Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Starting my own hosting company, any advice?
20 points by foxtrot on June 21, 2010 | hide | past | favorite | 38 comments
I have been in the hosting industry for about 7 years now, working in lots of different positions: Telephone Support, Technical Supervisor, Team Manager, Product Executive for one company then Technical Support for another (Systems Admin would be a better description)

I've been wanting to start out on my own for a while, just to generate a little extra money and my ethos is:

Transparency is the best policy

I've built my website, created my products (aimed at the beginner) and now I'm stuck. I'm worried about the failing my customers. Any advice on what to avoid or do to stay motivated and on my goals?




Do you have any experience in the marketing end of the business? I might suggest signing up as an affiliate of one of the big ones, or whitelabeling, and then trying your hand at that prior to sinking hip-deep into a capital intensive industry.

The reason: hosting is saturated and the marketing is cut-throat. Most of the companies are highly dependent on affiliates, and the type of affiliates who operate in hosting are often about one level more scrupulous individuals than the ones in PPC. (It means Porn, Pills, Casino in this context.)


Don't know what is your planned offer, but:

1) Focus on a niche. There are thousands of similar php standard setups and it will be hard for you to get out with yours, while there are far less reliable python/django/pylons, ruby/rails shared hosting platforms. If you know (or have someone who knows and will work for you) how to set up for this technologies, you can rock in this market.

2) Hire an uber-geek who can give instant support, fix stuff, install and configure needed frameworks, dbs, libraries, deal with problems, etc. If you are small, don't act big - be accessible, respond to every email, have an emergency-only phone number, give support, engage with your clients. If you have lots of devs as clients they'll probably help you (willingly or by accident) tune up your setup, you'll know the common needs, and so on.

I'm on a local shared hosting who meets both of the points above and I bring all my clients to them when it's possible.

An affiliate program would be nice, it can be simple - give discount points to devs that bring new customers. I don't even pay a dime for my shared thanks to the clients I've brought.


I'd also suggest to focus on niche. One particular niche that I can probably think of right now is "pay as you go" or "prepaid" hosting which is only provided by nearlyfreespeech.net as of yet.

I think you can experiment with billing and can come up with an innovative sales pitch. Also honesty matters a lot in this industry. It's the reason why people who host on nearlyfreespeech.net, prgmr.com and linode.com swear by their service.


Couldnt agree with you more here, I have been around the industry for a while now I have seen all the crap that goes on. The constant lies and deceptions, which is why I am making the effort off the bat to be honest. I have a section on my website which clearly says "What is not supported" this gives them information straight away with them assuming it has it and ordering the wrong package.


And avoid a mistake many beginners in business make: charge enough! Charging enough means you can do a generous affiliate program and still make money.


Seriously. With the amount often billed by consultants, clients aren't going to balk at double, triple, or even 10x the hosting costs of something like GoDaddy. (assuming reliability is there)

Also, I'd play with offering "unlimited" plans vs "cheap to upgrade" plans. Maybe the market you are targeting (if its a technical market, like django hosting mentioned above) associated "unlimited" plans with bad service or other trade-offs. A/B test until you're blue in the face.


its limited packages, aimed purely at the tiny business website. I have three packages, 10MB, 30MB and 70MB. This serves a number of purposes.

1. Limits the amount of risk from overselling 2. Increases the profitability of the server 3. Prevents abuse (people uploading mp3's, videos etc) 4. Easier Maintenance

I am flexible however, so if a customer needs a little more room I wont charge the earth for it.


70mb? You're more likely to oversell resources other than disk...


What is your value proposition? Why should anyone give their money to you rather than say Dreamhost? Transparency? What would a beginner do with that? You'll write on your blog that their site was down for three days because a hard drive failed while you were on a beach somewhere, and, no, you can't afford automatic fail-over at $9/year, what did you expect? (devils advocate, of course -- but hosting is extremely competitive)

Anyway: what's your differentiator? Once that's clear, identifying clients and delivering that value to them is much easier.


Thanks for the reply.

My target is to not oversell my systems, what most people fail to do most of the time. Offer spam protection at no extra cost, means less server administration for me and happier customers overall.

My differentiator is that whilst I offer very small hosting packages, aimed at the small business website I offer a package addon which means they can pickup their phone and just ask for something to be done, a page updated, email address created, sent a backup of their site etc etc (this is within reason). My aim is to take out the worry of maintaining a website for the new user, and allow them to focus on running their own business instead of fiddling around with a hosting control panel.


If you want to make your clients not worry, take over the entire process of being on the internet, not just hosting (there's a lot of worry in getting a website made). Get a network of freelancers and farm out the actual work, but you're the face of the work that the clients always talk to.

But there's a pitfall here: While you can make money on developing website, it's very hard to make money from value-added hosting. If you charge maybe $25/month, you can afford maybe 15 minutes of support a month, and you still face the risk of the client moving to some crappy host that only charges $5/month (80% savings!!) and they won't know what they're missing (you) until it's too late.

Either way, you're going to end up in the people business, not the hosting business -- and is that what you want?


It is, indeed, very difficult. I co-founded a niche service that offered hosting and support for about a year -- until we succumbed to scaling problems. Just be VERY sure of your business model and cost of doing business beforehand.


these days its where the money is when it comes to hosting. Im ok with being in the people business that relates to hosting.


> My aim is to take out the worry of maintaining a website for the new user

Focus on your core. Let another company deal with the hosting part, (unprofitable, requires investment and 24x7 availability) and offer your services with, say, a 24h turnaround time.


You're not in the hosting business at all then. You're in the: Get people on the internet, the easy, cheap, and fast way. This part is a bigger differentiator. Just use someone else's hosting here.


Yeah I think you are exactly right. Whilst I have done my own server before this time I have a dedicated through another company. So starting my own "hosting company" may be taken different ways.


If you do not oversell your systems, you HAVE to charge more than the average. It's simple mathematics.

Read the blogpost Dreamhost did on this, basically saying: yes, we oversell.


You might ping Hiten Shah, or at least listen to his interview on Mixergy, where he talks about how he and Neil Patel lost $500,00 trying to start a hosting company.


What I am trying to do is run the company whilst working for another company. I imagine the cost they lost was on hardware? That is where I am fine as I am running a modest server at the moment for my own websites, with enough space to sell and an easy route to increase the resources/migrate to a larger server when the time comes.


To me, that seems less like a hosting company, and more like increasing utilization.

I'm another one of the "been there, done that" crowd, back in 2002-2005 - and it was much easier then. I eventually sold for a small loss.

Rather than target the general public, try to find a couple of higher-end clients, and take care of them like they were your mother. It was those clients that ended up bringing in the most value.

Just my .02, and good luck, no matter what you do.


Thanks for the response, much appreciated.


That's what I cam here to say, I listened to that interview Friday on my way to and from work, and it's a good one. With that being said, if you have an understanding of a specific niche in the market you could probably do well.


I ran a hosting company on the side for about 10 years before starting http://www.Olark.com. Basically, hosting is pretty much a commodity right now, and big companies are willing to spend upwards of $100 to acquire a single virtual hosting customer.

Here's what I would avoid:

1) competing on price

2) building a commodity

If you focus on customer service, you won't fail your customers, your product can be mediocre i.e. (http://www.tiptopwebsite.com) and with the right connection to your customers you can still make money.

If I was going to start a new hosting company today. My cheapest plan would be $25 a month. I'd shoot for superb service, and try to find a really wealthy under-served part of the market to focus on. I'd try to build something that was unique, i.e. not a commodity. (That said, as an executive in a webhosting company, you probably know how tough it is to compete in webhosting)

Good luck!


Thanks for the comments. Your olark.com website is really nice.


Write a description of what your target customer wants and needs, (Note - may be different.) How does your service help that customer do what they really want to do?

How much will it cost you to deliver?

How much will it cost you to acquire said customer?

How much is that customer willing to pay? How much do they pay now? Why will they pay you instead?

If there's a mismatch, change what you can and go through the exercise again.

I suspect that there's a huge market one level up from hosting, but I'm not in the biz, so YMMV. (No one says "we want hosting" - they want something else that they get via hosting.)


I've been wanting to start out on my own for a while, just to generate a little extra money

To be frank, that scares me. I want my webhost to live and breath hosting, not treat it as a side-project.


There are certain people that I will not appeal to, and I want to make sure that is easily visible to everyone. One of the largest problems I have seen in the industry is inaccurate selling and deceiving customers. I don't want to attract large websites I am actively aiming at small, low traffic, low impact websites. Those are the easiest customers to support and the easiest to please.


Web hosting is the closest thing to Perfect Competition, in the economic sense, that I've been able to find. I wouldn't go anywhere near the field. There may be some niches (cloud?) that haven't reached the level where there's no producer surplus, but cloud hosting isn't as straightforward as vanilla hosting.

http://en.wikipedia.org/wiki/Perfect_competition


My initial idea would be to find some local design/web development firms and try to become their "default host" (maybe offering a good referral fee).


A great idea, now to find them and persuade them.


Sometimes when I tend to lose motivation on a project I've been working for long, I concentrate and force myself to focus on it and launch it soon to get feedback. That way when a couple of people like it, it gives me a push to continue working on it :)

P.S: If you want a niche idea I have one: start a cheaper heroku, I'll be your first customer and can get you one more for sure


Don't start one. Hosting is dead or dying.


Have great customer service!


I've got a few tidbits on the operations side of hosting that may be of interest. Since you've been around in the hosting biz a while, some or all of these things may be old hat to you. My experience is largely at mid- to large-size dedicated server hosts (1k-35k hosts), so this may not apply to your particular model, but hopefully it's useful in some way.

- Make sure your policies/procedures are clearly written and do not have any gaps or gray areas. Keep in mind that you will probably have to train a new hire from the ground up at some point, and the less hand-holding needed, the better. This goes for everything from operations to sales to billing.

- Automate EVERYTHING. Linode is a great example of how to do this correctly (although automating VPSes is a touch easier than bare-metal servers). Softlayer's web panel is pretty good, as well. The more your clients can do without opening a support ticket, the better.

- Monitoring is important. You should be notified of problems instantly so that they can be fixed very quickly, ideally before any clients notice a problem.

- Proprietary software/hardware for core offerings is generally a bad idea, unless you're hosting MS Exchange (and Openchange should eliminate that issue eventually). Keep in mind that you may have to migrate every bit of data someday in the future, and implement your stuff accordingly. This also ties into automation: proprietary stuff tends to be harder to write code for, harder to troubleshoot, and more expensive to maintain in the long run.

- Do not skimp on facilities, hardware, or network architecture. Always have hot spares to replace your live gear in case something gets fried (switches/routers, power supplies, hard drives, RAM, server chassis). This requires some investment, but telling clients "we're waiting for a new powersupply shipment from Dell, you're down for X hours" will make them spend X hours researching their next hosting company.

- If your organization is responsible for deploying hardware in datacenters, be absolutely sure that you are not overloading your power drops. If you can, get intelligent power strips that allow you to monitor load on each circuit. Know the maximum load for your hardware, in case everyone on a circuit gets slashdotted or similar.

- Do not roll out new services/datacenters/hardware without stress testing them first. Launching new stuff that doesn't quite work 100% (or will work with minor adjustments) will cause headaches for staff and clients alike.

- DO NOT LIE TO ANYONE, ABOUT ANYTHING, EVER. Transparency may be your policy, but integrity is pretty high on everyone's list, too. Admit mistakes, especially the embarrassing ones. Don't make promises you can't keep without breaking a sweat.

- When mistakes are made, take systematic steps to eliminate their causes, permanently. Examine procedural failure before human failure; the former generally leads to the latter.

That's just a few things I've gleaned from the last 6 years fixing broken servers... I may have left a few things out, but that should be a good start. Feel free to drop me a line sometime (email is in my profile) if you want to talk more about this kind of thing :)


hit the nail on the head there mate, very accurate and whilst not all applicable some are and I have thought of the processes for them.

The automate everything is tricky, I would like to offer this, but it also steers me away from the client a little as a package could get setup and they have not spoken to me about their needs. Automations with package setup all so leaves you open to abuse which incurs more cost.

Softlayers panel is pretty cool, a little outside what I want to do but inspiring all the same.


Don't.


I hear that, hosting is all I know and think it may be a good place to at least try.


No, just because it's what you know it doesn't mean that's a good place to try. If there's no opportunity then it doesn't matter how much expertise you have, it's like drawing blood from a stone. You might well be better off starting in something which you have less expertise but there's a bigger opportunity.




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

Search: