The enabling tool they built internally during development BECAME the actual product

Just like Paypal, the Hotmail founders (Sabeer Bhatia and Jack Smith) originally set out to build an entirely different product. They sought to make a simple plug-n-play database that anyone could use to store structured information and then expose it via the web. That product was called JavaSoft and sounds as if it would have been Google Base ten years before its time. Both founders held their corporate jobs while developing Javasoft. When their IT department implemented a firewall, it killed their ability to access their personal email (which they had been using to collaborate during development). They were forced to use paper and floppy’s to trade data and it crippled their development efforts. They were driven by necessity to find another way to access their personal emails remotely and it occurred to them that web sites could be accessed from anywhere. They immediately began developing a web-based email client.

The parallel here for us was that the original concept for the hardware version of JumpBox was born when Kimbro was working a consulting gig for a major hotel chain. He stepped into a situation of utter chaos with twenty consultants on the job using excel spreadsheets to track bugs and “free climbing” with no source control. He entered the situation and implemented a bug tracker, source control and a wiki for documentation to put some structure to the development effort and after the three days it took to setup this open source infrastructure, he thought “wouldn’t it be neat if a pristine instance of all this stuff could be cloned on small form factor machines and used every time you parachute into a gig and need development infrastructure?” This was of course before the concept of virtual appliances existed and he was thinking of a hardware-based appliance at the time. But he essentially arrived at the JumpBox concept in an effort to “scratch his own itch” - a common theme throughout the Founder’s At Work series.

Using the JavaSoft idea to test the waters with VC’s

Once they had the web-based email working they thought “other people probably have the same problem… we should share.” So they opened it up and began offering web email access for others and (just like Max posting his Palm Pilot security application) they had immediate visibility and rapid adoption. They gradually realized that the web-based email idea was bigger than the Javasoft product idea, but given the low barrier to entry they feared that telegraphing their intentions by distributing the business plan of the web email to VC circles could leak the concept to someone like Netscape and blow them out. They needed to tread carefully in how they approached potential investors so they used the JavaSoft story to gauge their reaction and broached the web email idea only after they had confidence in the VC.

Just-in-time scaling

The Hotmail story was very similar to the “Hot or Not” story in that they experienced such explosive growth early on that their growth nearly killed them. They managed to stay just one step ahead of being overwhelmed with traffic by relentlessly adding servers. This is a problem everyone wants to have - growing so fast that you can’t keep up with the new traffic - but at the same time server outages not only prevent new growth but destroy your credibility with existing users so it’s an all-or-nothing game. Services will typically throttle new signups in that situation to preserve the quality of the experience for existing users rather than risk blowing it but Hotmail managed to keep performance satisfactory while keeping open signups throughout.

We have the luxury now given the nature of our offering that scaling is much easier relatively than it was for other companies. First, our products currently run locally on user hardware so ours is not an SaaS offering in that we don’t do the hosting. Secondly, because we’re now beginning to move all production apps to a virtualized environment, it should be much easier to deploy to faster hardware as necessary. We do face issues of bandwidth in making these large files available for download from our servers but we’re mitigating that with spill-over hosting on Amazon’s S3 (and we’ve also moved our dedicated servers to a higher-bandwidth data center).

The viral component

Hotmail is probably the classic example of viral marketing. Every email that was sent via their system had a tagline appended to the end of the message that said “This message sent via Hotmail. Get free email on Hotmail.com.” The users of Hotmail were inadvertently advertising the product to each of their friends with each message they sent and given that the signup process was frictionless and free, it immediately went viral. Viral advertising is trait of most of the companies that experienced logarithmic growth in FAW. Our offering fortunately takes advantage of this aspect now as well and we’re seeing some immediate results.

An unassailable lead

With the same premise of the Mythical Man Month, even with Microsoft having 16,000 engineers and Hotmail only 60, they had an six-month head start on the product, it’s scalability and a loyal user base that was promoting it virally each day. Their lead proved to be insurmountable for MS and on New Year’s Eve 1997, Microsoft acquired Hotmail for $400MM. The story of how the negotiations went down reminded me of a negotiating gambit called “deference to higher authority” and how silence can be the strongest bargaining tactic you have.

The lead we have with the JumpBox technology is by far insurmountable at this point but with the viral component and our commitment to provide our customers massive value, we strive to build a loyal following before other competitors enter the virtual appliance space.

This is the first in a 32-part series of posts on each of the vignettes from the book Founders At Work.

The initial direction is never the final direction

Paypal is an interesting story on many levels, probably the most striking of which is the fact that they originally set out to build an entirely different product. Founders Max Levchin and Peter Thiel originally intended to build a wallet application for the Palm Pilot that would allow its users to securely beam money back and forth. Max had a background in security and had written a piece of software for his Palm Pilot that allowed him to ditch the many challenge/response security keycards that he had to carry around school by emulating the algorithms. He reverse engineered some ten different algorithms and then posted the application to a site to share with others. He was “scratching his own itch” and sharing the solution with others - a common trend throughout the FAW stories. It was apparently a prevalent enough problem for fellow nerds and he received a ton of downloads and offers from people to pay him to add certain features. He decided there was a business there somewhere and it began a winding road that ultimately led to building a company that was acquired for $1.5BN.

The Palm Pilot was a relatively wimpy device resource-wise and Max had to make things lean and get creative with the interface to make the experience responsive enough for the user. His thinking was that mobile devices would become pervasive in big businesses and since security is important to enterprises, the devices would need to be highly-secured and he could sell the libraries to make that happen. It was a valid hypothesis but the demand for such secure devices never materialized. Rather than stick with an original flawed plan, they listened to the market and went back to the drawing board to see what else they could do with their investment in building security for PDA’s.

What immediately resonated for me on this story was the similarity with how JumpBox got started. At first our original plan was to be able to offer what we had been calling “On-site software as a service.” We knew that SaaS represented huge value to customers in that they could eliminate all the overhead of managing their own hardware and application updates. We also knew that there were disadvantages with SaaS that outweighed the advantages for some- privacy concerns with storing sensitive data on a third-party system and connectivity concerns (what happens when your T1 is down and your entire customer database is stored remotely?).

Our plan was to build a vessel that would allow us to package software applications on small form factor hardware and deploy them in such a way that they offered all the benefits of SaaS like automatic backups, monitoring, updates and required no maintenance from the users. So we originally set out to build this custom hardware device (check out the original header graphic) but quickly learned that
a) the hardware is a tough game to get into - the minimum order size to be able to turn a profit was prohibitively large (like 10k units)
b) our hypothesis that there were a ton of people fleeing SaaS from the privacy concerns from all the identity theft breaches was not accurate- the demand fo
c) appliance sprawl in the server room is an issue for companies and they’re actually rebelling against adding new disparate hardware because they want a unified IT infrastructure

So we had to go back to the drawing board on how we would carry out our plan.

Max and Peter were looking for a different application asking the question “what else would demand bulletproof security on a handheld device?” They considered storing credit card and other sensitive info but realized it still wasn’t practical for most people- there are too many justifications for carrying your credit card with you. They eventually arrived at the idea the ability to store money itself on the Palm and beam it around via infrared. That turned out to be one of those concepts that sounded neat in theory to techies but was again impractical- it required that both people have a device with the application installed. This was happening during the dot-com exuberance though when 90% of the ideas were more ridiculous and they were able to raise $4.5 on this story of beaming money. They actually used a Palm Pilot to receive that $4.5MM over the lunch table with their VC’s.

Listen to what your customers are asking for

They got their funding, built the download-able Palm application and posted a demo of it on the site. What they found though was that people were actually trying to use the website to send money instead of downloading the Palm app. Hundreds of people were flocking from eBay mistakenly thinking that this was a web-based method of making a financial transaction with a stranger. With only a handful of people actually using the Palm app they had an epiphany that they were building the wrong thing. They went heads-down to build what the eBay customers were asking for and a year later had essentially the service that is available now.

Again the resonance here with JumpBox was clear - our original concept was not actually demanded. The security and privacy concerns we had forseen weren’t as painful for companies as we expected. And IT admins didn’t want another piece of hardware in their data center to manage. About this time we were hearing more about the trend towards virtualization in the data center. Pacific Gas & Electric had just announced a 50% rebate to ISP’s who were willing to move to a virtualized infrastructure. The epiphany for us came when we realized virtualizing the offering could solve both the objection of the IT admin to having extra hardware as well as the problems we were facing with having to manage inventory and the high sunk capital costs of purchasing the hardware up front. We immediately changed tack and had a virtual version of the appliance software put together in a week using VMware.

How they survived the stiff competition

An interesting aspect of the Paypal story was how they survived the onslaught of competitors. As soon as the space for online payments was validated, immediately other competitors emerged. They only reason Paypal won in the end was fraud- they figured out how to deal with it while the competitors didn’t. In 2000 Paypal was losing $10MM per month to fraud- it was still only a fraction of their revenues and business was growing fast enough that it wasn’t immediately prohibitive but it was clear that the trajectory of it would eventually kill the company. They realized at that point that their business wasn’t so much facilitating online transactions anymore as it was preventing fraud and Max set out to write software to detect the transactions that were prime candidates for being fraudulent.

Tracing a chain of fraudulent transactions was a tedious task and given their tool set at the time involved printing out reams of paper of transactions across accounts and having the investigators manually sift through them. They were able to build a tool that allowed investigators to slice through this process and visualize the flow of money across many accounts. Their tool helped the investigators take a significant chunk out of the problem- enough that they survived while their competitors succumbed to the chargebacks.

This was another theme that was common across stories in the book - companies that cling to their original value proposition gradually get washed away as the space becomes commoditized. The ones that shine are the ones that “skate to where the puck is going to be” as Clayton Christensen says. Paypal recognized the severity of the fraud issue and rightfully pinned the future of their business on successfully solving that problem. Other competitors were busy focused on growth ignoring the fraud issue and ultimately had to close shop from the burden of chargebacks. For us the virtual appliances industry is nascent and still relatively wide open. We have a decent head start on developing the technology but that will evaporate as more competitors enter the space. The key for us will be in keeping our ear to the rail and finding what those pivotal challenges are that weed out the competition and place our focus on solving those well. Distribution is everything and once we transition to sales, creating the VAR network for distribution, building out the library of appliances and tackling the hosting opportunity become paramount. Our equivalent of the Paypal fraud could come in different forms and it will be up to us to keep our eyes wide open to recognize it.

The viral component

It’s interesting to note that many of the successful companies in FAW had a viral component to them. Paypal was a clear example of this in that, if you sent money to someone they would get an email and need to create a Paypal account in order to retrieve it. Viral wins because it’s promotion that comes at no expense to the company. Once it’s out there it propels itself and relies upon the networks of its users to spread. With the latest release candidate of JumpBox we’ve added a viral component to our offering. The unregistered version of our software applications displays a navigation bar across the top that has a link back to the JumpBox site. Anybody is welcome to download a pre-packaged blog or wiki application for free and run it on their own system forever if they like, but in doing so they will simultaneously advertising JumpBox to all their viewers. We determined that this was an important feature to have and helps offload some of the promotional responsibility from us to our non-paying users. We’ve already seen good response in traffic and there are a handful of new JumpBox instances that pop up each day.

The Paypal story had an incredibly successful outcome resulting in a $1.5BN acquisition from the company that was sending them the most users- eBay. Their success can be attributed to the adaptability of their founders and their recognition of the importance of executing on the above four elements.