Splunk, another overhyped software company

Splunk is yet another totally overhyped Silicon Valley software company.

In their S-1 document, they claim they make software that deals with "machine data." The phrase "machine data" is repeated over and over again in their S-1, like a kind of mantra.

SPLK subdivides data in their S-1 as follows:

The growth of digital information estimated by IDC may not be indicative of the growth in machine data, which is our primary market and only one of the three types of data included in the estimated growth figure by IDC.
• Business application data is the digital information used by organizations to conduct their daily operations, such as payroll, supply chain and financial data. Most business applications rely on traditional relational database technology and software that have pre-defined data structures, or schema for organizing, storing, accessing and reporting on structured data.
• Human-generated content is the digital information derived from human-to-human interactions, including email communications, spreadsheets and documents, and in recent years, mobile text messages, video, photos, recorded audio and social media messaging. Human-generated content typically comes in the form of unstructured data, which means that it is not optimized for storage in a relational database.
• Machine data is produced by nearly every software application and electronic device. The applications, servers, network devices, desktop and laptop computers, mobile devices and various other systems that organizations have deployed to support their operations are continuously generating information relating to their status and activities. Machine data can be found in a variety of formats such as application log files, call detail records, clickstream data associated with user web interactions, data files, system configuration files, alerts and tickets. It is generated by both machine-to-machine as well as human-to-machine interactions. Outside of an organization's traditional IT infrastructure, every processor-based system, including HVAC controllers, smart electrical meters, GPS devices and radio-frequency identification tags, and many consumer-oriented systems, such as mobile devices, automobiles and medical devices that contain embedded electronic devices, are also continuously generating machine data. Machine data can be structured or unstructured.


Elsewhere in its S-1, SPLK describes its focus as follows:

While our software may be used to collect and index all types of data, including business application data and human-generated content, our primary market opportunity is focused on enabling our users to harness the value of their machine data.

So, we are told that SPLK’s software is designed primarily to deal with machine data and that machine data is distinct from both human-generated data and business application data.

Now, a case can certainly be made that there is a difference between human-generated content and the other two types of data. It would be quite a feat, for example, if software could read this message and tell just from what I have written so far whether I think SPLK is a buy or a sell. But SPLK does not claim to focus on human-generated data.

This leaves us with the question of whether machine data is really all that different from business application data? In my opinion, it is not. And that is where SPLK’s business plan breaks down. In the modern world of computing, almost all data is machine data. When you press a button to purchase a book on Amazon, is it “business application data” that is generated or “machine data?” Obviously, some kind of record is generated, passed around through the ether, and eventually stored in some persistent store. But, is this essentially any different from what happened a couple of years ago?

The other thing you can say about machine data is that, since it was generated by machines, it is relatively easy for machines to deal with. That is, unlike human-generated content, machine data generally has a format that is easy for other machines to process and act on. A clickstream record has a particular, well-defined format. It may not be a relational row, but that doesn't mean it can't be broken down and stored in a relational database. In other words, what SPLK and the IDC say about machine data -- that it “can be structured or unstructured” -- is nonsense. If it were unstructured, it wouldn’t be machine data. As the S-1 says, “machine data can be found in a variety of formats.” This is quite different from saying that it is not formatted or structured at all.

So, what new functionality is SPLK actually bringing to the table? In my humble opinion, nothing much. In their S-1, they claim to have several patents. But my guess is that these patents are going to be, like so many software patents, largely unenforceable in practice. The lawyers will get rich fighting things out, but in the end other software companies will find a way to do what SPLK is doing (if they are not doing it already).

“Machine data" is the new buzzword. It's just like the other buzzword that we read about so often these days, “big data,” or older buzzwords like "web services" or "software as a service" (SAAS) or "cloud computing." These are brilliant marketing memes, but the actual value of the software behind them is quite a different question. Hasn’t the last decade taught us anything about how to value software companies? Didn’t the internet bubble teach us not to drink the Kool-Aid?

So, right now SPLK’s revenues are growing quite nicely. In my opinion, that is because they have managed to create a great marketing meme that has gone viral. But, eventually, other companies will catch on and replicate this meme and market their own, similar software products under it. When that happens, SPLK will fade fast. Right now SPLK’s p/e ratio is up in the hundreds. Bubble? You bet.