfect fit for Twilio, if they were expanding into something well outside their core offering that would be concerning. If you're a customer of Twilio there's a good chance you'll also find Sendgrid useful and vice versa.

Comment; I find myself using Nexmo and Mailgun a lot. Perhaps there's an acquisition that needs to happen there, too?

The site is not responsive. I can’t read on my mobile.

$2 Billion for an emailer...

Anybody know the valuation pre GDPR?

Completely off-topic, but it is intriguing me: why is this post remaining #1 on HN for some time if the post about Paul Allen has 3x more votes and is one hour younger. Both with an ordinary number of comments.

My assumptions about HN's algorithm seem to be way off

It’s wild that the announcement of a $2 billion deal isn’t mobile friendly. The font is so small, I bet desktop customers also have trouble reading the press release.

I just wonder what these people think in terms of designs when their fonts are way too small for anyone but teens to figure without getting closer to the screen.

Lol I thought the same thing. Bounced out of the site in a second.

wow, it's great to see a player from the modern era acquiring the bahamut from the past.

Is there a modern/past player here? Both companies were founded within a year - 2008 and 2009

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact