Take a Deep Freaking Breath, The Cloud Still Wins!!!

Coping with the B2B Cloud Valuation Meltdown

Woman breathe fresh air

Businesses to Business Cloud Computing valuations have dropped off the cliff in 2016. And I mean a big cliff. Its bad! If your heavily invested in B2B SaaS, you have been puking on your shoes for all of 2016. The Bessemer Cloud Index that does a decent job of tracking B2B companies is down 33% year to date versus S&P500 being down 9.1%!! And the biggest independent SaaS companies are way down: Salesforce -24%,; Workday -42%, Splunk -47%; ServiceNow – 42%; Netsuite -49%. Athena Health -26%; Even market darling Atlassian is down 33%. If this were a prizefight, they would stop it! If cloud stocks were their own market, rule 48 would have been invoked in 2016 and circuit breakers would be getting a frequent workout.

Things have gotten so bad on valuations that revenue/market cap ratios for SaaS companies are similar to those of old line on premise software companies.

Now, I am an operator and company builder, not an investor or valuation expert, but this blog post is here to say that multiples of B2B SaaS companies should NOT be similar to old line on premise hardware software companies.

Why? Because the new B2B SaaS companies will eventually get a majority of the old line hardware/software company’s customers and revenue. Cloud computing is simply a vastly superior way of delivering value for clients.

So it’s worth revisiting why cloud computing wins over the old paradigm. I gave the “Why cloud” speech hundreds of times, while I was evangelizing the cloud for Google in 2007-2013. In these troubling times, it’s worth revisiting my old notes, which are more true and relevant today.

Cloud Computing has four incredible, mind-blowing benefits over the traditional on premise hardware and software model. I will list them in order of importance: 1) Speed of deployment 2) Elasticity, 3)Security and 4)Cost.

Speed of Deployment

On premise hardware and software programs are unbelievably slow to write, deploy and update. And by unbelievable slow, I mean we measure cycles in increments of decades!!! No, I am not sh*ting you here. It is highly typical for companies to take 5-10 years to deploy new applications or even new functionality on existing applications. Don’t believe me. Read on!

The best software company in the world is undoubtedly Microsoft. There are other great software and hardware companies, but none is a good as Microsoft. They have ~$100B in Annual revenue, Earnings of ~$25B; Market Cap of $400B. There is a reason Gates is the richest man in the world. His company has the undisputed king of the on premise software world.

Now how fast was Microsoft with on premise software? Fortunately we can easily track how fast they developed software because they did us a favor my naming their products by years. Here are the major release cycles of the world’s most popular application program, Microsoft Office:

  • Office 95
  • Office 97.
  • Office 2000,
  • Office 2003,
  • Office 2007,
  • Office 2010,
  • Office 2013,
  • Office 2016

So, on average they release software every 3 years. Every 3 years!!! And that’s from the best software company that the planet has ever known

Worse than that, customers would not or could not easily deploy new software, and they would often be years behind those release cycle.

When we launched Google Apps in 2007, I went to see General Electric. This company is the hallmark of management practice, an example for all of how to run companies. GE bragged that it spent $4B on IT. Surely they were faster. Nope. When I went to see them in 2007, they were using Office 2000 products and would do so for another 3 years. GE users were forced to use software that was 7 years old. Can you imagine using technology that is seven years old? Can you imagine a mobile phone that is seven years old? Would you buy from Amazon if you were using their website from 7 years ago!!!!

Ever wonder why, it takes the United agent, 500 keystrokes to change your flight or your seat. Its not the agent’s fault, he is using on premise reservation software that was originally written in the Nixon Administration!!!

The cloud allows massive improvements in speed of development, deployment and end users get functionality quickly and seamlessly. Applications design and development is less monolithic. Cloud applications can be updated in real time, you don’t’ batch up features and release them every three years; you introduce them as they are developed. At Google we could put new features in every day or every week. Secondly, because true cloud applications are essentially centralized version NOW, multi-tenant instances, they are orders of magnitude less complex to deploy.

I could go on, but you get the point. End users are now accustomed to constantly updated cloud applications liked Facebook, Google, Amazon, Snapchat, etcetera, they are not going to put up with using enterprise software that is a freaking decade old!!!!


Cloud applications, at least properly architected applications ones offer the ability to cheaply and easily build applications at scale that can withstand heavy loads without significant performance degradations. Better yet, you can rent the computers and software stack from cloud vendors and only pay for what you use. If your application is a flop and there is not much usage, you don’t pay much. If you application is a hit, you application won’t crash while you are ordering hardware and installing software.

This is especially important for your most crucial applications, your customer facing applications. These applications might need to scale to hundreds of thousands or millions of users and have very spiky workloads. On premise hardware and software model was largely designed for internal applications – those used by company employees. In the pre-cloud days, only GE, Wal-Mart and GM type companies had to worry about scale on applications. But when the cloud age hit, many companies delivered computing to their customers where scaling is paramount, and performance is critical. Further the Internet age was accompanied by globalization, so for the first time IT departments had to build globally scalable applications. Most modern cloud applications come with built in global scalability, if you build on premise applications, you will be building data center capability globally.

In the on premise world, we would estimate workloads, build capacity for peak demand, deploy hardware and software into multiple datacenters and hope the application was a hit. If it was a hit, our servers crashed and we scrambled to build more expensive infrastructure. If the application was not a hit our software and servers sat idly by, burning up cash by the minute.

In the cloud, you can let others worry about this. Hardware and software infrastructure are much more elastic. Build the application, run it in cloud and forget about estimating workloads, determining server capacity and factoring in lead times for hardware and software deployment.


It is still sacrilege to say this. But I said it in 2007 and I will keep saying it. The public cloud is much more secure than traditional on premise software and hardware. Security is a benefit of the cloud, not a liability.

I am not saying that you should not do deep security due diligence on your cloud vendor. You should. You can build crappy cloud apps, but most people don’t and securing cloud applications is much easier than securing on premise applications.

The reason that cloud applications are more secure than premise applications is that security for on premise applications is a virtually impossible task. Let me explain how impossible it is for even the richest and most competent IT organizations.

Everyone will agree that all software is vulnerable. It must be continually patched. We are now paying outsiders to continually hack our software, identify vulnerabilities and then quickly issue fixes and parches for that software.

So the key to secure applications is the ability to patch quickly.

Traditional on premise applications are highly decentralized. They exist on many servers. We are talking 10’s to 100’s here. And software is deployed to desktops where there are 100’s to 1000’s of those computers. So even with virtualization and automated patch systems the task is massive. If you are Fortune 1000 company, you may be able to cobble the resources together to be half way decent at this. But if you are a mid sized organization, you cannot hope to be able to handle this. (I have a great story from an old customer on this, but that’s another blog post)

Secondly, the application stack that most companies deploy is incredibly complex. Operating systems, database, application servers, countless layers of middleware. It is very complex. And those layers don’t typically come from one vendor, so when one vendor identifies a security vulnerability and sends you a patch there is no guarantee that you can deploy that patch because it may not work with the other layers of software. They will likely need to wait until other vendors catch up before they can deploy the patches. And patches are further delayed because IT departments need to do testing on the application to ensure that the patch does not break something.

Lastly, most on premise software applications are not designed to be patched and updated without “taking the system down”. This is mind boggling to young people. They can’t believe this. Amazon updates it software without closing its store, Google updates it software without taking search down, You never go to Facebook and see a message that says “the system is down for routine maintenance”. But that is how we live in the on premise world. We have to take applications offline to patch them, but our maintenance windows are getting very tight. Business is 24/7 globally, and applications are now critical to the functioning of the business. No applications, no business. So many organizations, just batch up their patches and roll them our periodically in order to ensure application stability and reduce downtime.

But that means your systems have known vulnerabilities for extended periods of time. Forget about zero day exploits, the bad guys can hack systems by using known vulnerabilities.

With on premise hardware and software, deployed in the traditional fashion, it is virtually impossible to keep your systems highly secure. We know this in IT we just don’t like to talk about it.

Cloud based systems are highly centralized, easier to patch and typically designed to be patched and updated with no downtime.

Cloud based vendors also have extremely large and competent security teams. I am not saying they are perfect, but they are usually larger and better than the IT security teams that non–IT companies can build. Security specialists are in short supply and do not come cheap. You can do your best at building a great IT security team, but in the end, on premise IT systems are like storing money in your mattress. You might feel good because your data is at your location, but its only safe if you can vigilantly protect your house. Might be better to store your money at the bank where they have professional security.


Everyone talks about this as he the main benefit of cloud. It is a benefit, but it is fourth of my list because you should do cloud, because of speed, flexibility and security. Lower cost just happens to be a side benefit of cloud and a pretty damn nice one at that.

When we introduced, Google Apps in 2007, we charged $50 per user per year. The average loaded cost of a Microsoft Exchange email box was $500 per user per year. Our systems were 1/10th the cost of on- premise software! And much of that $500 expense was not going to Microsoft license or maintenance fees. It was the cost of hardware and an increasingly large IT staff that are need to deploy, maintain , and secure internal IT systems.

Its not just Google Apps, Workday is cheaper than PeopleSoft or SAP, Salesforce is cheaper than Siebel, Netsuite is cheaper than Great Plains software. ZenDesk is cheaper than Remedy

It is not only less expensive to buy and run cloud software, but it is less risky to procure. Price entry points of traditional software are very high as large perpetual licenses need to be bought, followed by 20%-25% maintenance fees. On premise software is so expensive and risky because of its high upfront costs. On top of that, large deployment and customization charges are typical in the on premise world. And once you have deployed software, switching costs are very high. So the vendor has a lot of power in price negotiations. Maintenance fees were about 12% when I got in the business in the mid eighties. They are now 20-25%.

Cloud based software is purchased annually with no exorbitant upfront costs. If you don’t like the software or your needs change, you simply don’t renew. You don’t need to marry your cloud software, you can just date it.

The cloud will win. It is a simple matter of fact. I don’t know what is going to happen with valuations, but I do know who is going to win this battle.

So if you are an investor, entrepreneur or employee of a B2B cloud company it is ok to hyperventilate or puke on your shoes. But take a deep breathe because the cloud wins!

Atlassian or Box.com; Is “Sales Light” or “Sales Heavy” the right model for you?

5 Questions to Guide You

Atlassian is about to go public in what might very well be the best Enterprise Tech IPO since Workday. On top of great products and profitability, the headline of this IPO is their sales model; or better stated – there lack of a sales model. Here is the paragraph of their F1 that everyone is talking about. I have added the highlights for emphasis:

We founded our company on the premise that great products could sell themselves and we have developed a unique approach to the market that is centered on this belief. We begin with a deep investment in product development to create and refine high-quality and versatile products that users love. We make our products affordable for organizations of all sizes and we transparently share our simple pricing online. We pursue customer and user volume, targeting teams in every organization, regardless of size, industry or geography. To reach this expansive market, we distribute and sell our products online without traditional sales infrastructure where users can get started in minutes without the need for assistance. We focus on enabling a self-service, low-friction model that makes it easy for users to try, adopt and use our products.

Wow! Talk about disruption They are selling $319M or Software and SaaS with no traditional sales force!

As a result, Atlassian sales and marketing expenses are only 12-21% of their revenue. Compare that to other Enterprise SaaS companies who spend between 50-100% of revenue on sales and marketing. See Redpoint Ventures Tomasz Tungus’ excellent analysis of Atlassian’s Superior SaaS metrics.

Now let’s look at Box.com who went public this spring and is doing quite well.  Revenues are growing at 38%, losses are narrowing and profitability is on the horizon.  That having been said, they are the polar opposite of Atlassian. Sales and marketing expenses are still 81% of Box revenues and at times they were spending an astounding 200% of revenues on sales and marketing.

So, can we all just copy the Atlassian flywheel model. Can we finally dispense with a golf playing, wine drinking, first class flying, face to face field sales force and let the computers do the work?

It’s easy to look at the Atlassian and Box numbers and declare Atlassian the winner on not only efficiency, but in effectiveness. It is a larger company than Box ($319M vs @280Mish in trailing 12 months revenue). And Atlassian is growing faster (46% to 38%)

But before we rush to any conclusions, it is also worth noting that the undisputed heavyweight champion of SaaS, Salesforce.com, uses a “sales heavy” model and is fact now headed by Keith Block and Tony Fernicola. These Oracle alumni are industry veterans of the “sales heavy model”. And it’s not just Salesforce and Box. Most of the leading Enterprise SaaS companies use a “Sales heavy” model. Workday, ServiceNow, Netsuite, Marketo, Eloqua, Concur and Successfactors all employ a lot of salespeople. And even companies who target the lower end like Hubspot spend a lot on sales and marketing.

As for other good examples of “Sales Light” models, we should look at Amazon Web Services. They have sales people now, but a lot of their business was built in the low cost frictionless sales model. Another example would be the Google At Work unit,that I ran for eight years. We certainly had lots of salespeople, but large sections of our business were designed to be low touch frictionless sales and mid market and enterprise units were run below industry average expense levels. The other non-public examples of sales light model would be Github who has grown to 10M users worldwide. But even they hired a VP of Sales 18 months ago and stated that their latest $250M fundraise will be used in part to expand sales.


So, what is right? “Sales Light”, like Atlassian or “Sales Heavy” like Box.com and Salesforce? The answer of course is: “It Depends” and “You may need more than one sales model”.



But I am not going to cop out with the ubiquitous “It Depends” answer. Let me give you 5 factors to consider when making your decisions about your sales model

Factor 1: What is your target customer’s buying process?

We talk a lot about sales process in Enterprise Tech, but in a hyperconnected world where buyers possess access to almost infinite information, we need to think about how customers buy, not how we sell. You need to ask how they have traditionally bought and how they are buying in the new normal.

Small businesses have usually bought quickly and with smaller levels of consideration. Frequently there is a single decision maker and they are focused on ease of implementation, low cost and end user simplicity. They are not often interested in competitive evaluations unless it helps drive price down.

Large enterprises have traditionally bought slowly with very high levels of consideration. There are many stakeholders involved in buying technology. They often have established and cumbersome procurement rules that must be navigated. They will have very complex decision criteria and will focus more on total cost of ownership and ability to integrate with existing complex infrastructure. They will almost always require a competitive analysis as part of the buying process that involves significant dollars.

Medium sized business is traditionally a mix of the above- more than one decision maker, but not hoards of people; Some structure in the buying process, but not too much.

Buying processes also differ by industry and region in the world. Buying processes in government and regulated industries (Finance, Healthcare, Pharma) are even more stringent. European buying processes can differ from American ones and Asia Pacific buying processes are clearly different.

So, the general rule is, if you target large companies and regulated industries you will trend towards sales heavy and if you target smaller companies and less regulated industries, then you can trend towards “Sales light”.

Lastly, you should consider who your target buyer is within the company. The sales light model has worked well for Atlassian partly because they target engineers and developers. This target buyer is highly skeptical of salespeople and marketing claims. Taking them golfing is not likely going to help the sale move along. They want to see the product, look at the nuts and bolts and decide based on hands on work. Furthermore, developers can often choose tools individually or in small groups without engaging with a large team of stakeholders. And once, you convince an engineer or developer about your product, they are likely excellent sources for word of mouth referrals. If you are selling compliance software to the CIO or Chief compliance officer, you are much more likely to need spend a lot more sales time doing belly-to-belly meetings.

Factor 2: How Mission critical is your product?

Every company wants their product to be called mission critical, but be careful for what you wish for. Mission critical is often defined as “Will the business stop if this application is down” or “Will there be significant impact if this application is down“. ERP software for manufacturers is mission critical, Patient record systems are mission critical to hospitals; Websites and ecommerce systems are critical to online companies. Points of sales systems are critical to retailers.  Email and Payroll systems are usually mission critical for everyone. But there are a whole host of important software that is not mission critical.  If you are selling “succession planning software” or “recruiting software” or “e-meeting software”, then you fall into the important, but not crucial category.

The sales model implications are as follows. Mission critical software requires a more sales heavy approach and important, but not mission critical software can trend more towards sales light.

Factor 3: Is your first sale a “Bunt single” or a “Grand Slam”

Certain software sales are an “all or nothing’ sales proposition. Customers decide once. There is usually a large upfront sale for all users. If you win, then you are golden. If you lose then you are likely out of the account for a very long time. Salesforce.com is a good example of this. People make decisions about CRM, Service and Marketing automation once and then work very hard to succeed with it. They do not revisit that solution every year, nor do they allow departments or divisions to deviate from the corporate decision

Other software is bought is small chunks. It can be bought by a department or as a corporate trial. Customers make a small purchase, they watch how adoption goes and then expand use over time as they see value.  File sharing software from Box.com, Dropbox would be a good example of this. E-meeting companies like Webex and “Go to Meeting” are another example.

If you are a Grand Slam company, you will trend more towards Sales heavy and if you are “Bunt Single” company, you can trend more towards sales light.  But be aware if you are sales light, you likely need to invest more in customer success than the Grand Slam company.

Factor 3: How Viral is your Product?

Viral is an overused and poorly defined term. But I will use a simpler definition. How much effort is needed by your company to spread usage of your product? If it a lot, then you are less viral, if it is almost none, then you are very viral. Some products are not naturally viral. ERP and Patient Record systems are not viral. They take a large amount of resources to “get live” and then the company mandates them. No viral success needed. Other technologies like e-meeting software are more viral. Customer buys some limited amount of licenses, they start hosting meetings, people have to have access to software to attend meeting. Other departments or divisions see you using software and want to come on board. This is a really good model if you can make it work. Unfortunately we have many potential enterprise software categories that are viral in nature, but whose products are not user friendly enough to go viral in a big way. I mean have you ever left a Webex meeting and said “Wow, that was great, I can’t wait to do that again.” So be careful here, if you want to get the effects of viral distribution you have to have the right product and it has to “delight users”, not just meet requirements. Another way to say this is that you need “Net Promoters” at the end user level- people who rate you 8 out of 10 and above. And it has to be at the end user level, not the IT or support level. Most Enterprise companies are doing net promoter stuff now, but they are surveying the wrong people-Central IT or Project Management staff.  You need to survey the end user.

So bottom line on this factor, if you are viral, you can trend toward more sales light. Use a sales force to land the big account and let the product and customer success teams/process take it from there. But be careful, if you are not truly delighting end users, then you will need heavier investments in customer success than your CFO will like.

Factor 4: How competitive is your market?

Every market is competitive, but some are more competitive than others. If the customer has many alternatives in your space, then the sales model will need to be heavier. Customers who perceive that they have many choices, fear making the wrong decision and want to make sure they get the best price. If your product is a newer category, where customers perceive uniqueness, then you can be more sales light. To use a consumer example, If you want to buy a luxury sedan, you will feel required to look at BMW, Audi, Mercedes and then likely Infiniti and Lexus, GM, Ford and Chrysler. But if you want a luxury electric vehicle, you will stand in a (virtual) line in a get a TESLA.

Factor 5: How hard is to implement your product and What is the time to value?

Products that require lots of professional services to get started are tough to sell in sales light environment. If the customer cannot take some very quick steps to get live and start getting value, then you will need a much heavier customer engagement model. Some products are naturally heavy in implementation services because they require configuration and business process design. These products usually fall into the “Grand Slam” category because customers cannot “trial implement” them or adopt them in chunks. You will need a sales heavy model to assure customers about your solution because it will cost them money and time before they see value. There is higher risk, so they need more belly-to-belly time. On the other hand, solutions that can be quickly implemented and do not require a lot of expense to get value, can trend towards a sales light model because customers will perceive lower risk. If it doesn’t work or users don’t like it, we are not out much…so let’s just buy it”

Please note that as you move up the stack to sell to the largest companies it is not likely that you will be “full self service implementation”. You will be required to integrate into to the customer’s proprietary environment like directories, single sign on systems, mobile security systems and archiving systems, just to name a few.


There is no one right answer to the Atlassian vs. Box.com question. But there are clear questions that you can ask yourself to determine your sales model. Further, if you plan to tackle the whole market from small to large and across many industries and geographies, then you are likely to have more than one sales model.

I also recommend that you stretch the boundaries of the sales light model as far as you can. Sales light can be used beyond small business and there is evidence that it can work much further up the food chain, so don’t default to sales heavy too soon.

You can also deploy hybrid models. In the mid market in Google Apps , we used a sales light model for the first step of the buying journey, but deployed a team based field approach to the final step of the sales process. That change had a big impact on our close rates.

Lastly, your sales model is not a standalone business strategy. It is very closely linked to your product strategy. The simple fact of the matter is that Enterprise software has been so horrifically bad that sales light was not an option for most companies. If they saw or understood the software without a thick and expensive sales force translating it for them, then there was no chance of a sale or adoption. We simply must make better enterprise software.

Your sales model strategy is also tightly tied to your customer success strategy. In the past,we didn’t call this customer success, we called it support. We published a 1-800 number, we fixed bugs and we let third parties implement our product. We are now developing robust customer success departments that do implementation, end user adoption, project management and support.  And we are learning to do customer success functions in a high velocity, frictionless environment as well. If you are a “bunt single company” that depends on land and expand, then you will often need to invest more in customer success than your “grand slam company” and you will have to work hard on your product to ensure sales expenses are simply not replaced by customer success expense.

So there you have it. Picking your sales model is not black magic. It also should not be a religious war between the wine drinking golf playing, first class flying old guard and the “all on line” new guard. Answer these five questions and pick the right blend of sales models for your product and target markets.

Defending the (Enterprise) Unicorn

The unicorn is under attack. Not a day goes by now where the tech press is not publishing a “Who will be the first dead unicorn” story. Every tech conference contains lots of fireside chats where we opine about the overfunding of late stage companies. We also seem to now be drawing parallels between what we are seeing in 2015 and what we saw in 2000-2001.



So while everyone is freaking out. I would like to write a defense for some of these unicorns and more specifically for the Enterprise unicorns. But first some caveats. I am not saying that all Enterprise unicorns are all huge winners and I am not saying that some of the Enterprise unicorns are not potentially overvalued. I am also not saying Enterprise tech companies are not immune to dying or being sold for scrap. Indeed, while Webvan and Pets.com were B2C companies and became the poster children for Web 1.0 bubble, There was plenty of Enterprise tech carnage to go around. Remember these names: Ariba, CommerceOne, Epiphany, i2? I don’t need to go on, it is too painful.

So with those caveats, let me lay out 5 reasons why there is no overarching crisis in Enterprise Tech valuations.

1. Enterprise Tech valuations are high, but not as high as B2C tech companies.

The really big private company valuations – the so called decacorns- are Consumer plays. Uber, Xoami, Snapchat, Pinterest, AirbNB, Flipkart all have valuations over $10B. The only pure Enterprise company in the $10B plus club in Palantir. And that is a very special and secretive company. Dropbox is a 50/50 play and valued at $10B.

2. Even if B2B Tech companies are overvalued, a Web 1.0 meltdown won’t happen.

This is is the most interesting thing about the unicorn attack phenomenon. I believe that unlike 2001, Enterprise tech companies are real businesses with great recurring revenue and many happy referenceable customers. In 2001, companies bought a lot of “exchange and procurement “ software from CommerceOne, but few if any every got live and the chance for repeat customers and new customers from references dried up real quick.

If Enterprise unicorns are slightly overvalued, it will not really disrupt their business. The investors, founders and employees will all make a little less money. Even the late stage investors appear to be protected by so called ratchet clauses. So in the end. if valuations are toohigh, I don’t think we have a fundamental problem.

The only risk here is that companies might be raising too much money at high valuations and not moving towards businesses with strong gross margins, acceptable customer acquisition costs and high customer retention rates. But the Enterprise SaaS model is getting well proven and I think highly valued Enterprise tech management teams understand this and are not making the types of mistakes that we saw in the first meltdown.

3. Enterprise Tech SaaS business models are inherently strong and will stand the test of time.

Enterprise SaaS is a great business model.  Contracts are typically annual commitments and renewal rates tend to be quite high. Modern Enterprise SaaS is easier to implement, time to value is quick and enterprise customers are not fickle as consumers. Ie – they don’t change their technology stacks very often. The bottom line is SaaS companies have predictable repeating revenue which makes a great business.

Secondly, gross margins tend to be good. The business is pretty simple. You charge customers  for usage. Your cost to deliver that service tends to be predictable and has high economies of scale. Most Enterprise Saas is multi tenant, so you make the software once, run it in a single or few instances. So at worst , costs scale with revenue, but more likely marginal costs drop as you add customers. That is very good!  Also, they don’t run our their data centers anymore, they run on Amazon or Azure or Salesforce. We don’t have big upfront expenses to get our business going, because we just rent compute and storage.

Customer acquisition costs can be high, but nowhere near the high costs in consumer tech ( see Fanduel or Draftkings). There are now well established models to acquire customers without blowing your bankroll and with proper investment in customer success teams, you can keep and expand customers for a decade. That makes for great margins.

Not only is this model, way better than the on premise model software model, it is much simpler that the B2C tech models that exist. Don’t get me wrong, I would love to have a advertising model like Google or Facebook, but Consumer Tech business models do not offer the type of predictable revenue and solid margins that Enterprise Tech does.

4.The trend to replace on premise software and hardware is irreversible and we are still early in the cycle in most categories

Enterprise Tech Unicorns are largely SaaS plays replacing or extending functionality provided by  on premise software vendors. The on premise software model is hopelessly broken  In those models, customers pay for a perpetual licenses and then pay about 20% per year in “maintenance fees”. Customers are also heavily incented to pre-buy software and even buy ‘all you eat” licenses. This led to a ton of shelfware, where many,many licenses go unused. The on premise software model was also famous for long expensive implementations, where time to value was measured in decades. The old model was famously inflexible. Even if you got the software implemented in was incredibly difficult to upgrade due to the massive complexity of a traditional on premise software/hardware stack. Lastly, the on premise software model is an incredibly insecure environment where tracking and applying multiple patches to many layers of the software stack lead most companies to have long periods of time where known vulnerabilities remain unpatched.

Enterprise SaaS solves almost all the problems caused by on premise software and with the possible exception of CRM software, most categories are early in adoption/replacement cycle.

5. Almost all traditional Large Enterprise players are not well positioned to compete with Enterprise SaaS.

The new Enterprise unicorns largely compete to replace solutions from Old tech – IBM, HP, Oracle, SAP, Cisco and Microsoft. With one notable exception, these companies are not well positioned to compete in SaaS organically. They have typical innovator’s dilemma problems and their size and age makes them less than nimble competitors. They also typically have little large data center, cloud experience and have difficulty attracted the employees with the skill set to compete in the new world.

I must say, Microsoft is the exception here. Their Bing and former Hotmail consumer products gave them some great cloud DNA and they have done a decent job with Office365, Azure and Dynamics. So don’t count them out.

But other than Microsoft, Old Tech is not likely to outcompete Enterprise unicorns. And in fact, their large cash hoard and slow growth rates make them ideal acquirers of Enterprise Unicorns. ( See Eloqua, Taleo, Responsys, SuccesFactors). Stock buybacks can’t buy you top line growth, so stay tuned for a more robust M&A market.

So there you have it. Enterprise Unicorns are in good shape at a macro level. Quit fretting, If you want to worry, go worry that Snapchat’s $16B valuation in a company without a reliable revenue model. Leave the Enterprise Unicorns alone. They will be just fine.

I hope Aaron Levie is wrong, but I am worried he is right!

Enterprise End Users deserve both Security/Compliance and Beautiful Simple to use Apps.


For those of it who missed it Drew Houston of Dropbox and Aaron Levie of Box got into a little war of words this past week. Competition is good for the market and I admire both these companies, but some of things Aaron Levie said worried me a little.

Here is what he said.

“I don’t think it( meaning Dropbox) will work at scale in the enterprise.  There are a lot more security, compliance and legal measures that need to be ironed out.”

We then went on to suggest that Google Apps is not enterprise ready

“Google Apps has millions of small businesses, but Microsoft is what is becoming the standard in the Fortune 500 and larger enterprises. That’s just because the DNA of the companies are just very different.”

Here is the problem I have with these statements. Aaron is suggesting that there are two kinds of enterprise companies with two different kinds of DNA. He is suggesting that companies like Microsoft and Box understand the enterprise security, compliance and legal issues and then there are consumer or user focused companies like Google and Dropbox who do not.

There is certainly ample evidence that this divide has existing for some time. Enterprise applications have long erred on the side of security/compliance. Up until recently applications were procured by the CIO’s office and enterprise software companies treated the needs of IT ahead of users. Users have hated Enterprise applications for a long time. If you have ever used SAP,  or Siebiel, you know what I am talking about. If you have ever walked up to the counter of an airline to change your flight and watched the person make about 500 keystrokes  to get you an option, you know how bad enterprise software has been. That SAP or reservation system is secure and compliant, but users hate it.

There is also evidence that application companies that focus on end users have been dismissive of enterprise needs. Up until recently, Apple was the poster child for this unfortunate approach. Steve Jobs did not like designing products for CIOs or corporate middlemen. Users loved Macs, but Microsoft won this segment handily. End user focused companies have had difficulty valuing enterprise requirements. Their engineering teams don’t like working on the boring plumbing issues that large enterprises need. They want to focus on design and end user features, but they don’t want to hear about developing an API to interface with multiple Single Sign On solutions.

Aaron Levie was saying that Box and Microsoft get the Enterprise, while Google and Dropbox do not. He is saying that the features that users love about Google and Dropbox cannot be used by enterprise.

BUT, I personally hope that enterprise end users won’t have to face this false choice between – compliant, secure apps OR user oriented apps. The users of Fortune 500 companies should get both a beautiful user experience and a secure/compliant environment.

When Aaron says companies have a certain DNA, he seems to be implying that these companies cannot change and that Enterprise users will forever be faced with a choice between applications that security and compliance professionals love and applications that end users love.  We simply can’t allow that to happen. We are in the process of a huge application shift that is seeing enterprises trade in their old on premise apps for cloud/mobile apps. We should aspire to the consumer grade usability standards as we make this shift. If we don’t, we will have failed our enterprise end users.

I think we have made some progress in this area. Traditional enterprise companies are improving their design teams. Take a look at Microsoft’s mobile Outlook app. It is arguably the best mobile email client. Great design. And despite what Aaron says Google and Dropbox have made great strides in building scalable enterprise IT feature sets. When I was at Google, we bought Postini for $625M to add granular policy control and archiving feature sets that IT needed. Facebook and LinkedIn have recently launched programs that enable enterprise users to utilize their technology for corporate use. Look for them to add these enterprise IT features. And even Apple has seen the light. The iPhone has added enterprise features and Apple created a partnership with IBM to help them with Enterprise requirements

So, I call on all cloud based B2B tech companies to ponder this question and achieve excellence in both areas.

If you have end user design DNA. then you need to acquire the enterprise skills. You need to learn about Directory integrations, Single Sign On Systems and Mobile Device Management solutions.  You need to learn about Granular Policy Control and Archiving capabilities. And you need to learn about compliance rules of different countries and industries and help your customers meet those compliance requirements. You can’t just rely on your outstanding design and tell the CIO/CSO/CCO that their requirements are secondary.

If you have a lot of enterprise DNA, you must get out from selling and servicing the IT department predominantly. Ultimately your customers are your end users, not the CIO.  Do not rely on the CIO or compliance office to select your apps and then foist them on end users. Hire some designers, especially mobile designers and value and prioritize end user simplicity and beautiful design.
Come on people. Its 2015! I am getting same day delivery at my house. I press a button on my mobile phone and a driver shows up. We have robots, drones, wearable computers and pretty soon my car is going to drive itself. Please don’t tell me we can’t build beautiful enterprise apps that are secure and compliant!