This "rethink," as Redmond calls it, takes place against a backdrop of pervasive mistrust and skepticism. Even people who don't believe HailStorm is evil aren't crazy about sharing such exploitable personal data as appointment calendars and contact lists with Microsoft.
It's not that Microsoft lacks experience in services. MSN counts in that category, and the company has learned much from operating some of the world's most-browsed (and most frequently attacked) websites. Passport is neither bulletproof nor widely adopted, but it works more often than not. Microsoft has proven that it can place and keep high-traffic services on the air; what's lacking is experience in hosting end-user-targeted Web services, and no one's got that expertise yet.
Microsoft is uncomfortable being first in any market, but this venture raises some uniquely thorny questions. Per-transaction billing requires an infallible tracking and accounting system that's as complex as the services being sold. Customers will be largely unaware of their volume of HailStorm traffic, so billing disputes are inevitable. Tech support overhead will be enormous, and Microsoft can't pawn that burden off on resellers or make support a profit centre as it does with its software. Microsoft should take a lesson from the @Home failure: When it comes to services, end-users may be easy to get but expensive to keep.
The HailStorm service, as originally devised, is effectively a remote store of indispensable personal data. If prospective customers aren't completely convinced of that store's reliability and security, the service won't fly. With MapPoint .Net, a set of web services aimed at commercial website operators, Microsoft tries to bridge the trust gap by offering service credits of five times the number of transactions missed during downtime.
Microsoft clearly understands that its reputation is a liability as it enters the services market. But the MapPoint .Net model makes far more business sense than HailStorm's. MapPoint .Net billing is yearly, paid in advance, sales and support are handled by partners, and has none of those finicky end-users to deal with.
In the end, the best course for HailStorm is to follow a route Microsoft knows well: Sell it as a server instead of as a Microsoft-hosted service. Hosting providers that already have the necessary end-user infrastructure can better afford the risk, and users will be free to shop around for rates and service levels. Companies that want to host HailStorm internally can buy the server instead of struggling with the issues of privacy, control, security, and billing accuracy raised by connecting to a Microsoft-hosted service.
Yager is the technical director of InfoWorld's Test Centre.