If you just look at MongoDB’s high level metrics, you may miss an impressive transformation. 40% revenue growth, burning -32% on the EBITDA line — 0.36x GM-adjusted magic number, 8% rule of 40— any investor would say “ok growth, fine / not great sales efficiency”. Dig a level deeper and you see two things:
- On an operating cash flow basis, MongoDB is only burning 7% of revenue, so rule of 40 looks more like 33%, not 8%. (This is primarily due to non-cash adjustments associated with stock-based comp and deferred revenue, which most investors are comfortable backing out.)
- And the real kicker is the mix shift to MongoDB Atlas — in just 2 years, MongoDB Atlas has grown from $60M in revenue to $271M in revenue. It’s an incredible growth story that’s hidden from cursory view because of the 24% growing “other subscription” business (sales of their original self-hosted database product).
So what exactly is MongoDB Atlas?
It’s the traditional MongoDB database, but with hosting managed by MongoDB itself — what Mongo calls a “DBaaS”, a database as a service. Sounds simple, but it’s really transformational for many organizations.
I’m actually a user of Atlas — it’s where I built and run our sourcing technology platform at Brighton Park. The beauty of it for me was:
- I can stand up a database in 1 minute with zero server management, configuration, etc.
- Atlas offers a free tier, so unlike other managed databases services where I’m at minimum spending $500/month on underutilized infrastructure, Atlas has a very low entry point, and allows me to scale from there.
- Atlas manages the scaling my database as I add more data. No need to stand up more servers in a cluster as we need them.
- Atlas manages data replication, so if one of the database servers goes down, another one seamlessly steps in to take its place.
- Atlas gives me a nice UI to manage user accounts that have access to the database.
- The platform includes serverless triggers and functions, so I can stand up basic data processing tasks without needing to run a separate server just to manage scripts.
- I get the benefit of the NoSQL Mongo database, which let’s me store arbitrary information and add structure/logic to it later. No need to plan out what my data structure is going to look like, deal with database migrations, etc.
Why is this all so important?
Infrastructure management is hard. It’s hard for small organizations because there’s a high fixed cost of entry to getting an application running. And it’s hard for large organizations because at scale, you run into complex issues (query latency, sharding, server management, etc.) and reliability (data replication, failover, etc.).
But then you ask “What about the big cost of the cloud providers? Aren’t companies pulling infrastructure back in-house?”. You are not alone. Sarah Wang / Martin Casado of A16Z wrote a post about it. Zach Kanter had a great response:
Yes, companies spend a lot of money on “cloud”, but for that expense, you get so many hard problems solved for you, which frees up your resources to operate your core business, the thing(s) that actually differentiate your company. Not only that, but it gives you massive agility in easily building new applications, scaling to support your customers, etc.
And just from the perspective of basic economics, the most efficient economic outcome is one with specialists. Companies “ought to” outsource complexity to those who are better prepared to deal with it.
There will be exceptions for the largest companies that can afford to hire the best talent to manage infrastructure in-house, but for the vast majority, I would expect this shift to cloud to continue in perpetuity.
Back to MongoDB
Looking at some of the recent services MongoDB has launched in/around Atlas, you can see they’re headed toward an integrated ecosystem of products similar to what the big cloud platforms offer:
- MongoDB Atlas Search — Full-text search built on the MongoDB document database model. Compare to Elasticsearch, AWS CloudSearch, and Algolia.
- MongoDB Charts — Compare to Looker, Tableau, and other BI tools.
- MongoDB Realm — Compare to Google Firebase.
- Triggers & Serverless Functions — Compare to AWS Lambda and GCP’s Coud Functions.
Then the question becomes: How does MongoDB differentiate itself when going up against the big 3?
Just a hypothesis, but I would argue the following:
- MongoDB will obfuscate server infrastructure. I talk about the importance of this here. Other cloud platforms increasingly do this (e.g. Amazon DynamoDB, Amazon Aurora v2, Google BigQuery), so it’s not without competition, but Mongo has a head start and clearer positioning/messaging on this idea.
- MongoDB innovates on distribution with “try before you buy” pricing models and far more approachable documentation (vs AWS/GCP/Azure). This pushes them down market, but creates a “land and expand” dynamic. You can see this in their customer metrics — total # of customers spending >$100k grew only 30% in FY 21 vs revenue growth of 40%, so ACV is coming down. But net ARR retention has been north of 120% for several years and I would guess newer cohorts are showing higher expansion numbers than those prior.
- MongoDB‘s open source product gives it a large and active developer community off which it can drive growth. It mirrors something like Postgres, which is used ubiquitously and very popular. Comparing it to DynamoDB, it’s clear MongoDB has won the first part of the war on developer mindshare in NoSQL/document databases.
It’s not without risks
- The drop in sales efficiency is notable. You can see the GM-adjusted magic number came down in 2021 from 0.49x to 0.36x. This is either due to Mongo spending forward to capture the opportunity around Atlas (so will later adjust back), or it’s due to the market being so noisy / crowded with offerings similar to Mongo’s.
- Product marketing in this market is very challenging. Related to the above point on sales efficiency, in a market with so many competitors offering similar/substitute solutions, positioning and winning mindshare is challenging. Mongo has done a great job here so far, and has an advantage with its strong dev community & engagement on its open source product, but it has a long and expensive road ahead — and you can see it in the above Google Trends chart. I suspect most people still do not understand all of what Altas offers, and Mongo needs to tell this story in a massive room with a bunch of well-capitalized behemoths shouting similar things.
- The market for necessary R&D talent is hyper-competitive. Scaling infrastructure to solve global-scale problems for specific use cases has been solved well in companies like Facebook, Google, and Netflix. But building infrastructure for global-scale that needs to solve arbitrary use cases is probably even harder because of the dimensionality of the problem. Mongo has recruited some great talent to do this (e.g. Mark Porter who ran Aurora at AWS), but Mongo needs to continue to recruit the best of the best engineering talent to make this work. And all the other mega-large tech companies have a lot of resources to compete for this talent, so this will not be easy.
- Look at what happened to Elasticsearch. AWS took the open source tech offered by Elastic and started hosting it on their own platform, to compete with Elastic’s Atlas-equivalent. It probably isn’t a business killer for them, but it doesn’t help. Fortunately for Mongo, AWS has already made its bet in the NoSQL / document DB universe on DynamoDB + Mongo creatively changed its licensing in 2018 to the SSPL, which effectively prevents the cloud providers from offering MongoDB as a service.
It’s difficult to understate the scale of the opportunity MongoDB has ahead of it. AWS alone in Q1 had $50B in run rate revenue, still growing 32% y/y. GCP had $16B growing 46% y/y. Microsoft does not break out Azure, but it sits within a $60B run rate “intelligent cloud” business unit, and noted Azure grew north of 40%. Let’s just assume this market conservatively is $100B in run rate value growing 35%. Atlas made up $337M of that in Q1 = 0.34%, growing 66%. Atlas is taking share, and has a huge amount of room to run.
The road for Mongo will not be without its challenges as noted above, but what great opportunity isn’t.
Disclaimer: I am a personal shareholder of MongoDB.