The Mall @ One Market Street
October 8, 2005

The AppExchangers

Salesforce.com opens its shelves, and people rush to stock them.

The Retail Buying Experience

Salesforce.com has succeeded so far because it has brought the retail buying experience to the purchase of enterprise software. "You want enterprise software? Pull out a credit card and go to it," they tell their customers, and the customers love it.

At the heart of this appeal is what one might call an impatience that Salesforce.com customers feel with the imperfections of their IT departments. The Salesforce.com buyer wants to buy and use direct; they don't want IT to get in the way. The good Salesforce.com customer simply goes around IT. The great Salesforce.com customer coopts his or her IT department, forcing them to go along by arguing that they don't have the resources to serve his or her departmnet properly.

As a business, this has worked great. As a business model, this idea has a some pretty obvious limits. To begin this piece, I want to remind you of one. To justify the huge investment in mass-market marketing and retailing that Salesforce.com is making, it ought to have lots of products to sell. But no matter how you add, Salesforce plus Supportforce doesn't count as a lot of products.

"No problem," says Marc Benioff, at this year's Salesforce.com user conference. "Let's just sell other people's software through our site. We'll put a button on the site that leads to a product catalog. We'll give the buyer free trials and allow the buyers to post user evaluations. We'll call it the AppExchange."

It isn't e-Bay for software (though it is often described that waY), because Salesforce.com is there after the sale--and even after the delivery. A product purchsed through the AppExchange provides a software experience is seamless with Salesforce. The new software appears as a tab within Salesforce, and when you click the tab, even though you've moved to a new host and a new provider, the interface looks the same and your session context has been transferred without your having to do a thing. eBay provides the context for a transaction. Salesforce provides the context for the use of the product.

Whose software is offered through the AppExchange? Well, practically anybody's. Entry on the wholesaler side is almost as easy as entry on the buyer side. There's no charge to be listed, and Salesforce.com takes no cut. It's sort of like the camel market in Pushkar. Show up with a camel, and go to it.

It's an interesting concept. So, when I was at the Salesforce.com conference, I talkedto 10 or so AppExchangers, trying to get how it was going to work. My sample was not comprehensive, but it was enough to get an idea of the possibilities. One thing seems clear even from this small sample. The idea is nowhere near as silly as the many nasty comments from competitors might make it seem.

At the same time, said possibilities are somewhat amorphous. Salesforce.com has not worked out all the details yet, as we shall see. But even if they don't get all the details completely right, it will be something we can all learn from.

Mash-Ups

The first thing we can learn is a new word, "mash-up." A mash-up, from a developer's point of view, is a commercial application that can call a different commercial application, preserving the context of the original application.

Say, for instance, you're one of those evil people who likes to send out mass e-mails. The Salesforce application doesn't let you send out more than 250 e-mails at one time, clearly cramping your style. You'd like to get around this silly limitation. Enter AppExchange. You look on the Exchange for a service that will let you use Salesforce contacts or leads, but send out a lot more than 250 e-mails. Lo, one appears, VerticalResponse. You sign up for VerticalResponse, and it then appears as a tab on your Salesforce desktop. You hit the tab, and you go into a new application, hosted by VerticalResponse. You create the e-mail (NEWS FLASH -- NEW REPORTS FROM B2B ANALYSTS AVAILABLE NOW), and bash it out all at once, using the contacts list that you have in Salesforce.

In a very real sense, you are hovering between the two applications, never really noticing or caring about which application you're in and using data from each. A mash-up.

Underneath, the operation is pretty simple. You call the new application with a URL. APIs provide your context to the new application and let it call the contact list. The new application uses style sheets provided by Salesforce.com, which makes the interface consistent. To go into more details, I'd be handwaving, but if people are interested, I'll find out.

From a technical point of view, there's nothing new. Everything involves extensions of capabilities that earlier versions of Salesforce had provided. For some time now, people have been allowed to write Java script (or whatever) and put the URL for the script in a custom field. (This is what Salesforce.com calls "customization.") The tab access, the style sheets, the APIs, and some other mash-up capabilities were part of multiforce and customforce. The term is new, but it's leveraging capabilities that have been there for a while (at least a while by Salesforce.com standards).

(There is a new capability coming, support for Ajax, which lets you write client-side apps. But everything I'm talking about works fine without Ajax.)

Several of the charter AppExchangers that I talked have actually been exploiting these capabilities for some time. For them, the AppExchange was not a new technology; it was just a new channel, a new way of putting their product out on the shelves.

"Mash-up? I thought those were composite apps," I can hear some of you saying. And I think you're right. If there is a technical difference between mash-up and composite apps, it's invisible from the user's point of view. But there's a big marketing difference. For Salesforce users, "mash-up" sounds a lot cooler and less IT-ey than "composite app" or "SOA."

There is, of course, one fundamental technical difference between a Salesforce.com mash-up and a composite application Powered by Netweaver. With Salesforce.com, nobody has to do any work at all to set up the environment or to find the data. The mash-up provider simply points its connections toward the data center north of Market Street. This is a not inconsiderable advantage; it means the cost of entry for a developer of mash-ups is way less than it would be for a developer of SAP composites.

Interestingly, the fact that this is true exposes some hand-waving on SAP's part. They have argued that enterprise customers want control of their data, and that's why SAP doesn't move to the on-demand model. But even this simple example shows that there is a value available from the on-demand model that has nothing to do with where the data resides. Providing system context and system access OnDemand is greatly simplifying, and logically, that service is independent of a service involving hosted data. If SAP really wanted to make it easy for people to create apps that composite with SAP, they would be exploring ways for SAP to host access to apps, even if the apps themselves actually run on a customer's machine.

The AppExchangers

The companies that have jumped onto the AppExchange seem to fall into three categories: the infrastructure providers, the sales extension services, and the partners-in-crime. The first two of these have a pretty natural fit with Salesforce.com; the third has a more opportunistic connection.

The infrastructure providers give you some of the plumbing that Salesforce and AppExchange users will need. The sales extension services are trying to sell new services to salespeople. The partners in crime share only an ideological bent with Salesforce.com. They are there partly because they like the idea of cheap access to customers who are proven to share the ideology. Otherwise, they have no natural reason to be hitching onto the Salesforce.com bandwagon.

Let me talk about each category in turn.

The infrastructure vendors I talked to included Five9, which provides worldwide telephony for call centers, Fox Technologies, which provides single sign-on and account/user management, MessageSecure, which provides encrypted messaging, and the company most likely to profit from a successful AppExchange, OpSource, which provides hosting services for OnDemand applications.

Almost all these guys are small, but growing fast, and they are already seeing Salesforce as a source of significant revenue.

Five9, for intance, provides hosted call center telephony, typically to call centers of less than 500 people. Instead of buying a switch from Cisco (as Salesforce did) or Genesys, you call them up and rent the switch from them. They have about 450 clients overall, 100 employees, and are growing, they say, at 30%/month. 10% of their customer base comes from Salesforce.com, and they work closely already with the Salesforce.com sales force. "We send as many people to them as they send to us." For them, AppExchange is probably a relatively minor sales channel; the important channel is Salesforce.com.

For Fox Technologies, which already has 1000-plus customers, Salesforce is a new partner; both see the other as a help in sales to large companies (100 users or more), and Fox sees the AppExchange as an entree into smaller, 50-user installations. Fox sels to companies like Cocal-Cola, JP Morgan, Ing Financial, and Bank of America, which use it as a tool for central password management. With this in place, the Salesforce user and password management system is a slave to the Fox tool, probably something that's important to larger companies. Putting Fox in adds about $5-6/month to the cost of Salesforce.

MessageSecure is smaller and more focused on tightly controlled industries, like pharmaceuticals and financials. They provide a service that is probably necessary if Salesforce is to make lots of sales into the banking or financial services industry, where highly confidential information will end up on Salesforce.com. You don't want to be using Salesforce to send and receive messages that might contain this information, unless the message is encrypted, so you turn to them. The product goes for $20/month a user, and so far, they're encouraged by the response.

OpSource provides hosting services to companies that want to be on the AppExchange (or any other OnDemand provider), charging a percentage based on total revenue. Importantly, using OpSource means that the AppExchanger doesn't have to be certified by Salesforce.com. More on what this means in a later section.

In the sales extension service category, I'm going to include a company that might be viewed as infrastructure, Convoq, which provides online meetings. Also included are two of the many companies that allow a salesperson to send documents, Esker and VerticalResponse. And then there's Before the Call, which provides background on prospects.

Convoq's claim to Salesforce.com's attention is not simply that it provides meeting services--a commodity--but that the meeting service capability is closely integrated with Salesforce. You can, for instance, set up internal meetings using Salesforce's information about roles and push the schedule onto other people's Salesforce task list. The charge is around $7200 for up to 20 users per year; clearly, the target is all those small Salesforce installations, Convog estimates that about half the 16,000 Salesforce clients are targets. AppExchange should help Convoq, because it brings them to the attention of people who might not otherwise know about them.

Esker and VerticalResponse provide different kinds of document delivery services. Esker is most focused on fax, though it also does snail mail and e-mail. VerticalResponse is the mass e-mailing provider mentioned above; they also do design-your-own postcards. Both have been around for a while and provide these services to other CRM and ERP users. Esker numbers SAP customers like Microsoft and Air Products among its clients, but it has embraced the OnDemand idea. It is a Salesforce.com customer, turning to them after a failed Siebel implementation and frequently speaking on their behalf. They have around 5 Salesforce.com customers so far, including Gateway. Vertical Response is smaller, but serves smaller companies and already claims that 10% of its 7-digit revenue comes from Salesforce users.

Before the Call, as its name implies, provides that elusive thing, relevant market intelligence, to salespeople before they make the call. When I talked to BtC, they were going up on AppExchange soon (they are very small), but this has not yet happened. Both this company and Convoq strike me as benefiting greatly from being put on the Salesforce shelves; people who want them and can use them will be far more likely to find them.

The partners in crime that I talked to are Ketera and OpenAir. Both sell OnDemand applications, but in spaces that have relatively little overlap with Salesforce.com.

Ketera is an OnDemand provider of sourcing, whose most obvious competitor is Ariba. I spent quite a bit of time with this company, and I found its understanding of how a software business model changes when you go to an OnDemand environment really exemplary. (Execution seemed a tad more problematic.)

OpenAir sells an OnDemand project management for services providers, with a particular focus on companies that provide software implementation services. For them, an important potential customer is the services organization at small software companies. Here there is some overlap just in customer base, because many software companies already use Salesforce in their sales organizations.

Notwithstanding the somewhat tenuous connection with Salesforce.com, both companies have gone to some trouble to create a Salesforce-like interface. Obviously, they see some advantage to Salesforce providing first-level access and context services. But at the end of the day, both companies didn't care all that much about this. The primary benefit for them of being on the AppExchange is exposure to customers who are already known to be in the market for OnDemand solutions.

There you have it. A million stories in the Naked City, all of them faintly reminiscent of the stories you heard in the early days of SAP or the early days of the dot-bomb. If you find this account wearying, my apologies. But part of my purpose is to impress you with the number and variety of people who have rushed to become charter members of the AppExchange.

A couple of observations about the group as a whole:

• With AppExchange, Salesforce.com is taking a necessary step for a growing software company; they are creating an ecosystem. Big customers simply expect this; the successful creation of an ecosystem was one of the big advantages for SAP in the early days. But the costs for Salesforce of setting up this ecosystem are nowhere near what they would be for an OnPremise provider. Salesforce.com, very sensibly, is passing on the benefits of this by giving partners a low price of entry, a much lower price, it seems to me, than the comparable cost of becoming Powered by Netweaver.
• The charter AppExchangers, anyway, aren't giving anybody any grounds for the derision I have heard. They provide reasonable services, and the services pose no technical challenges that I can see. They are leveraging what Salesforce is providing them in a reasonably sensible way.
• The fact that partners or customers can post on the AppExchange seems to me a clear advantage of the system. Bulletin boards where customers of OnPremise vendors can post and download reports or minor enhancements have been around for years. Doing this as a mash-up appears to me to involve no technical risk and offers some clear advantages. Again, it's the retail experience. We've all downloaded bad software. We just erase it (we hope) and go on.
• For many of the early AppExchangers, there is clear synergy on the sales side. Salesforce.com representatives do in fact point out these applications to customers who might use them; in some cases, the AppExchanger is providing entree to Salesforce; and in others, there is an established process for going to market together. At the very least, therefore, the AppExchange is helping Salesforce sell to larger customers.
• There are still a great many unknowns. As I write, the AppExchange still doesn't actually work except in demo mode. And many questions need to be resolved before it can be as wonderful in practice as Marc Benioff the master of spin can make it sound in theory.

Costs and Benefits

Benioff announced that after the conference, he and his team were going around the world drumming up developers who could now develop their product in a back alley in Pune and have immediate, free access to the American market.

"Free" is a bit of an overstatement. To join the AppExchange, there is no direct charge, but you must be certified, which involves filling out an amazingly long questionnaire, demonstrating (through your answers) that you will have acceptable levels of reliability, availability and performance, that you won't bring the Salesforce system to its knees, and that your heart is in the right place :-). You also have to pay Salesforce.com a $10,000 processing fee, which probably doesn't do much more than cover the cost of reading the answers.

This fee and questionnaire would be enough of a barrier to deter most back-alley hackers in Pune. But they do have another option. They can get OpSource to host for them, which means, according to OpSource, that no further certification is needed. And OpSource doesn't charge any $10,000 upfront fee.

Alas, there is one other little tiny cost. Think of it as a kind of tax or use fee. If you buy and use an AppExchange package, you must first pay Salesforce.com to be a Salesforce.com user.

Let's say, for instance, that you have 10 Salesforce.com users and you want to use that $50/month conferencing package for internal meetings. For the 10 users already on Salesforce.com, you pay $50/month. But for the 10 users who don't use Salesforce, but are involved in meetings, you pay $115/month.

For a partner in crime like Ketera, whose users are going to be in the purchasing organization, this is a very serious tax. For VerticalResponse, whose users are going to be on Salesforce anyway, it's probably fairly trivial.

More generally, as long as you write a product that only appeals to existing Salesforce.com users, Salesforce takes very little cut at all. But if you write a product that has users who don't use Salesforce currently, your customers have to pay Salesforce a whopping fee for potentially small amounts of service and also pay you in order to use that product.

Mind you, this is the idea. I asked Marc Benioff why he's not charging for leads, and he said he wants more users instead. Sensible in theory, but very complicated in practice for the AppExchangers. How do they price it? You don't want to make it too cheap for existing users, but for new users, you have to consider the fact that they have to buy an expensive Salesforce.com seat, too.

I talked to some of the partners in crime about this, and off-the-record, they were willing to acknowledge that with this pricing system, they'll have to offer users the option of going to them direct. But if they do this, a system that is heavily weighted against them is now heavily weighted against Salesforce.com, which is providing them with a free channel.

Obviously, this policy is unstable. It appears to me that Salesforce.com has only two choices. One, it can continue to charge the $65/month tax on any application purchased through the AppExchange that brings it a new user, in which case the sales extension companies are the only ones that will really last, and they will always be small. Or it can find a charging mechanism that charges for the value received, taking a reasonable cut of the revenue from new customers.

Of course even that is a little complicated. There seem to me to be three different values that are received. For sales extensions products, the biggest value is the fact that Salesforce.com holds up its end of the mash-up. You can write half an app instead of a whole app, because Salesforce is providing you with data structures and data that you're using. What's a fair charge for this? I don't know.

For all products, there is also a value in the fact that Salesforce.com manages access, provides context, and gives you IP in the form of stylesheets. This doesn't seem to me to be worth $65/month, but it's worth something. Finally, for all AppExchangers, but particularly for the partners in crime (and to some extent the infrastructure vendors), there is a value in the Salesforce.com sales channel. Here, again, I don't know how to price it, for instance, whether a commission model or a retail-style mark-up model is right. But the $65/month use fee is surely not the right approach.

One other note on this. Salesforce.com also has to worry about a reverse tax, an excessive charge made by a partner who provides significant sales support or provides a unique service. Several of the companies I talked to seemed to have quite an exaggerated idea of the value they were providing. Prices of $20-$30/user/month were not uncommon for products whose value proposition seemed to me much less clear than Salesforce's was.

Well, who knows. Maybe they work it out, maybe they don't. In a sense, it doesn't matter. Here is a company who has taken the idea of composite apps and commercialized it, even while the other providers are still trying to recruit developers. Whatever they come up with, the rest of the industry needs to watch and learn.

The Salesforce.com Bazaar

One final set of issues. Salesforce.com has now entered a market where their other competitors are Microsoft, Oracle, and SAP, the market for software developers who use their platform. This entry is in some ways similar to their entry into the SFA market. They don't have the strongest product, but they appeal strongly to people who want easy, simple, and cheap access both to the software development environment and to the customers.

So far, I'd say they're actually pretty successful. The ecosystem doesn't have the size and breadth of SAP's. (It's forty booths on a small part of the floor, not the entire convention center.) And the companies aren't necessarily the most sophisticated. But they see a rocket ship going up (as one person told me), and they badly want to ride along.

Let's assume the success continues for as long as Salesforce continues to grow. Is that all there is to it? Alas, no. With success, Salesforce.com will have to deal with other challenges besides pricing..

To begin with, they're going to have to do more than the current AppExchange does in helping users to find and evaluate the products for sale there. The current system feels like eBay in the days when they had nine sellers: it's a list of products placed in a rough hierarchy. eBay had to work hard on its interface and work even harder on the reliability of the exchange. So will Salesforce.com.

Salesforce.com will also have to pay attention to risks, never a Salesforce.com strength. I have no idea if it is at all reasonable to worry about whether an evil AppExchanger can cause damage or steal information. Despite the questionnaire, it's not clear to me Salesforce.com has much more of an idea than I do.

Of course incompetence may be more of a worry than maleficence. As I said in a previous piece, the underlying design of Salesforce.com makes it pretty hard for people to bring it to its knees accidentally, but having more and more applications using APIs to make calls on Salesforce.com, any of which may have built in places like Bellevue, where software reliability has never been a priority, will mean that there are more and more tests of this design.

Finally, quality control is an issue. In eBay, ultimately, the buyer needs to beware (one reason my wife doesn't use eBay any more); eBay does protects the transaction, more or less, but that's it. With Salesforce.com, Salesforce is necessarily involved both in the transaction and in the later use of the product, so a similarly hands-off approach to product quality probably can't work.

If people buy a mash-up from Salesforce.com and it sucks, they're going to be mad at Salesforce every time it doesn't work. It's as if an eBay representative hand-delivered all those products that weren't what the web site said they were. People would get ticked at them, even if it wasn't their fault. And no matter what, anyone who tries several products on the AppExchange and discovers that they a're no good will stop using the channel, just as you stop going to a grocery store that sells rotten food.

Quality control, protecting yourself from your partners, and catalog management--all are funny businesses for a software company, er, a no-software company to be in. But if Salesforce.com want to be a retailer with lots of products, that's pretty much the direction they'll have to take. It's hard to gauge whether the current enthusiasm for AppExchange will evaporate when Salesforce.com's growth rate goes down. But so far, it looks like it could be a pretty good deal for all concerned.

To see other recent Short Takes, click here for a listing.

The retail buying experience, phase II.