Comment by ttul | Hacker News Reader

Comment by ttul | Hacker News Reader

My company (MailChannels) sends email on behalf of more domains than any other transactional email provider (see https://trends.builtwith.com/mx/transactional-email/traffic/...) because our customer base is hosting companies rather than individual senders. The whole transactional space has matured in the past year with SendGrid's acquisition by Twilio and Mailgun passing through a couple of different private equity portfolios. Mailgun is simply trying to make a profit; removal of the generous free tiers had to happen eventually.

In their S-1, SendGrid disclosed that it earned about 73% gross margins, spending roughly $18M in cost-of-goods-sold to earn $52M in revenue (six months ending June 30, 2017). Around that time, we estimate that they were sending roughly 30 billion messages a month, suggesting that they were spending $18M to send 30 billion messages - that's about $0.60 for every thousand messages.

If SendGrid's costs are universal in the industry, then that 10,000-message free tier, therefore, is costing Mailgun perhaps as much as $0.60 x 10 = $6.00 a month. If you have thousands of free accounts, the cost adds up rapidly on the P&L statement. I could be off by an order of magnitude about the sending cost, but you get the point.

The thing that annoys me is the standard operating procedure in tech is to offer stuff for free until a bunch of people depend on it and then to start charging as much as you possibly can once there's a large enough customer base that a reasonable percentage won't have any choice but to bear the updated costs.

Mailgun did a pretty nice job by offering a pay-as-you-go tier, but they still took features from the free plan and moved them into a paid tier where you have to spend a minimum of $420 per year.

I use Mailgun for less than 100 emails per month, but happen to use the receive filtering that's not going to be available with pay-as-you-go. I don't expect them to give me anything for free, but I'm not going to pay $420 / year for something that costs $.72 using your estimates.

It really has become a norm and is very frustrating for customers, while there is simply a better way to do this - kill free plan for NEW customers, but grandfather existing customers. Or at least introduce some break-even (low profit margin) plan/tier to move existing customers to, no need to force them off the platform completely.

Yes, spammers are a big issue for all ESPs (my own biz in this space is no exception), but no need to throw the baby out with the bath water.

Why? It's their business. You didn't sign a contract saying "free for life". It's sad, but there is no free lunch, and you shouldn't build your own business off the assumption that some resource will be free forever.

Free tiers are just another type of market research where the company is collecting data to determine the stickiest features.

$8 (or less) per month to keep using the feature with competition (sending). $35 a month to keep using the feature without competition (receive filtering and forwarding). Is that a coincidence?

Features and bugs. The free tier is closer to unpaid beta testers, except the free tier comes with minimal support.

How are they being forced off? Their plans are low profit already, and if customers are getting value then it's fair to ask for payment, even if it was free until now.

What value are they getting out of carrying a bunch of freeloading accounts?

Or put another way: what value are they risking by force-deplatforming the freeloading accounts, besides a bit of splashback from frustrated (non-paying) users?

The value might be a reputation for trustworthiness and fair dealing.

Those "freeloading" accounts are offered by companies to users, for the company's benefit, in order to strengthen their market position and lure users into a trap where they'll have little choice but to pay extortionate prices once they've developed a dependence on the service.

See the reputation of Microsoft and Google in backwards compatibility and support. I don't think thats the sole reason for Azure's success over Google Cloud but its certainly a factor.

I'd love to have a business (of any sort) with customer lock-in ... sadly none of my start-ups have ever achieved that position.

> The thing that annoys me is the standard operating procedure in tech is to offer stuff for free until a bunch of people depend on it and then to start charging as much as you possibly can [...]

Maybe someone more knowledgeable can help me.. but from a laymans point of view, how is this different from the practice of "dumping?" Shouldn't the FTC be more active in protecting markets from this type of strategy?

It’s not anticompetitive to offer a freemium product, unless your market position is so powerful (Microsoft vs Netscape late 1990s) that competitors cannot also afford the same giveaway.

With the consolidation of resources into Big Tech, this is quickly becoming the reality across all domains. I imagine that they will always keep cloud computing costs juuuust low enough to have plausible deniability in a court case, but still.

I don't know if emailing them would help, but I'd try it. They probably have a system to grandfather people in, who reach out. I would hope.

^^^ I agree with this. There is a very good chance that anyone who bothers to reach out to support and explain their case can get grandfathered.

"offer stuff for free until a bunch of people depend on it"

Nobody can really be surprised by that, though? Meaning people who sign up are aware that it could happen eventually?

Sustainable self-funded SaaS founder here.

Everybody should realize that:

* there is no free lunch

* a VC-funded business is looking to maximize user base and then get sold, so as a customer you are likely looking at a time horizon of ~3-4 years: plan accordingly

* if a niche B2B SaaS offers pricing below $40/month, it's not sustainable and will disappear, so plan accordingly

* if anybody offers you stuff for free, something is wrong, so take advantage at your own peril

But most people do not realize all that, and expect things to be free forever, pay for software once, or for complex B2B applications to be available forever at $5-$19/month.

I get how it works. I just dislike it. Did you have to compete against any VC funded companies?

My biggest issue with the whole model is that by offering a free tier or by subsidizing everything for "~3-4 years" the VC funded companies make it difficult for anyone to build a business that charges fair value from day 1.

It's hard to deal with because so many people are choosing the "free" option that if you try to pick something that charges fair value, you could end up picking a losing side. Or, even worse, there aren't any alternatives and you get to choose between dying on a hill or running to the slaughterhouse along with the rest of the herd.

Look as Visual Studio Code vs Visual Studio Online. Watch as VSO slowly gets more features than VSC. It'll happen slowly, and intentionally, but VSO will eventually be significantly better than VSC. Then start to think about the integration MS can do with GitHub and Azure to streamline the development process and you get to the point where everything else is an inferior product.

Now we're paying $1k+ per year for a code editor that used to be "free". Right now it's possible to sacrifice time in place of money for almost everything, but I'm worried we'll hit a point where a lot of the development process has been usurped by SaaS offerings and the only way to play will be to pay.

I'm in a small niche, which is likely too small for VC funding. That said, it's something I worry about all the time: getting a competitor who can burn cash for several years and not worry about sustainability is my nightmare.

I agree that VC funding makes it difficult to charge reasonable fees. It hurts everyone: both businesses and customers in the long term.

I would not mix a discussion of SaaS into this. I personally believe that subscriptions are the only sustainable way to develop and maintain software these days. "Buying" software, "owning" it forever and using it on rapidly changing operating systems and environments is fiction.

The very obvious takeaway is to not expect free services to be available indefinitely.

Note his estimates were for revenue less cost of goods sold. So that doesn't cover any labor, interest, taxes, etc.

Makes sense. I think the 10,000 number was way too high, but completely removing the free plan is a bit jarring and frankly annoying. I'd be much happier keeping the free plan but only 100 emails a month instead.

I'm still trying to figure out my new costs, I send approx 30 e-mails a month. I think it's like a handful of cents a month which is annoying.

I always thought 'free tier' was meant to make development and testing feasible without incurring big charges.

Having a big free tier is just a growth hacking technique, and it worked: people know the product and developeed integrations.

But, all marketing tactics have a beginning and an end. If the market is mature, and it's ROI is not good enough or has decreased dramatically, it will end soon.

Amazon SES — $0.10 per 1000 emails.

I suspect many tiny senders will just use SES...

Because it's shite. I've blocked SES's IP range on all my servers, as they just ignore spam reports.

It's not actually SES' responsibility, the sender can use SQS to receive bounceback/spam notifications and act on it.

Responsibility, shemonsibility. I want my e-mails delivered, if SES has a poor IP reputation, it's devalued regardless of whose fault that poor reputation is.

Sorry to hear about that experience. One of SES's primary goals was keeping abuse off of the service, though no system is perfect. When I worked on SES, more than 50% of the team's total time went into fighting abuse. Part of that effort was comprehensive monitoring of a variety of feedback signals, including bounces, spam complaints, and other data. Individual senders are encouraged to monitor their own sending via the Reputation Dashboard and related facilities [1].

SES also monitors these signals, and if it detects signs of abuse, such as elevated bounces or spam complaints, then a sender's ability to send may be paused and the account's activity may be reviewed. FAQ on these processes [2]. Some of the tolerates are quite tight, and can result in action being taken even if much less than a percent of messages are getting these sorts of responses. There's a lot more that I can't describe in a short comment.

SES processes and reacts to spam reports sent in standard formats like the Abuse Reporting Format (ARF), and looks at signals like List-Unsubscribe and bounces, etc. But this does require the system to understand the format of the reports used by the feedback loops [3]. Do you happen to know if your feedback loop sent spam complaints in ARF format? If you'd like to discuss further, feel free to reach out to me directly.

[1] https://aws.amazon.com/blogs/messaging-and-targeting/announc...

[2] https://docs.aws.amazon.com/ses/latest/DeveloperGuide/e-faq....

[3] https://aws.amazon.com/blogs/messaging-and-targeting/email-d...

I submitted multiple reports to email-abuse@amazon.com including headers, and received automated responses. For some spammers I reported them 4 times, but the spam kept coming.

I think you've confirmed what I said in my parent comment...SES doesn't take a zero tolerance approach.

I think it is the shared IPs that I've blocked (54.240.27.189/24). You'll see they are on multiple blacklists.

Do you literally block (aka drop) mail from SES IPs? That's awful for your users.

If you really distrust mail from those IPs so much, why not just filter it straight to SPAM? At least you'd be delivering it and your users could whitelist senders (or domains) they trust.

My users are: me. I get 100% spam from that ip range. This prevents me wasting my time on this spam.

I also work in this space. We have many customers who leave email platforms on Amazon because the IP space is polluted. Many inbox providers are extra cautious with email from those IPs because spamming is easy. Even if your IP is not used for spamming, someone else in the same /24 space could be.

I feel like they could solve that by making people buy-in to shared IP pools where the low volume pricing is a relatively high one time fee.

For example, if you want a quota of 100 messages per day, you have to pay a one time fee of $.10 per message to get into the pool, so $10. If you want a quota of 50k messages per day you have to pay a one time fee of $.005 per message, so $250.

I don't love the pricing examples (cost prohibitive for many), but I like an idea of IP pool clusters where you pay to be on a certain tier level.

> Do you literally block (aka drop) mail from SES IPs? That's awful for your users.

My good friend has worked on the email system for a mid-range consumer ISP for about the past 10 or so years. He told me 93% of all email they receive is spam. And they have a 60PB NAS for their mail store. So for every 1PB of storage they had to discard ~14PB of junk. And that’s just the stuff they summarily dismiss. Then some of the remaining 7% they accept makes it to your Junk folder.

We looked at SES and it was a no go on reputation alone. We went from a few thousand emails on Mailgun to almost 7 million emails across 140 domains and we have never encountered any reputation issues over the years we have used them. It just works and I suspect this change is to help protect their email reputation.

Reply on news.ycombinator.com