I met Greg for the first time today and very much enjoyed the opportunity. I initially thought we might clash a little because he is pretty negative on SAP, but it just never came up. Greg makes some very good points about building a business around customer satisfaction and measuring your success with your technology as customer ROI. I get the feeling from Greg that he just doesn't mouth the words, he measures his actual results this way.
Greg is talking about why hosted delivery is a cheaper way to deliver software. He makes an interesting comment about the cost of hosting an implementation is cheaper than the cost of customer support for an on-premise implementation, even without charging for the hosting. I'm a little confused about this because Rightnow does a lot of on-premise implementations but I think what he is talking about in "old on-premise" is the kind of stuff that SAP does.
Question from Seth Henry about th 5x faster implementation and how that's possible. Greg uses the implementation at Intuit where he displaced Siebel, says it took 45 days to implement. Didn't realy answer the question but it does appear that what he is suggesting is that he has greater control over how the app is implemented.
In this mornings conversation Greg made a really good point about how hosted multitenancy delivery is good for SMB but larger enterprise customers want choice on not only the delivery option but also how much custom fitting they are getting. Interestingly he says that financial and government customers require on-premise more so... don't know why.
This is one area where I think SAP would agree with Greg, your architecture should support both kinds of deployments. The new SAP on demand CRM (which sadly doesn't have a more creative name than that) relies on the same principle, I just think we are positioning our messages wrong.
90% of RightNow customers host it
10% implement it on site
As a separate decision they decide how to pay for it, temr paid
monthly, net 30, or perpetual plus maintenance. (25% of their customers
pay monthly)
BTW, their hosting costs are 6% of their topline. This is an interesting number because in nearly all of the startups I have been involved with that host their app the cost of hosting almost always exceeds the initial model by a fairly large amount. What Greg's number would suggest is that there is a fairly significant economy of scale when you get above $x. Added later that 18 people manage 5,000 deployments in their hosting environment.
This morning Greg made a point about it not being enough to be multi-tenant, that you also had to be multiversion. Here's where he is providing some details on that, they have an automated upgrade system where customers schedule upgrades into sandbox -> production process that their customers are part of. The automated system is the key to making it scale.
Okay, upon some reflection and discussion it would appear that the multi-tenant/multi-version is really a big partitioning scheme as opposed to a shared data model.
Ross asked why forced upgrades don't work. Greg answers that the training requirements that should be part of organized upgrades are significant in systems like call centers where order taking is taking place. Although in the explanation Greg does implicitly acknowledge that he is forcing upgrades because customers only get fixes or engineering work on the latest versions. Their contracts also have standard terms about 2 versions back being current.
Integration. "Web services remove location constraints". Couldn't agree more. In going into somo of the detail here it would appear that Rightnow is channeling customization requests into configuration requests (signficant difference). They do, according to Greg, have an architecture that supports customization in a multi-tenant environment, which I would assume is the multi-version capability referred to earlier.
Good question about sales compensation. The comp model works against how a customer is paying for it, not how it is delivered.
What are the biggest challenges in processing transactions in a hosted environment? Didn't answer, but said that most of the technical issues have gone away as a result of web services.
Comment about whether or not any customers have moved over to Siebel or some other big CRM app. He says no, that they have poached customers from Siebel. However, in further discussion he does indicate something along the lines of they do lose to Siebel occassionally.
Ross asks a question about customers demanding on premise for security and single signon. Greg answers that they have had issues on performance that drove customers to want on-premise, and then goes on to say that they shipped a rich client pre-AJAX (which I guess means they built it in DHTML).
What's the strategy for building a partner network? Different for hosted versus on premise deployments. Greg answers that direct selling is the way to go to market so that would suggest he is not building out a reseller network.
What does he worry about? That Salesforce.com fixes their reliability issues.
Jeff,
Thought on a question you raise:
"Interestingly he says that financial and government customers require on-premise more so... don't know why."
- Might this be because of legal/regulatory issues regarding control of client data? Seems these would be two sectors especially subject to this type of regulation.
Posted by: Mark Crofton | Feb 07, 2006 at 06:17 AM
A reseller hosting ( http://www.opalhost.com/reseller.html ) network would work best only if there's a significant amount of referral traffic to the reseller's site.
Posted by: azala | Aug 19, 2009 at 03:09 PM