March 30, 2004

How to price subscription software—and how salesforce.com does it

by Bruce Hadley, SoftwareCEO

We think it's pretty much a foregone conclusion that if software companies aren't already offering a subscription-based version of their applications, they need to be looking at it.

That isn't to say it's the "right" or "best" way to go about licensing software; far from it. In fact, it may be entirely wrong for some vendors and some situations.

Still, it is on nearly every customer's mind, at least as a checkbox on their software evaluation form. (I.e., they may opt for a traditional perpetual license, but they want to know that you offer a subscription alternative.)

When and if you start down the subscription path, one of the first forks in the road is price. How much? How often? If you've been successfully selling perpetual licenses for $500, or $5,000, or $50,000, how do those numbers translate to subscription pricing?

To explore this subject, we went to the two most authoritative sources we could think of: Jim Geisman, founder of Marketshare, in Wayland, Mass., who has spent more than 20 years in the software industry focused on pricing and licensing issues.

Second, we called on the single software company that has probably done more to pave the way for the subscription model, and profitably so: salesforce.com.

Here's what they taught us.

Price point #1: First, review your non-subscription pricing.
"There are subtleties and nuances to subscription pricing that a lot of people don't think about," Geisman says. "You can't do a good job of subscription pricing if you do a bad job on pricing your product fundamentally."

Price point #2: Ask whether subscription pricing makes sense for your customers.
"If you have existing products, you have to think through whether you want to offer subscription sales as an option," Geisman says. "If you do, you have to think about whether people want to buy a perpetual license or a subscription."

The best way to find out: research. Don't rely on prospects who merely have checkboxes on their RFPs; talk to the people behind deals you've recently won. More important: Talk to some you've recently lost; would a subscription option have made a difference?

"Another way to do this is to decouple completely how much you charge on subscriptions from what you charge on the perpetual model," Geisman says. "But to do that, you have to differentiate the offering completely.

"For example, your subscription model may not give them the ability to upgrade to certain modules. Of course, this separation creates the difficulty of dealing with customers who might later change their minds and want to switch from one to the other."

Price point #3: Ask whether subscription pricing makes sense for your company.
Pricing is never as simple as plucking a number out of the air because it "feels right" to you, your sales guys, or a handful of customers. "If you want to encourage people to buy a car, you can lower the price of the car—and that drives you out of business," Geisman says.

"Or, you can offer them monthly payments—but then you have to determine how much you want to collect upfront. You have to look at your business objectives.

"And, you have to recognize the cash flow impact. If your perpetual license fee is in the upper ranges, the hit you'll take by going to subscriptions can be substantial.

"As with all pricing questions, it gets down to this: What are you trying to achieve here? For example, it may be that you'll make your money on add-ons, not the base product.

"With the perpetual model, you had to make your money up upfront, because you didn't know what people were going to do downstream—you'd depleted their budget with that initial purchase.

"Computer Associates did a massive switch to the subscription model, but they were in the enviable position that they had their customers over a barrel; they had the clout that they could do it. And, I think they rigged the numbers so that the change was perceived as neutral, financially speaking.

"But when you're in a hotly contested arena, like CRM, it becomes very tough. The subscription model is used as a lever against developers who don't want to offer a pay-as-you-go-model. A customer will say, 'I really like your stuff, but with saleforce.com I can get this and this and this for 30 percent less—so come on, give me a break.'

"I think you win with the subscription model by offering it as one of several ways of doing business with you. At the end of the day, the vendor is accumulating revenues, the customer is paying out money; the issue is, what is the best way for you to accumulate those revenues? A little upfront and a little over a long period of time, or a lot up front and a little over time?

"None of them is good, none of them is bad; it depends on what your organization is geared up for. You have to understand how you can run your business, and once you know that, you can talk to your customers about how they can fit into that model."

Price point #4: Consider contract terms rather than subscriptions.
"Financing the customer's purchase yourself or through a third party may be a viable option for a lot of people," Geisman says. "You'll take a bit of cash flow hit upfront, but that revenue month to month is nice. It looks like a subscription model from a customer's perspective; they still make regular payments.

"If they are faced with three options—a lump sum for a traditional license, versus financing for that license, versus a subscription model—how do you differentiate? Other than the fact that the payments never end [with the subscription plan], how do you make one model more attractive than another?

"Maybe the fact of subscription payments never ending allows the developer to lower the monthly fee, compared to a financing arrangement, where the payments wouldn't normally go beyond three years.

"But with the financing plan, it could be at a higher monthly rate, which helps the developer because it's not as much of a cash flow hit. It's a real honest-to-God commercial contract: Customers make the commitment, and they do it or not. In general, people don't default on those contracts.

"With a subscription model, it isn't as strong as commitment, which is a real potential disadvantage to the ISV. Are customers really going to commit more than a year?

"I think it's like a magazine: You're only committed to pay it for the first year, and then you can blow it off. Now, I know that with software it isn't that easy; it generally isn't like a magazine. But, still, there is a value perception. It's a more casual relationship.

"The real thing that swings this decision for all software is switching costs. If switching costs are extremely high, then it doesn't make sense to offer a financing option, because people will be prepared to pay year after year for your subscription. However, I think you have to be willing to convert some from monthly to quarterly or annual payments."

Price point #5: You have to make the new model play with the old model.
"If you have a discount schedule for volume deals on your perpetual licenses, how is that going to play out on the subscription side?" Geisman asks.

"The numbers may be so small on a per-user, per-month basis, that you don't want to nickel and dime people. If the price is that low, should your discount schedule be adjusted?" For example, it might make sense to offer a 10 percent discount when someone buys 10 copies of your $5,000 product; you can still cut an invoice for $45,000.

But if you're selling subscriptions at $50 per user per month, do you want to rent to 10 users for $450 a month? Do you want to deal with discounts, negotiations, and invoicing for a price differential of $50?

Price point #6: Forget the ASP business; stick to software.
"In the early ASP days, there were a number of flaws," Geisman says. "You've got all this infrastructure being given away for free.

"It's like the ASP guys were saying, 'I hope that people hang around long enough to pay for all the servers and people and facilities, and hopefully their usage will go up.' When you're entering an uncertain marketing you don't want to do that; you want to keep fixed costs to a minimum."

Price point #7: Allow for added expenses on the support side.
"The subscription model encourages a wide and diverse base of users," Geisman says. "You have to think about the impact on your support organization, especially as you have casual users coming online and asking dumb questions. The traditional maintenance and support fees need to be thought about differently."

One of the persuasive arguments for the subscription model is that it puts all users on the same page, so to speak, as far as version control. I.e., they're all using whatever version you tell them to use, which makes tech support a breeze, say proponents.

Trouble is, that's only true for a "pure" subscription software company (and sometimes not even then). If you're going to offer a mix of perpetual and subscription licenses, the task of tracking and supporting multiple versions and configurations actually becomes more complex. And, generally speaking, cost follows complexity.

Price point #8: Allow for added complexity (and possible backlash) with upgrades.
"Not only are users more familiar with your previous version, there also may be lots of people who don't want the latest version; how do you handle that?" Geisman asks.

"Which raises another issue, of course: If you install upgrades automatically, they darn well better work. With traditional software, under traditional licenses, you ship out an upgrade, and a few users install it at a time. Now, if you install it for them, it goes out to a lot of users all at once."

The extension of Geisman's point should be obvious, but we'll state it anyway: If you've got a buggy upgrade—and as we all know, all software has bugs—under the traditional scheme, the first few users install it, discover the problem, and let you know.

You can then jump on that bug, make the necessary fixes, and re-ship (or offer for download) a new release or patch. All those people who did not install your buggy first release of the upgrade were unaffected; most will never even know about it.

With the subscription model, everyone in your customer universe trips over your fallibility at the same time. Not only does this create a support nightmare, it hurts your credibility with your customers, as well as your PR with the rest of the world.

Price point #9: You need to offer a path back to perpetual.
"I think you need to offer customers the option to convert from subscription to perpetual," Geisman says. "The customer may say, 'I'm getting tired of paying for this. I don't want to switch providers, but can you let me out of this monthly contract?'

"If you say no, it angers the customer because it's a reasonable request. If you say yes, you have to be prepared to quote them a number you can live with. I don't think it has to be difficult; it could be X percent of what they've paid you, and you apply that toward the perpetual license.

"If you rig the payout in such a way that it's in your favor, then you say to the customer, 'It's your choice: You can keep paying this low rate, forever, or you can pay me this large amount of money to own the software now—with the addition of an annual maintenance contract, of course.'

"An underlying theme in all of this consideration of subscriptions is dealing with transitions: How do you move your installed base, how do you deal with your staffing with cash flow changes, how do you deal with subscription people who want to go back?

"As I said earlier, it could be that it's best to keep them separate for separate products." In other words, offer a particular product—probably entry-level—only on a subscription model, and offer your high-end or enterprise version only on a perpetual license.

"The real trick is when you approach a customer and you say, 'What are your needs,' instead of 'How much can you pay me?'" Geisman says. "Then, when you learn their need, you fit the package that's best for them."

Price point #10: Don't try to convert your perpetual licenses to subscription.
If a developer wants to move wholly in the direction of subscription, is it possible to force perpetual license owners into subscribers? I.e., can you convert those annual maintenance fees to license fees?

"No, I don't think so," Geisman says. "With saleforce.com, I think the issue there is that CRM is a bitch to implement, and lots of organizations spend lots of money and get no results.

"In that market, it's a way for the camel to stick its nose under the tent: 'Okay, we'll try it with 10 people, and see what happens.' It's the big bang theory in small whimpers."

Price point #11: Use subscriptions to win over skeptics, procrastinators, and small spenders.
"I think that's what the subscription model has going for it: It's easier to deal with in small chunks," Geisman says. "You can aim lower in the corporation, which means you have more candidates, and you reduce the perceived financial risk.

"The fact that they're on a pay-as-you-go basis, well, that's nice too, but there's more to it. Remember, salesforce.com started from zero; they are a great example of a business being built to match the model."

Thus, if you have lower-price, entry-level product, or a constrained of "Lite" version of your flagship software, that may be your best candidate for subscription pricing.

Price point #12: Even the big guys haven't figured out how to make the transition.
It's a much more interesting challenge, Geisman suggests, to consider how traditional companies are going to make the transition from perpetual licenses to subscriptions.

For example, Microsoft and Oracle have both announced, with differing degrees of fervor over the past four years, that they'll move to subscriptions. An Oracle VP, in fact, declared at the company's annual user conference in 2000 that "software is dead," long before salesforce.com made "No Software" its marketing mantra.

"Oracle really does rely on software, not the upgrade cycle," Geisman says. "They are counting on sales of software and sales of new licenses for their revenues. But customers are saying they're tired of upgrades, and I think the subscription model for them is a way around the upgrade problem.

"It's similar to the problem that CA faced, but CA made a conscious decision to go all subscription, all at once. I don't think Oracle will go all subscription, and that means they face the question of how much of their base will move to subscriptions, and when.

"Microsoft may get into subscriptions at the server level, but I can't see it at the desktop. The desktop has such a strong history of perpetual licenses; I think it will be hard to break that.

"I think Microsoft is trying, with their software registration procedures. When you go online to register, they know who you are, and they could, at that point, just as easily attach a 'right to use' to the 'who you are.'

"But the desktop is so mature. I don't use most of the stuff on my desktop, so why do I want to pay a subscription for a suite? So I can continue to not use it? If it's a minor toll, I may not complain about it, but if it's a minor toll I don't think Microsoft will do it, because the revenues aren't there."

Price point #13: Mixed models will work, but not without a dominant plan.
"Back when CA did it, analysts didn't get it, and CA's stock got clobbered," Geisman says. "Now, I think analysts do get it, but they still have difficulty with a software company that's got a mixed model.

"I can't think of anyone who's doing it successfully on a 50/50 basis; I think one model has got to dominate, by a sizable share.

"If you have two models, what does your sales force sell? What do they lead with? If you're neutral to both, you can take your chances, but I don't think the world plays out that way. If you do it right, I think you'll end up with an 80/20 or 70/30 mix—though which model becomes dominant is a decision each company will make on its own."

Price point #14: Don't limit yourself to standard subscriptions.
Because it is so new to software, many approach the task of creating subscription pricing by looking for clues and imitable models from other subscription businesses: magazines, cable providers, and the like.

But there may be other ways to get people into the tent, Geisman suggests. "It has never failed to amaze me that software companies offer 90-day demos, but they don't want to offer 90-day paid licenses," he says.

"What's the difference? In my book, the difference is the money you collect. I understand the logic of giving away a free demo or trial version, but when the customer comes back and says, 'I can't afford the whole thing,' why don't you say, 'Remember that 90-day trial? We'll sell you another one.'

"Or, there's another way of putting this: 'We'll sell you a one-year license, and if you cancel within the first 90 days, we'll give you all your money back. Most people don't want to sign up for a one-year license, however, without a trial.

"But if you can license someone for 90 days, and if you can charge for it, that ends up being a one-year term. It surprises me that so many developers can't get their heads wrapped around that concept."

Price point #15: Pay-as-you-go is a convenience, and convenience is worth a premium.
Geisman draws an analogy to buying a car: You can pay upfront, or you can finance, or you can lease it, or you can rent a car, or, if you don't like any of those options, you can take a cab.

"If you were paying your day rate on a car rental over the course of five years, it would be about four times more expensive than a purchase," he says.

"The freedom and convenience that you give someone for a rental arrangement is a premium; people will pay a premium to use it in a small chunk. But, they won't see it as a premium, because it's a good deal for them within a small period of time.

"The failure has always been that people ascribe no value to the convenience. When you buy pizza by the slice, you're always paying more than you would for the whole pizza. And, by the way, there comes a point when it makes more sense to buy the whole pizza and throw away what you don't want.

"But if you want to buy my software for a short time, I have some risk, and you have to pay me for that risk. I have to deliver it in small time chunks, and you're going to require more support, because with the traditional model support is delivered upfront, not at the backend.

"If you want to charge a low subscription price, hey, go ahead, but you'd better know what you're doing, because you don't often make it up on volume.

"I think the problem is that most developers assign value to the technology and product deliverable. Sometimes that's right, sometimes it is the bulk of the value; but if that's all you point to, I think you're losing sight of some levers.

"Is it convenient to buy, do you get immediate support, all those things that surround the core product? If you don't bring that out, then I think you're losing the opportunity to extract some additional money—and I don't mean extract unfairly. The subscription model is much more than a financing model."

Price point #16: Simple division isn't the answer.
"People tend to take their perpetual license prices and divide by 24 or 36," Geisman says. "Fundamentally there's nothing wrong with that, but you have to ask yourself, is that the right price for trading $100,000 in upfront money? Well, obviously, no.

"As you change your pricing or your pricing methodology, you're still going after the same customer—but with spread pricing you've got the ability to go after a whole new set of customers.

"Sure, you can be like salesforce.com: Start with no customers, get $60 million in venture funding, and rack up 90,000 customers. Yes, they're making money, after five years—but there's a really good example of what capital can do for you.

"This is a mature industry. Again, look at the car analogy: How can you start an automobile company without billions of dollars or incredibly cheap labor? You might be successful at a company that sells add-ons for cars, like navigation systems or a new style of headlights, but to sell cars, you're going to have enormous obstacles."

Price point #17: Unless you can undergo a major mind shift, you should perhaps forget about selling subscriptions.
OK, salesforce.com keeps coming up as the poster child for subscription pricing done right, so we decided to go to the source.

Founded in March 1999, salesforce.com got an initial investment of $17 million late in that year, and in May 2000 they got another $35 million; subsequent rounds brought the total to $61.1 million.

The company was cash flow positive in 2002, yet at the end of January 2004 their cash balance was $16 million—which means they burned through more than $45 million, supporting Geisman's observation that a lot of cash makes a lot of things possible.

This isn't to knock salesforce.com's accomplishments. On the contrary, we think they are a remarkable company, and have done more to redefine and revolutionize our industry than anything since the graphical interface.

The company supplies its CRM software via the Web to about 8,000 clients representing 110,000 users in some 70 countries, according to its pre-IPO prospectus. Tien Tzuo, salesforce.com's senior VP of marketing, who oversees all corporate and product marketing, says the number is now up to more than 10,000 clients.

Price point #18: Your best subscription models aren't in the software industry.
According to Tzuo, the subscription model works at salesforce.com largely because the company ignores software industry tradition.

"We look at a lot of different models when we set pricing," he says. "All of us in our normal lives are used to paying subscriptions: for our water, our cable systems, our phone service. We looked at those to figure out our pricing.

"There's something good and something bad about taking all the money upfront. If you take all the recognition upfront, you get that bump in revenue, but, long-term, it's not good for the relationship with the customers.

"With subscriptions, you have to pick what you think what is a reasonable amount of time for retaining the customer. We're fortunate in that we have five years of experience at this.

"It's not magic, but the model forces us to go deep into our relationships and make sure that those relationships are in fact successful. Software companies don't have that incentive. We all know that lots and lots of software is shelfware. If we had shelfware, we'd be out of business."

Salesforce.com sees see adoption rates of 95 percent or higher, Tzuo says. In other words, only five out of every 100 copies are not being actively used. With a universe of more than 110,000 users, that's a pretty incredible metric.

Price point #19: With subscription pricing, simplicity is key.
"When I was at Oracle, the joke was that the price list was as convoluted as possible so that the customer couldn't understand it," Tzuo says. "The sales guys had a couple of levers that they could manipulate to magically come up with the price that the customer wanted; that didn't create a good relationship."

At salesforce.com, it's about as simple as you can get: $65 per user per month, and there is no minimum; a business user can sign up one user or 100 at $65 per month. There's also a Personal Edition that's completely free; it can be converted to the Professional Edition or Enterprise Edition for the same $65 per month.

Price point #20: Raise prices only when you can prove value.
The company's subscription price started at $50 per user per month in 1999; in 2001 it was bumped to $65 per user per month, and has not increased since then.

"When we came to market we were a startup; our product was much lighter back then," Tzuo says. "We weren't yet in the CRM space. When you're dealing with the kind of size we've got now—after three years we had about 2,000 companies, and now it's 10,000—with that kind of market we have a lot of data, and can watch trends very carefully."

Price point #21: At the same time you slice up payments, slice up ROI claims.
It's easier to sell subscriptions on a monthly rather than an annual basis, Tzuo says. "If you offer once-a-month as the basis, you don't have to ask for such a big leap of faith.

"If your solution can carve out a 5 percent increase in productivity, or a 5 percent reduction in collectables, or whatever your solution does, it's much easier to show that every month. With the subscription model you probably end up justifying a higher proportion of value, because you don't have to justify the risk involved."

In other words, customers are more likely to find your ROI projections palatable when you can break them down into tiny chunks. Not only are the claims more credible, it's easy to measure them quickly, with a minimum investment on the customer's part.

Price point #22: Sell monthly, but offer longer term options.
At the same time, Tzuo says that many salesforce.com customers opt for opt for quarterly or annual payments.

"The bigger the company you get into, the less they want to deal with payments and invoices, while the smaller companies can just put it on their credit card. Many try it out for a month or two, then come back and ask for a longer-term contract."

Price point #23: At the end of the day, it still comes down to value.
Your actual subscription price will depend on value, Tzuo says: "If you go into a customer, and they're willing to say pay $3,000 for a one-time software license plus $600 every year for maintenance, how much are they willing to pay per month?

"You want to be able to say what value your software solution is providing. In our case, if the customer's sales organization is racking up $1,000 per rep per month in expenses, what's another $65 a month, if you can use it to increase productivity?

Price point #24: Subscription pricing requires a major mind shift.
Having worked successfully on both sides of the software license model, Tzuo says there are two main things to consider before you should move into subscription pricing:

"There's a leap of faith that if you do this your business will be stronger," he says.

Read that one again, and think about it. Tzuo is suggesting the equivalent of a religious experience: If you are offering subscriptions only because it's trendy, or because a prospect or sales rep is badgering you for it, or because you're grasping for new revenues, you will fail.

In the Tao of Tzuo and salesforce.com, only when you truly believe that subscription pricing will transform your company for the good, and create happier customers, should you take the leap.

His second point: "It's not just pricing; you have to see it through," says Tzuo.

"It's a leap, but it also leaves you exposed, because you're telling the customer that if this doesn't work out, he can leave anytime. At a very fundamental level, you have to change the way your software company does business."

Further reading: A few good related articles
Subscription software pricing is a hot topic, as you know; analysts are increasingly looking at other license models, too. While preparing this article, a few very good related articles crossed our virtual desk; we encourage you to check them out:

"The Power of Pricing," by Michael Marn, Eric Roegner, and Craig Zawada of McKinsey and Company, offers an interesting premise. The authors make a strong case for transaction pricing, claiming it's the key to surviving an economic downturn.

The article is part of McKinsey's Premium Member library, which means you have to pay $150 a year to access it. However, if you start with this page at Forbes.com, you should be able to access the full McKinsey article through a link that's inside the Forbes brief.

There's also a good story in eWEEK , by Jeffrey Rothfeder: "Supply & Demand: Software Pricing." It's both thorough and intelligent, though we think it somewhat overstates the rate at which our industry is changing its pricing and licensing practices.

Finally, there's an interesting albeit slightly dated article from the August 2003 issue of the Puget Sound Business Journal about Concur Technologies' shift from traditional to subscription pricing. It's called "Subscription strategy rejuvenates Concur," by Jeff Meisner.