I agree with you. I am starting to externalize my chat platform msgpad (http://www.msgpad.com) after using it with my chat application ScribbleHere (http://www.scribblehere.com).
I have thought about offering a black box solution, but I agree that customers who would benefit from a hosted solution are a different type of customer to those who want something they can run themselves.
I have received questions from users about having an offline/non-hosted version of Invoiceplace.com.
Being able to provide a non-hosted version would be ideal when internet access is unavailable or unreliable - but as you stated it effectively means developing two products (albeit with a healthy amount of shared code). Not having millions in VC funding to splash around this is simply not an option for me. :)
That was a very well reasoned post. My company (Projity) has a complete replacement of Microsoft Project as a SaaS solution. http://www.projity.com Your comments are on the money ! Btw: where are you in India? I recently returned from a trip
I think you can't generalize about which is the 'right' choice without looking at the issue of what your application does and the data it touches. The more confidential or mission-critical the data, the more resistance you're likely to run into.
There's no way in hell I would be willing to keep my financial data, for example, on an offsite SaaS service. Project planning or group chat, though -- that's a different story.
This article would do well to mention not just the initial costs but the ongoing costs as well. Supporting your customers on a hosted solution vs licensed offering is an entirely different challenge and will put a great deal of strain on any size company that tries to do both.
I laugh about that comment on financial data being "offsite" Your financial data is already "offsite" at your bank, brokerage, mortgage company etc. Per the article, most buyers have connected those types of dots and realize it's better to have someone else step up to infrastructure, security and regulatory compliance than themselves.
I agree that focus is key and trying to be everything to all people is a bad idea, but perhaps there is a middle ground worth exploring...
I think it was Fred Wilson (the A VC blog) that posted a while back on users increasingly wanting to control their own data. I think over the longer term he's right.
So rather than looking this at just hosted versus non-hosted, how about hosted with multiple storage options? i.e. separate the application and the storage component.
Why not offer hosted with storage on company servers and hosted with storage on an independent storage service like Amazon S9?
The vast majority of your application remains hosted and under your control. There's no need for the documentation and support that an on-site install would need. But at the same time you can allay some customer concerns about the security of their data.
Just a thought...
Great article! One consideration re: comment #3: "There's no way in hell I would be willing to keep my financial data, for example, on an offsite SaaS service." I think while this objection may be justified in some cases quite often it is based on a flawed perception of security risks. Managers and business owners have a dangerous tendency to consider their on-site applications more "secure" than hosted solutions whereas, especially in the case of small companies, their own network is the worst security risk and it might be a better idea for them to trust a third party that knows about data security and integrity with their mission-critical data. I would rather invest my time in convincing my customers that on-site does not equal more security than to give them a false sense of security by offering an on-site installation.
I think your analysis is right on. The cost is high especially if the desire for having both hosted and non-hosted is reactionary. In my experience, the pressure for having both options comes from salespeople, which of course came from sales leads asking "Can't we have both?" The sales team should be able to answer why their product's method works best. Things head downhill when the response to a potential client is "We'll go back to the office and inquire about offering both."
We have just released a very simple (for both vendors and testers) hosted bug-tracking application. We are currently running a beta. We actually want to only do the hosted version. Lets us see where it goes in future.
Regarding the data security, do you guys think if you would be concerned to have your bug reports residing on our servers? What kind of security features would you expect?
when i did the analysis of the customers profile for Enrich (http://www.nrichsoft.in/catalog.html) the Personal Finance Software that we offer, I was astonished. Many customers were from Class B towns and what was also common is that most of them did not have even an email id.
Given the low penetration of internet access in India and the incredible and growing number of PCs in the SOHO segment at the same time, there is a huge opportunity to offer quality and affordable products in the non-host model.
Timely post for startups like us. Btw, if you are still touring India, give a mail. Will meet up or even do a conf. call. I am based out of Chennai, but freq. with Bangalore & Mumbai.
Your points about a startup needing to find customers that are similar to each other is right on. Too many startups try to hedge their bets and be all things to all people, and end up spending a ton of time creating features and solutions to serve a large diverse group of customers and are really more of a consulting company than a product company. Focus on a specific market segment, serve that one well, and then grow from there.
You also have to explore the revenue streams of both approaches. Typically, when a customer has a piece of software installed locally, they feel as if they have purchased it and prefer to pay a one time upfront licensing fee with perhaps a small support fee.
When you have a hosted service, you have recurring revenue either through subscription or on a "per transaction" basis.
For an early startup, getting those chunks of money from licenses may make more sense because it will allow you to keep afloat.. however in the long run you may find that the cost of supporting many clients running varying versions outweighs the convenience of getting the money first.
Good stuff Dharmesh. I agree with you 100%. Especially when starting up and even beyond, I think the hosted model is the way to deliver software. Thanks for the great blog and post. Keep up the great work.
I'm developing a hosted app that will store our customers' contact information such as mailing address. Customers can login and change their information at will.
There may be some security/privacy concerns there. Is there a way to instill confidence in clients that it is ok to leave their information on our server? How did Amazon, as a startup, convince so many people to store their info with them? I think some of it is that people in some way "just" trust a site they have never seen before.
I know trust*e and BBB Online Reliability offer privacy programs but many people haven't heard of them. It isn't like seeing FDIC. I also don't place much value on those programs.