Own the Exit Strategy

At one point earlier in my startup career when I was VP of Marketing at a 3D graphics startup, I was advised by a well-known Silicon Valley VC that I wasn’t to worry about the exit strategy. Our job was to build value and the exit would come. Basically, go away and let the smart folks (VCs) determine the exit strategy. For many years, I generally accepted this philosophy but 20 years later when the world of startups and venture capital is in the midst of major upheaval, I believe entrepreneurs need to own the exit strategy as a fundamental part of the business model from day one. The product decisions you make and the organization you build must reflect your exit strategy. More critically, it determines your capital requirements over the life of the company and quite possibly the investors you target.

The accepted world view is that there are two basic exit options for enterprise software startups that have taken capital from institutional VCs or angel investors—IPO and acquisition. Given the IPO option is off the table for the vast majority of startups, the choice comes down to the strategy for an M&A exit. One approach which I label the Oracle “Mini-Me” model under which you develop a feature-rich product and all the sales, marketing, tech support and F&A infrastructure that you would find in a big company. My general assessment is that this takes 5-7 years to reach exit, $20-40 million dollars of capital and, if all goes extremely well, an exit at 8-10X capital in. It fits nicely with the timeline and capital available to large VC firms. Many large acquirers have the capacity to process an acquisition of this type and many actually prefer to have risk-free market validation through several years of growing customer revenues.

The other approach is what I refer to as “portfolio completion” (I’m certain there are sexier labels but nothing comes to me at the moment). Under this model, a target acquirer is looking for new products or technology that can fill the gaps in their product line to give them a more complete solution for their customers. They have established distribution channels (direct sales or otherwise) and a customer base that can be readily leveraged. The need for extensive market validation via proven revenue streams over several years is less critical since the acquirer has already determined the need for the product and/or technology and actually prefers a much leaner organization without the other functions that duplicate what they already have. The startup objective under this model is capital-efficient delivery of a basic product with solid market validation. These types of exits can occur in 2-3 years with capital investments well under $10 million and possibly under $5 million. The multiples are probably a bit lower in the 5-7X range but the IRR can actually be higher because of the shorter term. Not all VCs can operate under this model but smaller VC firms and angel groups can generally handle this approach.

Large VC firms will push startups towards the Oracle “Mini-Me” model because of the partnership structure (portfolio company to partner ratio), fund size and management fees which enable these firms to deal with much longer horizons. Smaller VC firms and angels (groups or individuals) may be more appropriate for the entrepreneur who believes the “portfolio completion” model is a better fit for his or her business model. A spin on the “rich versus king” issue also comes in to play here since the “Mini-Me” model generally has at least one CEO shift over the life of the startup before exit and it’s quite unlikely the founding team will be in charge or even have a role in the company after 5-7 years.

It’s not an obvious nor simple choice. I’m biased towards the “portfolio completion” model since I’m not convinced that business model in an early-stage startup should be driven too excessively by external factors like the minimum size checks a VC fund can write. In reality, it’s probably a hybrid model that demands agility when the context changes. At the outset, I believe startups should employ the “portfolio completion” model since it reflects a customer-driven product focus as the startup strives to find a market fit. Once the market demand has been validated, the decision to seek an exit or swing for fences with the Oracle “Mini-Me” model can be made.

In the final analysis, the exit strategy will be driven by the product, market sector, investor demands, capital requirements and external conditions but my point is that it must be acknowledged in the startup business model from day one. Own the exit strategy.

Angels vs. VCs

An angel investor in Knowledge Reef Systems sent me a copy of Basil Peters presentation (no link at this time) to the Angel Capital Association Summit in San Francisco this month. Absolutely spot on with one minor complaint (see a couple paragraphs below). Basil’s key points:

  • Target an early M&A exit (get out in 2-3 years)
  • Build a product that a larger company can grow as part of their portfolio (leverage their expensive sales organization)
  • Aim for the sweet spot of the M&A market ($15-30 million)
  • Run lean (a cliché but true)
  • Seriously consider angel investors even if your capital requirements are in the $2-5 million range

This is stunningly contrary to conventional institutional VC wisdom that demands entrepreneurs build Oracle “Mini-Me” clones. Why do they do so? My assessment is they simply have too much capital that must be put to work, they need to limit the number of bets they make and management fees tend extend the exit horizon beyond what is reasonable for angel investors. VCs have no choice but to swing for the fences on every investment even if a more rational strategy of base hits wins more games. Is there a relationship between this “home run” approach and the dismal VC returns of the last decade? The answer is far too complicated for my pay grade…;-).

As Basil points out, angel investors may be more aligned with founders than VCs are. It certainly gives me pause for thought as I get ready to launch my fund raising effort for veloGraf Systems. The VC path is the default but my strategy for building the company is probably more aligned with angel investors. I enjoyed working with angel investors in my last company and if I can meet my capital targets with angel investors this time around, it certainly beats being forced into the wrong business strategy simply because of excess capital in VC funds.

My exception to Basil’s presentation was his discussion of the “weekender” startup. I know I’m an old war horse when it comes to creating value in startups but I remain convinced that it takes more than a couple guys in a garage over a weekend to build a meaningful venture. If it can be done by these, why won’t another team (or ten teams?) be able to do the same thing the following weekend? What is the barrier to entry to a startup product/service that can be developed and deployed in a few days, weeks or a couple of months? I can’t help but believe that this type of investing is like buying lottery tickets. Maybe you can win the jackpot but the odds are much higher that you will spend a small fortune on $10,000 investments and never hit the jackpot. A couple small wins perhaps but your overall IRR will be negative.

I am in 100% agreement with Basil that the cost of building software solutions is far less than what it was 10-20 years ago but it simply isn’t $10,000. This “weekender” myth (also called startup “boot camps”) drives me crazy because it makes angel investment look more like a weekend in Las Vegas than an effort at reasonable due diligence on the capabilities of the founding team, the market opportunity and differentiated technology with a sustainable advantage. Yes, angel investing is risky but investing in this type of venture seems to heighten the risk.

Oracle “Mini-Me” vs. “Cool-Tech” Startups

Why is it that so many enterprise software startups continue to be structured as “mini-me” clones of Oracle (or pick your favorite large, established software vendor)? In a world of agile development processes, alternative distribution channels and customers accustomed to employing open source software that demands “self-help” evaluations, you would think that new business models would emerge. Instead, we see these very small companies, most of who have yet to reach cash flow break-even, adopt an extremely costly customer-facing organization (direct sales and support) paralleled by a rigid, long product development and release cycle. This seems particularly true in the data warehouse/business intelligence sector that our company, veloGraf Systems, is targeting.

My observation is that these companies are caught up in a vicious cycle based on the unproven premise that success in this space demands $20-50 million in capital. Assuming that is true, then only VCs who can (and must just because of the mechanics of large funds) write large checks invest in this sector. Having committed big sums and believing they are competing with large, established enterprise software vendors, these VCs insist on bringing in executives whose first reaction to the apparent chaos in many early-stage companies is to build a structure that reflects the large organization from which they have been recruited. A self-fulfilling prophesy. The burn rate takes off, the organization becomes less nimble and even more funding is required to build competitive sales and engineering organizations. Maybe I’m just showing my age and having a flashback to a database startup I ran using this model during the Internet bubble era but I found several contemporary exemplars of this approach while undertaking a quick competitive analysis exercise recently.

Of course, at the other extreme, there are the open source enterprise software startups who practice the “build-it-and-they-shall-come” model on the presumption that if you simply make enough noise tweeting about your cool technology and preaching to the converted cognoscenti, the sales will materialize. While usually practitioners of agile product development, startups operating under this scenario frequently lack a disciplined customer development process so critical to converting technology interest to product revenues.

Somewhere between the Oracle “mini-me” model and the freewheeling “cool-tech” approach is a business model that acknowledges the need for structure and experience but avoids a heavy-handed approach of jamming an inflexible large-company organization into a small startup. As I noted in earlier posts, I believe Steve Blank‘s agile customer development process and Eric Ries‘ lean startup principles can be applied to an enterprise software startup. With the right discipline exercised by the leadership team and investors, capital requirements can easily be half of Oracle “mini-me” model and quite likely even less. Even more interesting is that a lean startup with a differentiated product that is cash flow positive without the burden of an expensive, field-based direct sales organization potentially has a higher value to an acquirer looking to extend its product portfolio.

Biz Model vs. Biz Plan

Steve Blank defines a startup as an organization formed to search for a repeatable and scalable business model. He and Eric Ries have done yeoman’s duty in trying to move budding entreprenueurs away from the linear thinking used by most VCs to evaluate investment opportunities and startup processes that consume investor resources. Unfortunately, it takes wise VCs and smart entrepreneurs to accept iteration and “pivoting” as a natural element of the early-stage startup operation. I can envision a situation that is simply repeated iteration (spinning) that ultimately burns through the investment having collected a ton of data yet made no measurable progress towards the real end game–a business model on which to execute. The other extreme is simply being too cautious in iterating and making pivot decisions. Nonetheless, my thanks to Steve and Eric for articulating a model that really forces us to re-think the startup process. I will confess to having been in the startup game for almost as long as Steve, albeit with a later start, yet I still value the counsel of smart folks like them. It has certainly helped as I have gone though this exercise to define how we will organize and operate veloGraf Systems.

Which brings me to the title of this post–biz models vs. biz plans. You could argue that it’s simply a re-labeling exercise since the core concepts are addressed but just the term model is implicitly more dynamic, immediate and inter-connected. If I change one process in real-time, the outcome should be measurable elsewhere in the model. Plans seem inherently rigid, identify activities than can always be deferred, are never used after being written and absolutely not read by VCs (associates have this nasty task).

I began this series of posts exploring ideas for organizing and operating an early-stage enterprise software startup in a fashion that provides a rational alternative to the conventional wisdom that enterprise software is too difficult and takes boatloads of VC investment. I will confess this was a typical cursory Web investigation conducted over a few days but I have generally concluded that very solid guiding principles are offered by the following:

  • Steve Blank and his work in the area of closely coupling agile product development and customer development to find the right business model and product (www.steveblank.com)
  • Eric Ries’s thoughts on lean startups and minimum viable product (www.startuplessonslearned.com)
  • Alex Osterwalder’s efforts in business model generation (www.alexosterwalder.com)

Steve and Eric offer operating principles while Alex’s insights are really in understanding and organizing the basic elements of business operation. As I develop the veloGraf business model over the next few months, I plan to rely on these tools to direct my efforts.

Agile {Development | Selling | Funding}

In my last post, I was bemoaning the challenges of undertaking  an enterprise software startup, especially when you look at the distribution (go-to-market) difficulties and the heavy lifting usually required on the development side. Many VCs look at these investment opportunities and see a double digit (millions) investment over 3-5 rounds. The exemplar that triggered me to consider alternative enterprise software business models was GreenPlum, a startup in our market sector that has taken at least $45 million from VCs since 2005. I’m certain they have a solid product, good team and market traction in the data warehouse / business intelligence space but the bar for a successful exit is so high that it must be depressing for the founders.  I will say that I know bupkes about the company except what’s on their web site so they may very well be on the way to a $500 million exit. Cool for them if that’s on the horizon.

For veloGraf Systems, our new company, this business model is a non-starter so I crawled out of my self-imposed “build-the-prototype” hibernation and started exploring how other startups might be dealing with this challenge. Not surprisingly, there is a ton of stuff out there. Eric Ries Lessons Learned blog offer some interesting insights on “lean startups” and the Venture Hacks web site is worth visiting. However, what struck me was the very narrow focus on Internet startups. Not to diminish the value of focusing on capital efficiency, it’s a rather obvious that an Internet startup can be “lean”. This was reinforced in Babak Nivi’s interview with Eric Ries where they discussed what constitutes a minimum viable product. Eric was whinging about taking two weeks to build the final product (his debate avatars) instead of just a landing page to test interest. Jeez, Louise! A landing page as an MVP? And two freakin’ weeks for the final product? No wonder there’s a huge disconnect when VCs talk about software startups.

Nonetheless, the “lean” mantra has value and Eric’s definition of  MVP is predicated on an agile development model. For enterprise software startups, I think the best “lean” lessons may come fron Steve Blank’s customer development model as outlined in his book and the preso he delivered the Startup Lessons Learned conference. The principles can be applied to any type of startup and at least Steve understands from personal experience that a minimum viable product for many software companies may actually require a small team of really good software engineers many months to deliver just the core MVP.

I don’t have the definitive strategy for how we build veloGraf Systems. I am encouraged by all the smart people in this space willing to share their ideas about resource efficiency. The end solution really extends the agile development model to “agile selling” and “agile funding”. BTW, the latter is not the brain-dead, micro-managing tranching methods used by some early-stage VCs but really looking at what it takes to iterate over the next few development and selling cycles. This is actually harder than it may seem because of the conventional VC process, legal costs for closing a round and a willingness of investors to engage in the iterations inherent to the agile model.

I’m optimistic that we can leverage many of these “lean” and “agile” techniques to provide an operating model for developing the veloGraf product and raising money. As Steve Blank makes clear, we still must find the business model that make the company viable. I’ll kick that can down the road for a couple of months as we wrap up the prototype.

Is Enterprise Software Really Back?

I need to confess up front—I’m a sucker for enterprise software startups (launching and running them, that is). Why else would I be in the middle of the bootstrap phase for yet another enterprise software startup—veloGraf Systems—in a most unlikely location, Santa Fe, New Mexico?

You can imagine my surprise when Trevor Loy (@trevorloy) retweeted Kevin Spain’s (@kevinspain) link (http://tcrn.ch/axDwZf) to TechCrunch’s interview with Marc Andreessen’s where he stated his new firm is “very, very interested and active right now in enterprise software”. Can it be true? I share Marc’s view that most professional investors consider enterprise software investment opportunities non-starters (“dead” is the term he used). Is Marc really on to something or does he too suffer from an addiction to startups where serious innovation and solid engineering are table stakes before you even take on the even more daunting challenge of taking an enterprise-class software product to market? I’m thinking of LoudCloud/OpsWare.

Enterprise software startups are not for entrepreneurs who expect to cobble together a web 2.0 service during a short boot camp or investors interested only in consumer-driven home runs. However, for entrepreneurs and investors who don’t mind a bit of heavy lifting as the price for success, a few upside considerations …

  • As Marc points out, startups like these are often founded by “seasoned” (yes, that sometimes means “older”) executives who are way up the learning curve and less likely to make simple execution mistakes. They almost always understand the markets they are entering which can frequently be the most painful and expensive lesson to learn.
  • As with all startups, you want to swing for the fences every time but enterprise software leaves you opportunities for base hits because there is often a core product that meets “portfolio completion” needs of larger vendors who simply can’t afford to internally invest in nascent market opportunities.
  • In today’s software development environment, there are stunning amounts of really exceptional open source technology letting the startup focus on the vital, differentiating innovation and difficult engineering efforts required to build reliable, scalable products. This provides excellent leverage for a development team and can reduce the amount of investment required to get an initial product to market.
  • Contrary to what many believe, IT organizations do continue to adopt evolutionary technology. If an innovative product can meet a business need and the startup understands the rules of engagement for delivering a product to this market, I remain convinced there remains “white space” in the IT ecosystem. This does NOT mean wholesale displacement of the entrenched vendors like Oracle, IBM or Microsoft no matter how “breakthrough” your product may be. An unfortunate blind spot of many new entrepreneurs to this space is the unwavering belief that their technology is so innovative that CIOs will willingly discarded millions of dollars of development investment and untold measures of staff knowledge to adopt a new technology. Ain’t gonna happen. Think niche. Think evolution. Think playing nicely with others.

I’m certain there are many more “rationalizations” for investing in an enterprise software venture that I can add to this list since I need to hone the story much more before I hit the fund-raising circuit this summer. But on to the challenges…

  • The single biggest obstacle to succeeding in the enterprise software market is simply the difficulty of getting a product through the tortuous path to closing deals. While I do believe that we have tools today that can accelerate the sales cycle while reducing it’s cost, I’m still bedeviled by finding the ideal model to reduce the cost of distribution. Building a sales and marketing organization can be extraordinarily expensive and begins to rival product development costs at a certain point. When that happens, it’s easy to see why another company in our general market space (data warehousing and business intelligence) has taken at least $45 million from VCs since 2005. What is harder to see is how there can be a reasonable outcome for all stakeholders with that level of investment. The “base hit” option is clearly off the table.
  • The distribution cost challenge is closely followed by the risk of outsized product development costs which can easily spiral out of control if the product vision is too grand. Or maybe not carefully managed since I believe a “grand vision” is important to early-stage companies. Enterprise software suffers from those damned “customer prospects” each of whom has a set of must-have features. It’s very hard to manage feature creep in the face of a sales team prospecting for leads and investors pounding you for a pipeline.
  • Which really brings us to the overarching challenge–is it possible to build an successful enterprise software company for a relatively manageable investment that can offer VCs a good return even if it’s a solid base hit (let’s say a double) and exceptional returns if it’s a home run? The amount I have in mind is definitely a three-beer discussion but I think the number could be in single-digit millions of dollars). Yes, I know these numbers don’t work for firms that can’t write modest checks and have such unmanageable portfolio companies that partner leverage is limited to a handful of companies for each partner.

So what is the optimum model for an enterprise software startup that makes this possible? Probably not too different than it was 15 years ago but it is enhanced by leveraging the major leaps forward in software development technology and a communications environment for marketing and selling.

  • Think tactical (but don’t give up the grand vision). Agile development techniques should also be used for “agile selling” in the early days of the product. There is an incessant pressure to have the “complete product” in terms of all competitive features and support from day one. I may be overstating the flexibility of contemporary IT shops but I believe targeted products meeting specific customer needs can be introduced. Any IT team using open source software knows the product is never done.
  • The investment corollary is to take the right amount of investment to get to each stage. Yes, more is good but it often means building unnecessary features or spending money too soon on sales and marketing. And, yes, many VCs prefer to make fewer decisions and write larger checks. They may be the wrong investors.

Marc never mentioned whether his interest in enterprise software extends to improving the leverage. He may very well prefer the large-investment model depending on his investment strategy and fund size. I don’t. Been there, done that. If you have it, you spend it and not always wisely.

I think I have just outlined the manifesto for how we will run veloGraf Systems manifesto. Not my intention when I started this post and it needs further elaboration but it does articulate what I believe are the the basic principles for operating an enterprise software startup in 2010. Marc was right—enterprise software is back!