Why Hasn’t Open Source Software Disrupted Esri?
I recently tweeted that open source software is slowly but surely devouring Esri’s entire product offering after seeing a beautiful new tool from Mapbox that directly competes with Esri’s product StoryMaps.
The tweet is a riff on a common refrain among open source enthusiasts (like myself). The argument goes like this: Esri’s days are numbered because most of their core software is gradually being matched or bested by mature, free, open source alternatives.
Well, today is officially GIS Day, and I figure it’s as good an excuse to ruminate on this topic as any. After some reflection, I’ve come to the conclusion that while this line of thinking feels logically obvious (how do you out-compete free and open?), it’s flagrantly wrong. At least, for the subset of the market that employs “GIS professionals” it’s wrong. How successful Esri has been at penetrating markets where “GIS” is an annoying and slightly incomprehensible distinction from the broader challenge of data science and data engineering is a different topic for a different day.
The open source software movement, while formidable, is objectively not disrupting Esri’s core business, nor in my opinion is it threatening to any time soon. This question eats at me: how does Esri not only survive but thrive as enterprise-grade, open source software challenges an increasing share of its product offering?
Briefly, Who is Esri?
When I started working at Azavea in 2015, I had never heard of Esri. In fact, I had never heard of “GIS” in general (a term of art that enjoys incredible staying power despite being first-ballot-hall-of-fame corporate jargon). One thing I noticed pretty quickly after learning of Esri is that nearly everyone I met had a strong opinion about them. They have a near monopoly in the local, state, and federal government sector and a dominant market share in many other industries. It is also one of the most impressive privately held businesses I’ve ever encountered. As a relative newcomer to the industry, I find the degree of Esri’s success jaw-dropping given its relatively low profile. Today, they employ over four thousand people and reportedly pull in over $1 billion in annual revenue. The company proudly advertises that they operate with no debt, and I can’t find any filing with the SEC to suggest they’ve ever raised money with equity financing. How do they do it? How do they continue to profitably finance growth year after year when there are excellent free tools that do most of the jobs Esri’s products do? I have a few hypotheses…
Moat #1: Brand
One of my favorite frameworks for thinking about how great brands are built is outlined in a book called Play Bigger. The book’s hypothesis is pretty simple: in order to become a “legendary” company, you have to be the undisputed king of your category. The fastest path to becoming a category king is to define your very own category. Part of the reason this is such a powerful idea for marketers is that people start to think of the “category” as an objective, broader movement they can identify with more naturally than your distinct company. Instead of selling your product or service, you sell your category. For a terrific example of how this “category creation” strategy works in practice, I recommend the terrific memo We Don’t Sell Saddles Here sent to Slack employees by their CEO shortly before launch. In my opinion, Esri is indisputably the category king of GIS.
One way Esri preserves their status as the market leader is their maniacal focus on becoming the de facto tool for teaching GIS in higher education. I doubt much of their revenue growth comes directly from academic licenses, but I would wager that selling into universities has indirectly driven a significant portion of their growth over the long term. At this point, decades of college graduates have matriculated with a degree in one hand and hard drive full of ArcGIS projects in the other. When a GIS manager at an engineering firm makes the easy decision to renew their Esri license, that quiet moment is often the conclusion of sales cycle that started 30+ years ago when they were still an undergraduate following instructions about how to use ArcInfo in their first ever GIS course.
Conversely, the average open source project has limited brand equity. A seminal example is QGIS, the preeminent alternative to Esri’s desktop offering. Up until a few years ago, the officially sanctioned way to download QGIS for Mac was to visit the most incredibly suspicious-looking website I’ve ever seen in a professional context…
As a newcomer, I found it pretty confusing that a core distribution for one of the leading open source GIS projects in the world was hosted on a pink anime blog. Truthfully, this experience only endeared me to QGIS and its incalculably generous community (shoutout to William Kyngesburye, KyngChaos, whoever you are). I can only imagine what it would have been like to pitch QGIS to an IT department at a Fortune 500 organization knowing you had to eventually show them this site. Today, the Mac distribution lives on this much more, um, pedestrian webpage: https://download.qgis.org.
Moat #2: Corporate Procurement Practices
Even as QGIS and comparable open source projects have invested in better branding, that doesn’t mitigate the ultimate killer in the quest to replace Esri at most large organizations: misaligned incentives. The typical IT department at a large company or government agency is defensive by nature due to the asymmetrical risk of adopting new technology; you can get fired for championing a new software that winds up becoming a costly failure, but it’s tough to get fired for preserving the status quo.
One particularly insidious form of risk IT professionals worry about is the issue of vendor liability, and it introduces a tremendous amount of inertia into enterprise software procurement in general. If you’ve ever negotiated a contract with a 1,000+ employee company, you know just how desperately they want to protect their right to blame you when something goes wrong (no matter how little you deserve to be blamed). Imagine the thoughts going through the mind of an IT professional while hearing the pitch for a free and open source alternative to Esri:
- Am I going to be responsible for rolling this out without any vendor support?
- Does my department assume all liability when things go wrong?
- Now I have to write up a justification for why this is worth replacing software we’ve been using uneventfully for several decades?
Over the years, a handful of companies have formed specifically to address the problem of vendor liability for open source GIS, the most noteworthy of which was probably Boundless Geospatial before it was acquired and then unceremoniously dismantled almost immediately. Their business made a lot of intuitive sense — Boundless packaged up many of the major open source GIS projects in a tidy bundle, spruced up their branding a bit, and then offered them to customers along with support contracts and consulting engagements. Suddenly, the vendor liability problem was irrelevant — if your copy of QGIS wasn’t working, you could pick up the phone and scream at a recent college grad working at Boundless. Better yet, their business model allowed them to attract many of the most talented software engineers in the open source community who had the rare opportunity to get paid to contribute directly to the open source projects they were already hacking on during nights and weekends. But, when the dust settled, Boundless was reportedly acquired for close to $16M, or about an average week of revenue for Esri — clearly they weren’t able to capture a significant amount of market share despite offering a viable alternative to Esri with much more enticing intellectual property rights and at a more affordable price.
Moat #3: Licensing Strategy
The only way I know of to become a company whose customer base pays you a billion dollars each year without taking on outside capital is to retain your customers for a REALLY long time. This is why SaaS companies are so exciting to investors — if you execute well, you wind up with a cash printing machine that looks like compounding portfolio of annuities with no expiration date. Esri’s target markets (namely government, enterprise, and academia) are comprised some of the slowest moving entities on planet earth. They aren’t, in general, what you might call “early adopters” of new technology.
One of the ways Esri keeps customers from mutinying and replacing them with an open source alternative is through its licensing strategy. As my friend Sean Gorman recently pointed out, one of the most underrated parts of Esri’s overall business model is their prodigious use of Enterprise Licensing Agreements (ELAs). He probably knows what he’s talking about considering he once sold his company to Esri. The ELA is a stroke of genius because of its simplicity — for one flat price, you get access to practically all of Esri’s products and a bundle of support services. What this means in practice is that once an ELA has been signed, there is no incremental cost to deploy a new instance of an Esri product (either in terms of money or red tape). If you want to set up an open source service and use IT time to do it, the argument that it’s “free” and will therefore save the company money doesn’t hold water — the money’s already been spent. In some ways, it’s more expensive because there’s no customer support organization available 24/7 to help them when things inevitably go wrong.
A quick aside — many open source projects shoot themselves in the foot with well-meaning but overly restrictive licenses. In my opinion, if you’ve licensed your open source project such that commercial reuse is restricted or even outlawed entirely, you’ve probably sunk the ship before it has even left port. And that clever share-alike clause is probably a non-starter for most businesses, so that’s going to slow things down a great deal. For software, I personally think Apache 2.0 is the gold standard — broadly used, easy to understand, and truly open in that there are so few restrictions on how the software can be used, shared, and adapted.
Moat #4: Network Effects
The free and open source geospatial software community is incredible — thousands of software engineers all over the globe collaborate every day to make new tools and improve existing ones. And yet, I think it’s pretty clear that the community of people who use Esri products is bigger. The annual Free and Open Source Software for Geospatial North America conference (FOSS4G-NA), while a mouthful, is one of my favorite gatherings of the year — about 400–500 people usually show up. The international FOSS4G usually attracts around 1,000 people. Esri’s annual user conference, by contrast, brings together over 18,000 attendees, and it’s only one of a dozen significant events Esri hosts annually. It’s not a perfect comparison given how much broader the Esri UC audience is, but I think it’s roughly representative of the relative size of Esri fanatics compared to FOSS4G fanatics.
Esri doesn’t enjoy network effects in the traditional sense; each additional user of one of their products does not make the product experience better or more valuable for the rest of its users. But it does benefit from tremendous organic growth due to a form of network effects: talkative customers. The field of GIS applies to countless industries — geospatial data has proliferated and is impossible to ignore whether you’re siting a new retail store or conducting a door-to-door survey. People are happy to refer Esri products to other folks in their industry because, almost always, GIS is not the core competency of their business and therefore isn’t part of their competitive advantage. Meanwhile, in my experience, GIS professionals tend to be unusually passionate about their work. Many of them are working in Esri products all day long, so it’s only natural that when they get together that’s what they talk about.
Moat #5: Breadth of Offering & Partner Network
The final advantage worth mentioning is the sheer breadth of Esri’s products. The interface of ArcGIS Desktop alone must have hundreds of buttons and tools that all do different things — and that’s just one product! How does a product become so complex over time? Because every sales demo goes approximately the same way: a wise guy in the back raises his hand and says, “Sure, but can you do this obscure thing that isn’t useful to anyone else in this room?” At Esri, the response is usually for the sales person to smile wryly and say, “Oh, great question! In fact, we can that one of fifteen ways, let me show…”
Need to process drone imagery? Need a survey tool? Need to visualize 3D data? Need to make a floor plan? You get the idea…they’ve got a tool for just about anything you can think of relating to geospatial data. And if they don’t have that tool, one of their 1,600+ official partners either has it or will build it for you. Heck, even Esri might just do it for you themselves — an entire wing of their business is dedicated to GIS consulting. The idea that you don’t really have to think very hard to tackle any problem that comes up is extremely attractive in an enterprise IT context. It doesn’t matter much if the tool is very good or not as long as it’s manageable. Even if there were an open source alternative for everything Esri did (there is not), the fact they aren’t neatly cataloged and offered as a package deal is a huge disadvantage.
A Final Word
Whether or not you are a fan of Esri, I think the evidence is clear that the growing open source software movement has not slowed them down in the slightest. I think Esri’s ability to stave off competition from open source substitutes is an illustrative microcosm of a broader phenomenon. Chetan Puttagunta, a general partner at Benchmark Capital and investor in prominent “open core” companies like MongoDB and Elastic, put it well in a recent tweet:
I tend to agree; the very premise of most open source projects is that the world is not a zero sum game. I think open source GIS tools will likely benefit Esri more than they take market share away — they’ll actually grow the market for GIS by making it accessible to anyone with an internet connection, which in turn, will eventually grow the number of potential customers for Esri.
