This report covers the implications of the shift to a hosted messaging delivery model, focusing particularly on economics, architectures and vendor dynamics. Given the broad implications of the shift to hosted messaging, the report is relevant to IT management, as well as e-mail managers, procurement officers, security personnel and records managers.

• Driven by dropping prices and new vendors, the long-nascent market for hosted e-mail mailboxes is poised for rapid growth over the next five years, from 1% of enterprise seats in 2007 to 20% in 2012.
• The move to a hosted delivery model for e-mail will create opportunities for new e-mail suppliers.

• Prior to any large-scale e-mail migration (upgrades and vendor replacement), enterprises should examine the pros and cons of an externally provided e-mail service.
• Organizations should establish a list of criteria that will help guide the decision-making process around on-premises versus hosted e-mail.
• Organizations should examine market segmentation models, whereby one part of the company uses a less expensive (and less functional) software as a service (SaaS) service, while other employees remain on an on-premises system or a fully-featured SaaS service.

Hosted delivery models will change the face of IT architectures over the next decade. E-mail will be at the forefront of the change in IT delivery models, and market development should be scrutinized for both tactical opportunities and strategic implications.

This document is an updated version of the document published on 22 February 2008.

This report uses the term "hosting" to refer to externally provisioned e-mail services. SaaS is one form of hosting — typically one that emphasizes a standard feature, easy provisioning, and is often sold in large volumes. An example of e-mail SaaS is Google's Gmail service. Vendors that sell Exchange-based multitenant and dedicated server model mailboxes are not necessarily SaaS, but they are e-mail hosters. Hosting, however, does not cover outsourcers, which typically supply e-mail and other desktop services together, with both on-premises and off-premises server models. Therefore, the term "e-mail hosting" refers to a range of external provisioning models, excluding traditional outsourcing.

The vendor side of the e-mail hosting market — particularly for Exchange — has been exceedingly vibrant for the past 10 years. The problem has been, however, on the buy side. Over 75 vendors have been chasing a small pool of users — we estimate that there are about 1.5 million hosted seats of Exchange in service, out of a base of 150 million users (1% of the market). The barriers to market growth have been price and brand. Only the smallest companies — fewer than 1,000 seats — find that hosting e-mail provides an economic benefit (because they have such a small number of users to defray the on-premises fixed cost investments). And larger organizations are hesitant to hand over a mission-critical business function to vendors with scant revenue and even scarcer profits, in addition to the lack of economic incentive.

But events over the past 12 months, and over the next 12 months, will create rapid growth in the hosted delivery model:

• In February 2007, Google introduced enterprise e-mail and acquired leading e-mail hygiene vendor Postini in September 2007 to buttress that offering.
• In September 2007, Yahoo acquired Zimbra, a small but innovative e-mail supplier, which it will offer as a hosted service to businesses in the first quarter of 2008.
• In October 2007, Microsoft announced Exchange Online — a hosting service for organizations with more than 5,000 seats. Later this year, we expect Microsoft to extend the program to companies with fewer than 5,000 seats with a multitenant server offering.
• In March 2007, Cisco acquired leading Web conferencing vendor, WebEx, which we ultimately believe will drive Cisco to move into the hosted e-mail market.
• In September 2007, ATT's USi subsidiary announced an aggressive migration and hosting program for Exchange 2007.
• In February 2008, Dell announced its intention to acquire MessageOne, a supplier of hosted e-mail continuity, archive and hygiene services.

Consequently, a market which had largely been the province of small suppliers has been quickly transformed into a market where the largest IT companies are aggressively competing. A similar paradigm shift is occurring on the pricing side. The leading vendors in the space — Apptix, USA.net and Intermedia.com — have long offered small companies hosted Exchange services in the range of $10 per user per month. These offerings are based on what is called a multitenant server model — where a group of servers (say 30) provide e-mail services for tens or hundreds of small companies. Building a dedicated server model to support organizations with fewer than 1,000 users is uneconomical, which leads to the multitenant server model. But even the largest of these vendors still has no more than 200,000 end-user customers. By contrast, vendors such as Microsoft, Yahoo and Google have consumer mail platforms that serve tens of millions of users (Google) and hundreds of millions of users (Yahoo and Microsoft). While these are consumer platforms, the opportunity (and challenge) for these behemoths is to transfer the economies of consumer mail to enterprises.

Google struck the first blow in February with its $4 per user per month enterprise Gmail offer. We estimate that Google's own cost of goods sold for enterprise Gmail is no more than $4 per user per year — with the biggest expenses being electricity and storage. The challenge for Google is to demonstrate enterprise-grade functionality, reliability and control on a large scale. While Microsoft has similar economies for Hotmail (aka Windows Live), its core challenge is to generate the same type of economies for Exchange. To that end, the company is building a series of hosting data centers around the globe, and making architectural changes to the next version of Exchange to better accommodate a massively scalable multitenant server implementation of Exchange. Microsoft has already demonstrated that it can develop economies in Exchange operations: with its current dedicated server model version of hosted Exchange services, it offers MAPI-based Exchange for about $10 per user per month — which is comparable to the competition and what a typical 10,000-person organization spends to run Exchange on premises. But it offers Outlook Web Access (OWA)-only services (browser-based e-mail with truncated calendar functionality) for $3 per user per month — minimum 5,000 users — demonstrating economics not available to on-premises customers.

That $3 price reflects not only Microsoft's concession on license revenue, but also a lower cost of operations, partly because the need for high-performance storage is minimized since messages stay on the server. The $3 OWA offer allows Microsoft to compete aggressively with the Google offering on price. We also expect Microsoft to offer very aggressive Exchange pricing to small and midsize businesses (SMBs) with its multitenant server model, which we expect to be commercially available in 4Q08. Ultimately — within the next 10 years — we believe the single greatest cost for e-mail operations will be level one help-desk support, bypassing both licensing and operational costs. By 2012, hosters will offer enterprises with generic mail and calendar needs an opportunity to save significant money on total e-mail operating costs — $2 per user per month, versus $10 per user per month, for example. Given the current recessionary climate, demand for e-mail costs savings will accelerate interest in the hosted model.

The demand for e-mail storage will help create demand for hosted services. Most organizations cap mailboxes at around 200MB today, and allow the creation of local archives to accommodate needs for more storage. Many organizations would like to move away from local archives (generally for legal/discovery reasons) but don't want to spend the money on storage area network (SAN) storage to accommodate larger mailboxes. Hosting vendors have discovered ways to deliver economies on storage unavailable to most enterprises. So enterprises will be attracted to the hosting model if they can get five times the storage space for the same money they currently spend for in-house e-mail deployments. Users bridling under 200MB quotas relish the idea of more space.

A final reason for significantly accelerated uptake of the hosted model will be the role of Outlook. While the Microsoft e-mail client has supported Internet Message Access Protocol (IMAP) for some time, full-blown MAPI support (which gives the richest e-mail experience available) has eluded many vendors, including IBM, Novell, Oracle and Sun. But for the first time, full-fidelity MAPI support is available from third parties. Zimbra has good client-side Outlook MAPI support, which will clearly help Yahoo's push into the hosted e-mail market (although Microsoft's potential ownership of Yahoo could scuttle such an effort). One small e-mail vendor — Postpath — also has good server-side MAPI support, and that makes it a likely acquisition target — perhaps by Cisco (Postpath is party owned by mobile mail titan RIM). During 2008, we expect Google to announce better Active Directory and Outlook calendar integration services, and ultimately we expect it to release full-fidelity MAPI support, vastly increasing the appeal of its own hosted services. The Outlook phenomenon, however, has a finite life. By around 2012, we believe the combination of lower operational costs, near-ubiquitous Internet connectivity, full client functionality, and a thirty-day local mail cache (for offline use) will make the browser client the default mail reader for most users.

The uptake of hosted e-mail will start with small companies (the only area where it is successful now), move through SMBs, and only by 2012 will the model be as appealing to the largest enterprises (with more than 50,000 seats). The reasons are simple — smaller organizations pay more per seat for e-mail services when deploying on-premises, so hosted economics are more appealing. Smaller organizations also typically have fewer customization needs (multitenant server models make customization more difficult). Conversely, large organizations already have low operational costs and have more extensive needs for custom e-mail services such as compliance, disaster recovery, content filtering and encryption.

Large organizations, however, would expect some of the techniques and technologies that hosting companies develop for low-cost operations to be made available to them from vendors that offer both on-premises and hosted versions (Microsoft and Yahoo). While that may be true, it should be noted that 1) we expect these vendors to run hosting in the scale of tens of millions of users — not hundreds of thousands of users as we see in the commercial side — making hoster economics vastly different, and 2) it is likely that hosting vendors will develop proprietary software and assemble their own hardware, which they would be hesitant to let outside their data centers for fear of jeopardizing intra- and inter-vendor relationships.

Microsoft and Google are already using the education community as a test ground for their push into hosted enterprise e-mail. Both offer no-fee services to students, staff and faculty, and Microsoft recently extended its no fee service to Exchange for students (it previously had exclusively offered the Hotmail-heritage service, now called Windows Live). Microsoft will run Exchange for students on the hosted platform, while staff and faculty use the on-premises version of Exchange. We believe this hybrid model (on-premises and hosted) will become increasingly popular for enterprises — whereby a certain population of users (those without extensive mail and calendar needs) will use OWA "in the cloud" at a low price point — while the company maintains a population of Outlook/Exchange users at headquarters (examples include car company executives vs. line workers and retail store executives vs. store floor workers). Another variant of the segmentation model would be for users with generic needs to use the low-priced hosted option from a vendor, while other users employ a more full-featured version from the hoster.

The forthcoming push into the hosted e-mail market by the large suppliers will cause a fundamental restructuring of the hosted e-mail market. Traditional multitenant server vendors such as Apptix, USA.net and Intermedia will come under tremendous price pressure from mega-scale hosting vendors. Established dedicated server model hosting vendors such as HP, USi and Unisys will fare better, based on their ability to offer larger scale and more custom e-mail services. But in the end, both categories of hosters will find it difficult competing with Microsoft, Google, Yahoo and, we expect, Cisco. IBM has announced its intention to offer hosted Domino e-mail services. Some of these dedicated and multitenant server model hosting vendors will end up reselling e-mail services from the mega suppliers. While beyond the scope of this report, we believe that e-mail is only the starting point for a vast move to hosted services. Hosting vendors will bundle instant messaging, presence and teamware services along with e-mail. E-mail will also be increasingly tied to social software tools, enabling the automatic insertion of social data into messages from previously networked senders, providing another competitive advantage to larger vendors which straddle the consumer/enterprise continuum such as Google, Microsoft and Yahoo.