Mulesoft (MULE)

Mulesoft (MULE) was a recent IPO (a week or so ago). I took a small look-see position this week (just over a 1.0% position). All the information I had was the SEC form S-1 which was the IPO filing. The company has an easy to use program that allows companies to make all their applications talk to each other and exchange information. Before they came along this had to be done by hand, or by klunky programs sold by the legacy companies in the field like IBM and Oracle, who were in a no growth status in this field, due, apparently, to customer frustration. Mulesoft grew revenue by 91% in 2015, and by 71% in 2016 (off a larger base), so they were obviously doing something right. They have acquired some major businesses as customers in spite of their tiny size. These include Coca Cola, Unilever, McDonald’s, Citrix, Salesforce, Spotify, etc. Here were some of their numbers I found appealing:

Customers have grown

2014		  590		
2015		  839		+ 42%
2016	 	 1071		+ 28%

Dollar-based retention rate has been

2014		 110%		
2015		 121%		
2016	 	 117%	

Average Subscription and Support Revenue per customer in thousands of dollars

2014		   82		
2015		  105		+ 28%
2016	 	  143		+ 36%

Revenue has grown (in millions)

2014		   57	
2015		  110		+ 91%
2016	 	  188		+ 71%

Free Cash Flow was minus $7.2 million, improved from minus $48.6 million the year before.

Adjusted Loss was $43 million, down from $62 million the year before.

Stock based compensation was relatively small.

Then yesterday, Bert Hochfeld wrote an excellent article about them:…

As I read through the article he had a lot of good things to say about the company and its operations, but concluded that the stock price was too high and didn’t allow for much future appreciation, so he couldn’t recommend it.

Some quotes (slightly simplified):

the Average Contract Value (ACV) of their customers has risen from $77,000 in 2014, to $169,000 in 2016. At this point they have 30 customers with ACV in excess of $1 million. This is typical land and expand

… its gross margin on subscription revenues is a relatively high 90%, which is close to the top of gross margins that have been achieved by software vendors who sell subscriptions to their platforms…

… Quite unusual for a company of its scale and maturity, it had a cash burn from operations of just $2.7 million. Part of the result was, to be sure, a function of stock based comp but that was just $7 million. There was an increase in Deferred Revenues of $52 million…

…Most customers are looking for solutions that provide the benefits of Mule’s Anypoint platform. They need to ensure that their applications work seamlessly with each other and that the data that is generated from multiple sources can be appropriately aggregated so that the applications can actually be used as intended - something more difficult than might be imagined… All of that it be done with the tools that have existed, but it is indeed “donkey work,” it is laborious and prone to error. It is a significant bottleneck in the deployment of applications.

Typically, users will get their IT organization to deliver one-at-a-time custom IT projects with very tightly coupled connections between apps (APIs). These are “brittle,” inflexible and can only be used for a specific application. The next application that needs to be deployed will require the same laborious process to be undertaken, just using a different set of APIs’. …This is one reason why, even at its current small scale, Mule already has some large enterprises as customers.

Okay, you get the idea. He likes what the company does and how it is doing it. But he concluded:

It is likely that MuleSoft will continue to improve its profitability, but again, not fast enough to justify the current valuation. I expect it to meet and beat expectations on a consistent basis, but it is hard to see any great reason to own the shares at the moment. Should it miss a quarter and see an imploding share price, or should the shares stagnate and the company grow into its valuation, it would be of interest. And given industry consolidation trends, the downside is not huge. And it rarely pays to bet long term against Anne Winblad. But I think I will pass on this name at this time.

The Anne Winblad he referred to is apparently a very renowned, very successful, Venture Capitalist. And she has a LOT of shares in MULE personally, as well as through her company, according to Bert.

I will keep my 1% position for now in spite of Bert’s conclusion.


I haven’t looked into the Mulesoft. I saw Bert’s article and decided I wasn’t interested just based on the title.

I was the system architect for the largest data conversion project ever undertaken by the company I used to work at. The project took several years, not all that time was due to data conversions. There were a lot of procedural changes in engineering, manufacturing, procurement, quality assurance, finance and ancillary functions, so there was lots of training materials that had to be developed and delivered, so on and so forth - It was a big project.

After we completed work on this project, the organization started looking about to see if there was a reason for it to exist or if we should just go out of business. We determined that there was still a lot of data conversion work to be done and so we hung out a shingle as the place to go for moving data from one place to another. Most of this work was one-off. That is an application (or multiple applications) were being retired while new ones were being developed or purchased to replace it (them). Once the conversion was complete, there was no persistent interface.

This stuff can get pretty hairy. One of the projects we got a call on was to convert engineering design data from one system to a later version of the same system. The vendor had provided no conversion routines and the two versions were not compatible with one another. The problem was a lot of the design data was known as “type 1” which basically meant human life was at stake if the converted data did not exactly match the source data. How do you validate that what was converted is exactly equal to what it was before? Surface geometry? Not good enough. Center of gravity? Nope. And so on and so on. All manner of physical properties might be validated, but there was still no assurance that the converted design was exactly equal to the source design. We couldn’t do the job. If we couldn’t do it, no one could. We had the best brains in the company for this kind of work. The engineering group ended up maintaining two versions of the software with some design data captive in each. I seriously doubt that Mulesoft could crack this nut, not when a raft of government regulations have to be satisfied.

In fact, even more mundane problems like parsing text data and moving it into a disciplined relational database is not so mundane in most highly regulated businesses. It’s not just a matter of moving it, it’s the validation. In fact, our validation team had twice the number of people as we had on the programming staff.

I could go on endlessly about the kinds of problems one encounters in this kind of effort. Maybe my experience is esoteric and there really are simple problems that Mulesoft helps solve, but if the problems are simple, I’m not sure what the compelling case is for the product.

I’ll pass.


I was the system architect for the largest data conversion project ever undertaken by the company I used to work at.

I’m still trying to figure out what Mulesoft’s approach is, and what is unique about it. There’s this page on data conversions:

The ETL thing reminds me of what companies like Infomaticia do.

You point out the problems with regulated industries, but it seems that most of the use cases center around sharing marketing type data within an enterprise. There a number of whitepapers and videos on Mulesoft’s site (for which you need to supply an email since they use them as sales leads I imagine). I’ll probably try to get some of the data and customer stories in the next several days to see for myself.

This sounds like a competitor for Talend. Both are in the data integration business, and both are losing money. Not sure I am ready for either one right now.

There are a number of players in the ETL (or in some cases ELT) field, although some of them have been acquired by larger companies, it’s still the same basic technology. The most common use cases are moving data from a transactional production system to a data warehouse platform or an archival medium. These tend to run on a periodic schedule and consist primarily of mapping from the transactional system to the other platform coupled with simple conditioning and transforms (i.e. possibly transforming relational structures to XML hierarchies).

Data conversions are a different ballgame. If I were so motivated, I could write a book about the kinds of problems you run into with conversions, but most of them have to do with data pollution. I don’t believe there’s any tool that could be built to handle all the wrinkles that crop up in a system that’s been in production for 20 years or so. Humans can be very creative in finding ways to make applications bend to their will irrespective of all the edits and error conditions that have been anticipated by the IT staff.

I’ll admit, I’ve not looked into Mulesoft and maybe I shouldn’t be so skeptical, But after 30 years in IT, most of it devoted to application development with associated conversions I find it hard to conceive that there’s new silver bullet that can slay this monster. None of the older silver bullets ever killed it. But, technology marches on, maybe . . . .


This sounds like a competitor for Talend.

Hi TipTree, I’m not a techie at all, but I don’t think Talend and Mulesoft compete in any way. Talend organizes big data, while Mulesoft gets applications to work together.

1 Like

Mulesoft’s profile:

MuleSoft, Inc. is engaged in providing a software platform that connects a range of technologies. The Company enables organizations to manage and secure the flow of data between all systems in the enterprise. The Company offers organizations to create agile infrastructures. The Company has developed Anypoint Platform. The Company’s customers use its Anypoint Platform to connect their applications, data, and devices into an application network where systems are pluggable instead of glued together with custom integration code. The application network enables a self-serve infrastructure through discoverable building blocks, or nodes, that can be used and reused to rapidly compose applications. With an application network built with Anypoint Platform, organizations can transform into composable enterprises.

Talend’s profile:

Talend SA is a provider of open source integration solutions for data-oriented companies and integration platform under Apache Spark. The Company’s platform enables users to connect data and applications on-premise or in the cloud, in real time or in batch, Big Data or reference data. Its commercial products are Data Fabric, Big Data Integration, Real-Time Big Data, Data Preparation, Data Integration, Cloud Integration, Data Quality, Application Integration, Master Data Management and metadata Manager. Its free products are Data Preparation and Talend Open Studio, including Open Studio for Big Data, Open Studio for Data Integration, Open Studio for Data Quality, Open Studio for ESB and Open Studio for MDM. Its product, Talend Data Fabric, is a data platform. Talend Data Fabric combines big data, data integration, cloud, reference data management, data preparation and application integration in a single platform with a common development environment and management.

They both make integration platforms that connect applications, which really means they transform and connect the data in those applications.

Tiptree, Fool One guide


Hi Tiptree, I guess that you are correct and they do overlap. Thanks for pointing that out.


This sounds like a competitor for Talend. Both are in the data integration business, and both are losing money. Not sure I am ready for either one right now.

Hi Tiptree,

I was just going through Talend’s Investor Deck. Their slide #14 specifically calls out Talend and Mulesoft as “Small Start-ups” in the Integration Software Landscape, but very specifically shows Talend as Data Integration, and Mulesoft as Application Integration.

Then Talend’s slide #15 shows Talend in Gartner’s Leaders Quadrant in Data Integration. Mulesoft, on the other hand, isn’t listed in Data Integration but is in the Leader’s Quadrant in Application Integration.

I’d conclude that they might have some tertiary level products that overlap, but that they are basically in different, non-competing fields (otherwise, why would Talend have given Mulesoft that plug in Talend’s own Investor Deck).