This blog is called Find The Thread because it started out as an overflow from Twitter, and so it’s not always obvious what connects one post to another, without the context of the social stream. That said, one recurring theme is me getting all defensive when somebody says something nasty about the business I’m in - and this post is no exception.

What set me off this time is a piece entitled Selling software to big companies is a 'baroque tribal ritual bloodletting,’ says investor. The “investor" in the title is Martin Casado, who can be presumed to know what he is talking about:

In 2007, Casado cofounded networking startup Nicira, which raised $41.82 million in venture funding before VMware gobbled it up in 2012 for $1.26 billion.

At VMware, Nicira's networking-virtualization technology became a $600 million business, before Casado left for Andreessen Horowitz just a few months ago.

Now Business Insider is of course quoting the most inflammatory statement in the title to drive traffic - which is fair enough. Also, I don’t think Martin Casado is wrong as such. What I suspect he is doing is generalising too much from his own experience.

The Nicira technology that Martin Casado developed is deep infrastructure, hidden way behind the curtain of IT. It’s Software-Defined Networking, basically virtualising networking equipment in the same way we have become used to doing with compute infrastructure. This sort of development is well suited to certain approaches to development and adoption. To quote myself:

[Developers] wanted a tool that would perform a specific, often quite technical, task for them, and assembled a band of like-minded enthusiasts to work on the project in their off-hours.

This is great when the outcome of the project is something like the Linux kernel, which is safely hidden away from users who might cut themselves on its sharp edges. Problems start to occur when the band of hackers try to build something closer to the everyday surface that people see and use.

That describes Nicira - or Martin Casado’s example in the BI article, Mesosphere - to a T. People are trying to get something done, something quite specific that is an evolution of a well-understood domain. People in that position will indeed go looking proactively for a tool to achieve that specific thing.

So far, so good. The problem is that certain types of problem are exciting and trendy to work on. Others… not so much. When is the last time you saw a popular open-source CRM system, or ITIL-compliant service desk?

That’s why makers of less trendy software have to actually, y’know, go out and sell.

What's so horrible about selling enterprise software? Well, when a big software company like Microsoft, Cisco, or even VMware wants to sell a product, they have to go through "a very baroque procurement process," Casado says.

They have to romance third-party service resellers and consultants into offering their products, who then in turn wine and dine their customers' CIOs and IT departments, taking them to expensive dinners and out to the golf course.

The whole process can take months, if not years, and it hinges just as much on salesmanship and professional contacts as it does on the products themselves. The big tech titans have gotten very good at it, to the point where it's difficult for any kind of new startup to get a seat at the table in deals of any substantial size.

"I would say all of these large companies, their strength is selling to that sales channel," Casado says. "I've always thought that was the most difficult thing for enterprise startups."

This description of the process is exaggerated (especially the wining and dining bit - the days of big deals closed at the golf course are all but gone), but not fundamentally untrue. The point is, this is the channel that exists to get tools that are necessary but perhaps not as exciting as SDN to their customers.

VMware itself did not start out as the VMware colossus that we know today. Back in 1998, they were definitely a like-minded band of hackers. Even by the time I became a user in the early 2000s, compute virtualisation was definitely A Thing, but far from a standard. Now, the technology, the market, and the company have all grown and matured, and so the approach that made sense fifteen years ago is no longer appropriate.

A cynic might even suggest that, having sold Nicira at the height of the excitement around SDN, Martin Casado is holding everyone else to an impossibly high standard because he does not recognise or admit the unique advantages which he had.

Car Analogy

I like cars, so let me try a car analogy here.

If you’re running Tesla1, you do not have an easy job by any means - but you are at least selling into a recognised market. People already buy cars, so you sell a category of product that people already understand. You still have a lot of work to do in clarifying the differences and maximising the positives, but a lot of the interest is self-propelled: people are thinking of getting a new car, hear about this Tesla company, and add it to the list of possible cars they might want to consider.

If instead you are trying to get a flying-car company off the ground2, you are facing a much tougher battle. Nobody is looking to buy a flying car; they are thinking in terms of either a traditional car, or an aircraft. This means that you have to create a new interest where before there was not even awareness of the possibility, then nurture that interest into excitement and eventually (one hopes) a sale. You can’t just sit back and wait for people to Google you; everyone is at their local car dealership, on the Tesla website, or (for the 1%) playing golf with their Gulfstream sales rep.

So what do you do? You hire sales people, you develop a sales channel, and you accept the added time and expense of all this as part of the cost of getting your product out there.

And meanwhile, ignore the snark of the Tesla1 people, who can sit back and let the Google searches roll in.

Bottom Line

I have nothing but respect for Martin Casado and his technical achievements. SDN is upending network architectures everywhere. However, many of the companies doing those SDN implementations at scale are the same network operators as before. Their procurement processes are the controlling factor, more than any supposed lag by vendors.

The enterprise sales process is the way it is for a reason. There are a number of ways to disrupt and shortcut the more cumbersome parts of that process. However, doing so successfully requires a customer base that is equally ready to participate in that process. If you are selling to customers who buy that way, good for you - but don’t try to generalise your findings too quickly to other areas.

Image by Dmitri Popov via Unsplash

  1. I have nothing against Tesla or Elon Musk - although I do object to some of the more excitable and breathless of their fans. I am just picking them as a convenient example. In fact, given how differently Elon Musk operates his various businesses, I would guess that he understands the difference I am articulating very well indeed

  2. Sorry - not sorry.