Hacker News new | past | comments | ask | show | jobs | submit login
Amazon flipped a default and made me thousands of dollars (expatsoftware.com)
640 points by jasonkester on Oct 4, 2016 | hide | past | favorite | 114 comments



As a recently new customer to AWS. This service sounds really useful.

Quite possibly one of the best pricing pages[1] I've seen in a while too.

[1] https://www.s3stat.com/Pricing.aspx


If I'd had the forethought to build up a sock puppet HN account before writing the article, this is the comment I'd use it to write.

Thanks for that. Now I have two sites getting pummeled at once.


Every new venture needs to fill three roles: the technical founder, the business person, and the comedian.


and sock puppet


That role isn't covered by the business person?

/s ;P


because pets can't drive [1]

[1] http://www.forbes.com/2000/11/07/1107pets.html


So, does his post qualify him for the Cheap Bastard Plan?


Your pricing page was fun to read, thanks


AWS pricing is crazy complicated. I gave a presentation on this a few years ago (https://speakerdeck.com/jpsingleton/aws-pricing). It may be a bit out of date by now but I could update it and write it up as a blog post if people are interested?


I spent many hours with the cost calculator going through with our support company trying to find a _rough_ estimate of what our bills would be compared to Rackspace.

It's not easy initially but things like http://www.ec2instances.info/ definitely help.


> AWS pricing is crazy complicated.

I agree strongly with this. I've had clients before who would refuse to make the move up from VPSs despite numerous advantages because the AWS product descriptions and prices were incomprehensible to them.


It's not hard. Just figure out what your costs would be on dedicated or colocated hardware. Then multiply by 10.


Everything is hard if you try it for the first time.


I'd love to read an update if you're willing to put the time into it.


OK, I'll add it to the list and might combine it with some thoughts on vendor lock-in.


Thank you!



I'm not sure but I noticed it linked from somewhere else. Thanks for posting this! I'm looking forward to the additional parts.


It really is. Moving from RDS to Aurora was easier because we already had the metric data from AWS that we needed to make calculations. Moving from anywhere else would have been scary.


Definitely do it. I need someone to explain it too me. Too many ifs ands and buts for it to make sense.


It's a hilarious page, but I wouldn't call it "informatively" good, you can't easily eyeball the difference between the plans in the way the good ol' pricing grid would show.


There's a pretty 3-column pricing grid sitting unused on the server, after a day of painstaking pixel nudging then 30 days of pummeling in an A/B test against that ugly wall of text you see today.

I'm as surprised as you. But I'm not arguing against math.


> There's a pretty 3-column pricing grid sitting unused on the server, after a day of painstaking pixel nudging then 30 days of pummeling in an A/B test against that ugly wall of text you see today.

That's really interesting; how big was the difference and what do you think the source of the difference was?


I love how you repeatedly shame the cheap bastards who "cannot" afford a $10 service into taking the paying plan while still giving them the option to take the cheap bastard plan. :)


Because science! I love your response and creativity in making that page. It resonates with your audience.


I found it easy to eyeball, but it might depend on your monitor size? My monitor is big so the whole page was visible at once when I opened it.

* Less than 200,000 hits per day - $10 / month

* Less than 5 Million hits per day - $50 / month

* Over 5 Million hits per day - $300 / month


> We were initially hoping to provide this service free of charge.

Huh, the blog post doesn't make it seem like he was 'initially hoping' to do that at all.

I guess it's good marketing to make it seem like you just "need some help to cover our expenses," and aren't actually making a profit?


That hope did indeed exist, but it faded after a month or two when the first big accounts started to sign up. If you dig around enough you can find a version of that silly pricetag graphic on the server from when I hoped to cover those server expenses for $1/month.

But you're right. Sense prevailed, and I eventually started charging money in exchange for goods and services.


>>But you're right. Sense prevailed, and I eventually started charging money in exchange for goods and services.

MADNESS!


[flagged]


Try to remember to be more pleasant as well.


You're right, i could have been more pleasant, sorry.

I am personally offended that someone is using "We had hoped to provide this for free but need to make expenses" as a marketing strategy to make more than expenses. When others really are asking for money just to cover expenses.

It may not matter to some people. There's nothing wrong with charging money for providing a service in order to make a good profit. But the marketing is obviously targeted towards those who it _does_ matter to, if it didn't matter to anyone, you wouldn't need to give the impression that you were _not_ charging money for providing a service in order to make a good profit.

I am in the minority of HN in having this reaction I guess?


Just to confirm, do you, yourself, charge an hourly rate for your time? If you do, it is natural to expect that he can ask for both time and materials.


I charge an hourly rate for my time. I don't tell anyone that I had hoped to work for them for free, but it turns out I need to charge just enough to pay for my groceries, while in fact making enough to go on fancy vacations. Because that would be lying. I didn't hope to work for them for free, and I am charging more than just enough to cover my groceries.

I have no problem with him charging whatever price he can get for his service, that's business. I have a problem with marketing it as "We were initially hoping to provide this service free of charge. However, watching our bandwidth usage and server costs as people started using us, it quickly became apparent that we'd need some help to cover our expenses," if that's not true.

What's the reason to say this except marketing, because you think people are going to be more likely to pay if you say this? Does it not strongly imply that the guy is charging just enough to cover bandwidth and server costs? And is that true?

I have no problem with what he's charging, which seems a reasonable price, and really I don't care if it's reasonable or not, people can pay it or not, their choice, I have no problem with people charging prices that seem unreasonable to me, if they are successful at it anyway, good for them.

I have a problem with being misleading about your business model in order to get business, and in particular in pretending you are doing what some people _really are_ doing, but you are not, because you want to undeservedly get the good favor those people those people deservedly get.


Exactly. I just don't find the OP's idea that he tries to "spend as little time with customers as possible" and "charges the previous months just because he can" very comfortable at all. A bit too greedy-sounding in a sense.


I mean if it works for your as a business with happy customers, more power to you. We'd certainly all like to work less but still make enough money.

I just don't like it in combination with the marketting suggesting he'd providing this as a public service only charging enough to pay server expenses, when that ain't true.


This is one of those services that looks do perfectly executed I'm surprised they haven't been bought out by Amazon.

Looks like a great dev and a polished product.


I would have liked to have seen a call to action button at each price point.


I love the idea of the cheap bastard plan.


Here are my 2 cents about the pricing page you have @jasonkester

"Cheap Bastard Plan" is both little too much slang, and unpleasant to look at if you are in a demo to show this service to upper management.

I am not saying that you should have a boring, enterprisey pricing page with formal words, in fact I really like what you did there with free plan. Just saying that you can rephrase it with little more easy-going words. :)

Great product, congrats!


I would like to encourage you to rename it the "Cheap Fucking Bastard Plan" because discomfort amongst the corporate set is good for America.


The "I Embezzled the Entire Department Budget and Spent It On Hookers and Cocaine" plan


Hi, DOJ here, just wondering if you'd be willing to forward me a list of people who read that link and then clicked on it? Thanks!


Don't forget Friday beers with PFY.


You'll want Upper Management to be looking at the Features Page[1] instead. Though I should probably add some more check marks to it just to be on the safe side.

[1] https://www.s3stat.com/Features.aspx


The fact that some of the 'features' are two lines, but there's no way to differentiate rows makes that page kind of hard to read, fwiw. Not that it really matters, because who actually gives a shit about those aside from the # of check marks!


Just wanted to +1 this issue, OP can fix this if they increase the width of the container of the page.


There's a typo there - "complements", not "compliments". "S3STAT is a service that compliments Amazon's CloudFront and S3 offerings"


Or he could keep it as it is.. add a link to the page for a new page "For the Big Wigs" that's more enterprisey and uses words like "synergy"


Touche. Perhaps "Cheap Side Project Plan" would work. Most upper managers wouldn't want their project(s) to be considered mere side projects...


Not everything needs to be sanitized to protect the delicate sensibilities of America's boardroom.


Only those things that one is hoping to sell in those boardrooms...


$300 a month falls well within discretionary spend for even moderately sized orgs. This would need at most, VP level sign off.


In fact it's often far easier to get hundreds of dollars per month approved than tens. Everybody knows who's responsible for approving hundreds, no one really wants to think about who approves $10. All of the people who want to spend $10 are approved for nothing at all of the people who would approve and set up $10 are too busy for such small stuff.

I'd jack it to $750. To corporate, it's the exact same choice.


Hey Jason congrats on your product. Really liked your writing style. Very funny and informative. Nice job.

Do you worry about amazon releasing something to analyse the logs and killing your business? They just released Amazon QuickSight in beta and I think you will probably be able to import logs from their services. Of course the user will have to create the reports themselves but I guess your audience is tech savvy.


Absolutely. But it's been 10 years now, and the best they've come up with is a sort of stripped down Cloudwatch-style request count graph for Cloudfront usage. That did actually cause a bit of a dip in signups when it came out, but I think there's still a need for more detailed reports.

But yeah, the expectation is that they'll squash it dead any minute now. I'm hoping that the next thing I'm building will have replaced S3stat as an income stream by the time that happens.


Care to enlighten us on just what that is?


EDIT: Well that will teach me to read more closely! Continue below if you want to see me make an idiot of myself. Turns out I missed the point

So as I type this the votes are charging up the ranks (about 10 votes since I started typing this comment), so let me start off:

I can't possibly see how this is Amazon's fault. You're arguing that you offer a service for free (for a little while anyway), and that more people are taking you up on that because they have logs lying around, because Amazon has prompted them in the past to save these logs?

I mean, shouldn't this help your service?

a) now people can try it out because they already have logs and don't have to wait (which you've already identified was an issue), and

b) now people will end up with a bucket full of logs, think 'how can I analyse and use these logs', and go looking for a product like yours!


> I can't possibly see how this is Amazon's fault. You're arguing that you offer a service for free (for a little while anyway), and that more people are taking you up on that because they have logs lying around, because Amazon has prompted them in the past to save these logs?

You've got it backwards. The OP is saying that due to AWS UI changes, it's more likely that user's will have logging enabled, thus showing more value for s3stat. This leads to more subscriptions. It's a positive piece, not a negative!


I think you may not have read the whole post in detail. It's a positive post about an unexpected bonus from Amazons' change in default behaviour!


Well shit. I read the whole thing, not close enough apparently!


It's easy to be an idiot. Everybody does that sometimes. Kudos for your graceful reaction to being called on it. I think I'll borrow your approach next time I'm in a similar situation.


Although I mostly skimmed the article, I have to say that the feeling I got from the article changed a couple of times between positive v. negative. This left me confused and because of that I guess I missed the point as well.


For me, the words "amazon made me thousands of dollars" in the title rather set the tone.


I don't know whether the title has changed since your comment but I fail to see how "Amazon flipped a default and made me thousands of Dollars" could ever be a negative story.


It was written in a format I'm used to seeing as "grumpy rant", mostly, and much of the language used was overwhelmingly neutral.


It's not your fault though. The way the article is written it sounds(!) like exactly what you are saying.


Such a shame that Azure Blob Storage doesn't support "default pages" i.e. index.html

Then it could be used for static web site hosting just like S3 :(

See also: https://feedback.azure.com/forums/217298-storage/suggestions...


Wouldn't it be easier to use GitHub or GitLab pages for a static site?


Depends on the scenario really. If you want it version controlled, yeah sure. But there are other use cases that don't need that. You just want a "publishing target" that is low cost and minimal complexity. Plus GitHub/GitLab don't have the same number of availability-9's that Azure Blob Storage or AWS S3 can have.


True but if you don't have a billing account set up then SCM can be easier. I usually stick CloudFlare in front of it for availability (with multiple entries for the origin in the DNS). You can even add a page rule to cache the HTML in case the server does go down.


Yeah I've done the Cloudflare + GitHub pattern as well for a couple things. But not everything is suited to that, nor needs that. Cloudflare also isn't going to cache large static content for very long or at all. Or just having any cache in between at all might be the wrong solution entirely. Anyway... :) S3 supports this and AzStorage doesn't - this needs fixing!


Have you heard of Surge (https://surge.sh)? Pretty nails the 'low cost and minimal complexity' target for static hosting, although I don't know how well it scales.


Felt good after reading the story. That's a proper happy ending!


I really enjoyed the writing style :)


The way everything is formulated it sounds like successfully exploiting people, but actually it's the good side of capitalism: Someone providing a valuabe service and getting paid well for it.


Well, turns out your tool is exactly what we were looking for a few days ago.


The cheap bastard plan seems like it would be great for SEO, any stats you can share with us jasonkester on how well it worked out?


I'm actually quite amazed that a more fleshed out solution to AWS logging doesn't exist. I've worked on a small side project to pipe AWS logs to arbitrary locations for work purposes so that they could be visualized in whatever tool of your choosing

Mind if you could cover some broad details about how things are handled in the back end?


I've been meaning to write that blog post for a long time.

In broad terms, EC2 is the perfect fit for a service like this, that needs to run something like 100 hours of computing each day, but needs that all to happen during a 3 hour window before Europe wakes up in the morning. It's even more fun when something breaks and I get to spin up 200 machines in one go. For like ten dollars.

I use something like 8 different AWS services for various bits of the thing. Everything from computing to storage to queueing & mail. I even used them for payment at one point.

I'll try to get that writeup out the door.


@jasonkester Aren't you worried about attracting unwanted attention from Amazon's lawyers for using their trademark in the name of your business?

https://aws.amazon.com/trademark-guidelines/


Why should he be? Where in the document does it say Amazon has "S3" as a trademark? His service isn't called "Amazon S3stat"


Just came back here from this other HN thread (How many lines is Candy Japan code base) and your S3 stats seems like it'd be cool to learn about too. Have you ever published anything like this with some $$$ numbers?


I really hope that Amazon will provide at least basic per-bucket monthly statistics for free or a small price.

It is unintuitive that you need to parse logs or pay $10 month to find out how much you are paying for each bucket.


Jason's other blog posts are really great -- reminds me of patio11's writing style. It's a shame he's only blogging "semi-annually"!


Funny to see the americans in here torn between loving or hating the blatantly capitalistic OP ;)


Isn't a similar thing possible through ELK stack for those who are already maintaining one, the logs should be sent to Elastic Search for indexing. It might not be as customisable as this product.


Definitely possible but as is the case with most Saas products, It's much easier and convenient to just pay someone $10 a month to do it for you rather than hosting an entire ELK stack somewhere (And having to upskill to get it all working efficiently).


I agree. it works if companies are not already using an ELK stack. Looks like a great product!


Sounds a lot like https://qloudstat.com/ which I've been using for a while.


What a guy. Way to be an entrepreneur or businessman.


Very antifragile of you


The cheap bastard plan is excellent!


Off topic but about that blog: please make the text darker. I had to zoom it at 1 cm per line before being able to read it comfortably (some 400 dpi screen). My tablet is laying over a magazine and I can read its darker text at less than half the size.


Excellent timing. I was getting a bit grey-texted out while proofreading it this morning, and actually made the code change on the dev side.

But the blog started seeing a dozen requests per second before I could push the change live, and I've been in "nobody touch nothin'!" mode ever since.

But I snuck that CSS change in just now.


I think the problem is the font-weight. I wasn't really able to read the page until I changed it from 300 to 400. I don't think anyone should reduce the font-weight of their body text, it just looks too greyed-out.


Some fonts work well only with some specific values of font-weight. Maybe this is the case.


Please please make that site responsive. I do a lot of checking and things on my phone and if I can't read your site I will assume I can't use your project on my phone and go on with my life.


Tangent:

> I’d much prefer to keep those minutes for things like blowing off work for the day to go rock climbing because it’s sunny and I can do that because I run my own company. The less time I have to spend dealing with these customers, the better.

This doesn't really instill confidence in the level of post-purchase support I'd get if I were to buy in. I can easily empathize with the mentality and can even appreciate it if it's intended as humor, but all I see here is "I just want you to pay me."


Ah, but consider the support lag you get for most products, and the person from whom you receive a reply to your email (and their ability to actually do anything about your issue).

So yeah, you might have to wait until the next day for a reply. But that reply will be to say that your issue has been fixed (by the guy who built the product) and that no further action is required on your part.

People seem to like that. (And a guy can only really climb hard a couple days a week without injury, so it's entirely possible you'll find me in front of the keyboard.)


I appreciate this reply a bit more, though it's worth noting that this approach doesn't really scale all that well.

As you add customers, you'll eventually be stretched thin as you try to cater to different needs (assuming that's your goal—it might not be!). Assuming you eventually bring on other technical folks, I'd argue that you'd want to prevent this same language and mentality from persisting as a part of company culture.

But again, your goal might only be to reach a certain size and live comfortably. If so, then you're probably fine. I suppose my qualm was more with the image conveyed by your language more than anything else.


>I appreciate this reply a bit more, though it's worth noting that this approach doesn't really scale all that well.

That's OK for personal businesses. Not everything needs to be a billion dollar company.


Right, hence my third paragraph:

> But again, your goal might only be to reach a certain size and live comfortably. If so, then you're probably fine. I suppose my qualm was more with the image conveyed by your language more than anything else.


From the homepage:

> We're the sort of enterprise that is often termed a "lifestyle business" by people who feel that a software company should involve Venture Capital, sixty hour weeks, and "disrupting things". I'm sure we'd be doing all sorts of that if we were in the Bay Area, but we're not.


That answers my implied question sufficiently. Thanks!


I feel the opposite. When I hear things like "Customer service is number one at Comcast" it sounds like a hollow platitude, mainly beacuse it is.


I like the honesty. I look at it as that he has automated the most frequent stuff and that his allotment of time can go to further unique cases of customer service. Automated surface breadth allows deepness with finite time.


Well, in context he was talking about replacing something that he did for his customers in a time consuming way with something that was even easier for his customers.

I like support like that. :)


I was a user of S3Stat once. I emailed in about a problem about 4 years ago. Still no reply. I'm not upset at all as it was dirt cheap. But support is not a priority.


Ah, bummer. Sorry to miss a mail, since I do make a point of at least answering them. If you remember what the issue was, and still want to use the service, send me a mail.

I'll try to get back to you within a few months this time.


Zing lol


Not telling you are wrong, but do you really need support if product just works?


Do I ever know if I will or won't need support?

The conditions predicating the technical success of this tool aren't immutable.


You can make a guess that you're unlikely to ever need URGENT support for a logging tool. Let the dude go rock climbing!


I just read it as setting expectations, which is good.




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

Search: