Joel on Software, in Where there's muck, there's brass
We offer both kinds of FogBugz--hosted and installable--and our customers opt 4 to 1 to install it at their own site. For us, the installable option gives us five times the sales. ... we've got racks and racks of nice, well-managed Dell servers with plenty of capacity and our tech support costs for the hosted version are zero. Life would be much easier. But we'd be making so much less money we'd be out of business.
How to sell Qrimp is a decision I have had to think about a lot. On the one hand, I believe SaaS is the future
, but it's not the future yet. There are still customers who want their information locked up tightly behind a rugged firewall -- and I don't blame them. I'll attempt to justify their concerns, they are my concerns too, and I'll try to help you decide which to offer yourself.
Hosting is easier for the solution provider
Hosting is a much easier solution. The provider controls the environment, there's no need to fly out to the customer site to debug installations, you can concentrate on making the system good and powerful and avoid concentrating on the installation routine. That last one is only partly true -- every time someone asks for an account, we have to perform a little bit of work to copy files and create a database, but still, I can control all that from my laptop anywhere in the world.
For some systems, creating binary installables doesn't make a lot of sense. Can you imagine a locally installed version of Twitter or Youtube? Of course not. What about SalesForce? or Wikipedia? In some cases it makes sense, in others not. You'll have to look at your own product to determine, but in our case, both scenarios are viable. Due to Joel's post, I'm going to continue down the road we started and offer both a hosted solution and an installable version.
A big part of my complaint with existing SaaS providers is that they are only available from the vendor's website. If all my data is on that vendor's computers and the vendor goes belly up, what are my options? If I can install the application on my own servers, I have less fear of using the service. If it's locked up in a proprietary database written in language I've never heard of or can't be installed on less than a $40,000 server cluster, I'd be cautious putting my business critical information on it.
What if, at some point I grow to love the service and I'm at the point I want to host super sensitive information in the application? I won't be too comfortable knowing that some Database Administrator there could be sifting through it like they do over at FaceBook
. What if it's my source code? What if it's the minutes to my Board of Director's meeting? What if it's my patent library? Who would put trade secrets in a hosted SaaS?
I'm a real proponent of privacy -- especially my own -- so I think it is important to produce a product I would want to use myself.
This is why SaaS is the future
and not the now.
Scalability and Availability
The final issue I'll talk about today is scalability and the availability of Hosted solutions. One complaint a lot of SaaS customers have is latency -- especially at the end of the month. We're all procrastinators and we all rush in at the same time -- the last minute. If hundreds of thousands of people are hitting your servers all at once, scalability will
be a concern. Of course we all use the best caching algorithms money can buy and unbelievably massive clustered environments to help us respond to the load, but what happens when the internet goes down
? Power outages
The best of both worlds
So, what we decided here at Qrimp is to offer both. We're fortunate to have an architecture that supports it and an installation routine that isn't too painful. I do all the development on a MacBook Pro running Windows XP, so I had to keep the system lightweight and easy to move around. We require IIS 6.0 and SQL Server 2005 and the .NET CLR. The best part about this is that it allows us to work off-line, so if the Internet goes down, we don't lose access to all of our data, just the stuff we've added since the last synchronize.