What this covers
The episode is a business breakdown of Datadog, with Peter doing most of the explaining alongside Jesse. It starts before Datadog existed: teams wrote their own scripts or used open source tools, while vendors such as Splunk for logs and New Relic and Dynatrace for APM each covered one kind of monitoring, so an outage meant toggling between products. Peter connects this to the split between developers and operations before DevOps, and argues that Datadog's founders, who had lived the problem as customers, turned that organizational friction into a product with a single data model for logs, traces, and infrastructure metrics.
The middle of the conversation covers the mechanics of growth. Peter breaks the 50 to 60 percent annual growth into three levers: new customers, expansion within existing accounts, and new product launches, from infrastructure in 2012 to APM in 2017 and logs in 2018 and 2019. He also covers ease of activation through one agent and more than 450 integrations, and the "free" open source option, whose cost shows up in engineering headcount. A contested point is pricing: Datadog stays a la carte while some rivals move to a single usage fee.
The last stretch looks outward. Peter cites Gartner's roughly $44 billion market estimate for 2024 and suggests observability could extend to manufacturing or sales funnels. He sees Datadog's push into security meeting CrowdStrike, which bought Humio, coming from the other side. The main risk he names is a change in how software is delivered, such as no-code, low-code, or Web3, and he is open about how speculative that is.
Datadog's rapid, profitable growth comes from a land-and-expand motion fueled by a fast product development cadence, and its consolidated observability platform beats fragmented point tools, though a shift away from traditional internet-delivered applications is the main long-term risk.
- Net dollar retention above 130% means existing customers expand spend, driving most revenue growth
- Datadog spends more on R&D than on sales and marketing, using product velocity to make sales easier
- Bundling infrastructure, APM, and logs into one data model and UI was the insight that separated Datadog from single-purpose competitors
Land-and-expand levers and R&D reinvestment form a flywheel that drives Datadog's growth.
- The distinctive element of Datadog's approach is its very rapid product development cadence combined with the land-and-expand motion, and together these explain why it grows 50 to 60 percent a year, faster than any competitor.
“I think that what's unique with Datadog's approach is the really rapid product development cadence.”
- Datadog's R&D spend, now 30 percent of revenue and growing year over year, is what allows it to keep adding products and improving features, which is why it can afford to keep the rapid release cadence.
“Whereas if that's then allowed them to invest more in their research and development, their R&D line, they made up 30% of revenue last quarter.”
- Datadog's go-to-market is efficient because it reaches developers directly through conferences and developer communication, so that product improvement makes sales and marketing easier, letting the company keep growing while spending less on sales.
“They're great at reaching out to developers, they attend a lot of conferences, they have a lot of direct contact or communication with developers.”
- Datadog, like Twilio, uses a developer-led try-before-you-buy motion: any developer can sign up on the website with a low-cost credit card or a free account with limits, and then activate the platform by deploying a single agent across infrastructure.
“Any developer can sign up for Datadog through the websites. They can activate it at a low rate credit card or free account with some limits”
- Customer spend expands for two reasons: most customers are not digital natives and are rapidly growing their infrastructure footprint, and customers add products over time, typically going from one or two products to three or four, then six or more.
“most of their customers aren't being digital natives, are actually increasing the size of their infrastructure footprint rapidly.”
- Datadog's growth rests on a land-and-expand model with three levers: adding new customers at 30 percent or more a year, expanding spend per customer by 30 percent or more a year through net dollar retention above 130 percent, and launching new products at a 30 to 40 percent annual rate, which together produce 50 to 60 percent revenue growth.
“I think that what's unique with Datadog's approach is the really rapid product development cadence.”
- Datadog's sales and marketing spend fell to 26 percent of revenue in the most recent quarter from 33 percent a year earlier, while R&D rose to 30 percent of revenue and now exceeds sales and marketing spend, which is unusual among SaaS companies.
“So in the last quarter, they only had to allocate 26% of revenue to sales and marketing that's down from 33% a year ago.”
- Investors should not underestimate land-and-expand, which combines high new-customer addition rates with rapid expansion of spend per customer, and the founders' ability to anticipate the next product set is grounded in their technical backgrounds.
“One is absolutely don't underestimate the power of land and expand.”
- Datadog's revenue growth comes predominantly from expansion of existing customer spend, which accounted for 70 percent of last quarter's revenue growth, so sales and marketing efficiency matters less than expansion within the installed base.
“that spend expansion drives the majority of their revenue growth each year”
- Product development philosophy, specifically the rate of change, is a competitive advantage: being first to deliver the three pillars of observability and continuing to expand has produced the highest customer addition and expansion rates, and this feeds a business flywheel that Datadog can sustain because it reinvests more revenue into R&D.
“product development philosophy can be a competitive advantage. Datadog was the fastest to build the three pillars of observability”
- Datadog's adding customers, expanding spend, and adding products feed each other in a flywheel, and Peter argued this dynamic alone will guarantee a much larger market cap in the future.
“it's almost like those three, just continue to feed on each other, adding customers, expanding spend, and adding more products”
- Datadog's dollar-based net retention rate has been over 130 percent for four years, meaning existing customers on average increase spend by 30 percent or more each year.
“that's been over 130% for the past four years. And really why that's important is that means that customers on average, then increasing their spend by 30% or more each year.”
Datadog's scale and first-mover lead set it apart while hyperscalers stay out of full observability.
- Datadog processes over 10 trillion events a day across millions of servers and containers, and this scale is hard for competitors to replicate.
“they just got really massive scale that process over 10 trillion events a day and cover millions of servers and containers across all of their customers. And that scale is really hard to replicate”
- Datadog's advantage is that it was first to deliver the three pillars of observability, letting it keep adding platform capabilities faster than competitors, including user experience monitoring, network monitoring, serverless, database monitoring, and now security.
“I think that Datadog has the advantage because they were the first to deliver those three pillars of observability.”
- Datadog's competitors fall into three groups: open source do-it-yourself tools, legacy commercial vendors like IBM and similar providers built for companies running their own infrastructure, and modern cloud-era point solutions such as New Relic and Dynatrace for APM and Splunk and Elastic for logs.
“The competitors to Datadog kind of break down into three areas, the open source do it yourself solutions”
- Datadog's acquisitions of Sqreen and Undefined Labs add new product lines, such as application security and pre-production testing, that expand the total addressable market and give Datadog upsell options to existing customers.
“So these acquisitions are, I would say primarily extending the TAM”
- Security companies such as CrowdStrike are moving into observability; CrowdStrike bought Humio for log analysis, primarily to support its XDR offering, and has signaled intent to expand into traditional observability use cases, so Datadog's move into security will meet adjacent competitors coming the other way.
“CrowdStrike is a leader in security. They recently bought a company called Humio, which does log analysis.”
- The major cloud vendors offer only basic infrastructure monitoring and have not evolved full observability products, so they have chosen partnerships with independent providers like Datadog, which suggests the hyperscalers are not trying to compete in this space.
“the cloud vendors. They provide some very basic tools for infrastructure monitoring, but they really haven't evolved their observability products.”
- Splunk has grown through acquisitions that were bolted on, such as its APM product, which for a long time sat separately from its core log analysis, forcing users to switch between two interfaces, so Datadog's offering remains more complete and mature from a product standpoint.
“Splunk has been adding the capabilities around the three pillars of observability through a lot of acquisitions have been done in kind of a bolt-on fashion”
- Datadog has co-selling partnerships with AWS, Azure, and GCP that include marketplace listing, some level of consolidated billing, and system integrations, while the relationships remain neutral rather than pushing Datadog.
“Datadog has a co-selling partnership to all the major cloud vendors, including AWS, Azure and GCP.”
- Dynatrace is Datadog's biggest modern competitor because of its strong enterprise penetration and mature APM product, but Datadog has been out-executing it, with Dynatrace's revenue growth about half of Datadog's and customer growth of 23 percent versus Datadog's 36 percent.
“Dynatrace, their revenue growth was about half of Datadog's and their customer growth was much smaller than I was at 23%. Whereas Datadog it was 36%.”
One consolidated data model and interface beats fragmented single-purpose monitoring tools.
- Before Datadog, teams either wrote their own scripts and dashboards or used open source tools, while commercial vendors such as Splunk for logs, New Relic and Dynatrace for APM, and Nagios for servers each covered only one monitoring type, so troubleshooting an outage meant toggling between several products.
“they mostly focused on just one type of monitoring. So examples would be like Splunk for log analysis for New Relic or Dynatrace for APM”
- Datadog started with infrastructure monitoring in 2012, expanded into application performance monitoring in 2017, and then added log analysis, forming three capabilities that it calls the three pillars of observability.
“they started in infrastructure monitoring in 2012. They then expanded into application performance monitoring in 2017. They've since added log analysis”
- Datadog has integrations with over 450 external systems for data inputs, which makes setup easy for customers because it connects to the application services and frameworks developers usually want to use.
“have integrations with over 450 external systems for data inputs, which really makes it easy for their customers to set up Datadog”
- A website's stack has three layers: servers running on cloud hardware whose performance is tracked through metrics like CPU, memory, and disk; application software handling rendering, business logic, and payments; and a database storing product, pricing, and user data. Monitoring has to cover all three.
“For a lot of websites, first you have a set of servers. That's generally the hardware that is supposed to sit on a cloud environment.”
- Each server keeps its own event logs with timestamps, so across hundreds of servers those logs must be brought together and analyzed centrally to find common errors and user events.
“the problem is that across hundreds of servers, these server has it's own set of logs. So it's really important to bring all those logs together”
- Datadog's key insight was to bring log analysis, application traces, and infrastructure metrics into a single data model and single user interface, so that an issue like a non-refreshing leaderboard could be traced directly to a slow database without switching between three tools.
“Datadog's unique insight was the idea of bringing all three of those together into one interface.”
- Because Datadog bills across products, a customer can consolidate other security or monitoring vendors' spend into one Datadog bill, negotiate volume discounts, and displace a competitor's contract, while gaining one consolidated interface for all tools.
“Now I may have the ability to pull all of that into one bill with Datadog, where I can put a large customer and negotiate some volume discounts and displace that cost with that other vendor”
Customers like Instacart and Peloton show unified monitoring delivering concrete operational and security outcomes.
- Datadog's log rehydration feature lets customers pull in historical server logs going back several years, which automates security audit work that previously required logging into individual servers and writing custom scripts.
“before, prior to Datadog, all that work had to be done manually, they would operate this, would have to log into individual servers, pull those logs manually”
- Instacart uses Datadog log analytics to track each user session end to end, seeing which products were viewed or requested, the database queries made, and the response times and errors encountered.
“they can see which products they use or requested, the database queries that were made, and the response times and errors that showed up”
- Instacart used Datadog security monitoring to detect credential stuffing, where one user tries to take over other users' accounts by guessing passwords, surfacing these cases on a dashboard so operators could quickly block the offending users.
“Through its security monitoring Instacart was able to find those cases where a single user was trying to take over another user's account”
- The primary customer inside companies like Instacart is the engineering team, and specifically the DevOps function, which reports up to the CTO or VP of engineering and is responsible for uptime, usability, and scaling.
“The customer is primarily the engineering team, but it really leverages this concept of DevOps.”
- Peloton adopted Datadog's APM after becoming a customer in 2017, instrumented the endpoints behind its virtual leaderboard, and reduced response times on several of those endpoints by 80 to 90 percent, which improved user experience and cut server usage.
“they were able to reduce the response time of several of those endpoints by 80 to 90%, which surely benefits the user experience. And then also cuts down on server usage.”
Simplicity and easy activation let Datadog beat do-it-yourself open source monitoring.
- Datadog's leadership says most of its new customer wins are greenfield deals in which it replaces in-house open source solutions, rather than displacing other modern commercial providers.
“most customer wins are greenfield claiming that they are generally still replacing in-house open-source solutions”
- Customers will pay for simplicity and ease of activation, and Datadog's success in displacing open source alternatives shows that a solution covering most use cases and easily activated can beat do-it-yourself tools.
“customers will pay for simplicity and ease of activation.”
- Open source monitoring is free in license terms, but it carries a hidden cost because customers must pay operations and engineering staff to set it up and run it, which diverts those resources from building customer experiences.
“on the open source side, of course that's free, there's a hidden cost there.”
- Datadog is easy to activate because a single Datadog agent is deployed across infrastructure, and additional products such as APM, log analysis, or user experience monitoring are added with clicks in the admin tool, while prebuilt integrations remove the need for custom integration code that open-source users must write.
“It's a single agent, the Datadog agent that goes on all of your infrastructure and then adding on new products is just an exercise in the admin tool.”
- Listeners who want to go deeper can watch replays of Datadog's annual conference Dash, including the 2020 presentations on YouTube, and Datadog publishes free primers on topics such as observability and security.
“they hold an annual conference called Dash. It occurs in the second half of the year, but you can go back and watch the majority of the presentations from Dash 2020.”
Datadog's revenue, customer count, and large-customer share show strong growth.
- Datadog generated about $600 million in revenue in 2020, up 66% year over year, is tracking toward $1 billion in revenue for the current year, and is profitable on a non-GAAP basis with about $60 million in operating income and an operating margin of about 11% that is still rising.
“they generated 600 million in revenue, which was up 66% year over year. For this year and they're tracking towards 1 billion in revenue.”
- Datadog had over 16,400 total customers at the end of the most recent quarter, up 36% from a year earlier, after roughly 8,800 customers in June 2019, and its customer base is split fairly evenly across SMB, mid-market, and enterprise companies.
“they had over 16,400 total customers, and that was up 36% from last year”
- Datadog's large customers, defined as those with more than $100K in ARR, now account for about 80% of revenue, and their count tripled over two years from 590 at IPO, making them a critical growth driver.
“large customers, which they define as more than 100K in ARR, these now represent about 80% of Datadog's revenue”
Steady product launches and acquisitions keep extending Datadog's product breadth.
- Datadog has launched new products steadily, from infrastructure in 2012 to APM in 2017 and logs in 2018 and 2019, and new product launches add to customer spend as product adoption follows a predictable pattern where older products grow with time in market.
“products that were launched in the past that usage basically place scales by how long the product been in the market.”
- Datadog's M&A playbook is to stop selling an acquired company's product to avoid customer confusion, then move the engineering team and technology onto the Datadog data model and user interface and relaunch it as a Datadog feature.
“they basically will shut down taking on new customers for that company, and then really focused on the engineering teams”
- The Undefined Labs acquisition moves Datadog from a production-only focus into pre-production development and test environments, giving it a bridge into the developer workflow.
“Datadog's products have been primarily focused on a production environment. And what Undefined Labs does is pushes them into more of what we call it pre-production environments”
- Datadog's product development focuses on the intersection of development, security, and operations (DevSecOps), and its newly launched incident management product extends it into coordinating outage response, opening new avenues beyond monitoring.
“Datadog launched an incident management product, which provides a means for operations change, to coordinate and communicate around outages.”
- Datadog acquired Sqreen, a SaaS application-level attack detection platform, in February and folded it into a beta product called application security, which looks for code-level vulnerabilities such as SQL injection and cross-site scripting.
“the first is in February, they announced the acquisition of Sqreen, Sqreen is a SaaS based security platform”
The observability market is very large and extends beyond traditional software delivery.
- Datadog has not released blockchain-specific tooling, but Web3 applications such as dApps still rest on internet infrastructure that needs monitoring and that users expect to perform well, so Datadog has every opportunity to participate.
“a lot of that is still based on internet infrastructure. Those are applications that will still need to be monitored.”
- Observability is a roughly $44 billion market by 2024 per Gartner, and if five to 10 percent of cloud infrastructure spend goes to monitoring, with cloud infrastructure spending around $200 billion run-rate growing 36 percent, the opportunity remains very large relative to Datadog's roughly $1 billion revenue.
“the observability market is and Gartner estimate 44 billion by 2024.”
- The biggest risk to Datadog is that its business depends on software being delivered over the internet, and new delivery paradigms such as no-code and low-code platforms and Web3 models like dApps and blockchain may need less or different observability tooling, so the threat may come from a shift in delivery paradigm rather than a direct competitor.
“Datadog's business, truly dependent on this notion of delivering applications over the internet.”
- Observability is the general practice of instrumenting any system so its outputs can be measured and its state determined, and it could be applied beyond software to processes like manufacturing, sales funnels, marketing campaigns, and customer service, expanding Datadog's potential market.
“It's really just the idea of taking any system and instrumenting it and then making it so that you can measure the outputs and really determine the state of the system”
- Datadog is a monitoring and analytics platform for cloud-based application infrastructure that gives IT teams real-time visibility into software delivery stack performance, spanning from hardware, network, and operating system up to application code, user experience, and third-party dependencies such as databases, APIs, and CDNs.
“Datadog is a monitoring and analytics platform for cloud based application infrastructures using the Datadog platform, IT teams get visibility in real time into the performance of their software delivery stack.”
Datadog's founders built the product from DevOps silo friction they experienced themselves.
- Before DevOps, developers and operations were separate and often at odds, making it hard to tell whether a performance problem was an operations issue like a full database disk or a developer issue like faulty code. A shared toolset lets both sides see the same data and troubleshoot together.
“You almost see a lot of finger pointing where it was very hard if there was a performance issue on a site to isolate”
- Olivier Pomel and Alexis Lê-Quôc founded Datadog in 2010 after experiencing DevOps friction firsthand while working at Wireless Generation, where Pomel ran development and Lê-Quôc ran operations. They attended developer meetups and conferences and decided to solve the problem they had lived through.
“Datadog was founded by Olivier Pomel and Alexis Lê-Quôc in 2010. They were both working at a company called Wireless Generation”
- Datadog's founding opportunity came from an organizational inefficiency: siloed development and operations teams. Its founders turned that people problem into a product that offered a consolidated monitoring toolset.
“technology opportunities can really start as people problems”
- Because Datadog's technical founders had lived the problem as customers themselves, they could form a very clear picture of what an appropriate product should be, which is why the founders still leading the company are a competitive advantage.
“they've lived the problem, they were customers of this type of service. They know what customers want.”
Usage-based pricing keeps customer spend predictable, though hosting costs weigh on margins.
- Some modern competitors are moving to all-in-one pricing with one universal usage fee based on data processed or users, whereas Datadog keeps a la carte pricing because it believes most customers prefer paying only for the features they want.
“some of them are starting to move to these all in one pricing models”
- Datadog prices usage-based, charging per product per month based on a usage measure specific to each product: APM at about $30 per host per month, infrastructure at half that, logs at about 10 cents per gigabyte ingested, and security monitoring at about 20 cents per gigabyte, which makes spend predictable and controllable for customers.
“APM starts at $30 per host per month. Infrastructure is half of that. Logs can be ingested for 10 cents a gigabit.”
- Software companies that sell packaged software the customer hosts can achieve gross margins above 80 percent, whereas Datadog's hosted model carries cloud provider costs that reduce its margin.
“some software companies that just sell packaged software that the customer hosts where their gross margins are going to be over 80% or more.”
- Datadog's gross margin runs in the upper 70 percent range, above Twilio because Datadog does not pay per-message origination costs to third parties, and its long-term target is 75 to 80 percent, though it still pays cloud infrastructure providers that host its platform.
“Datadog, their gross margins run in the upper 70% range. Their long-term target is between 75 to 80%”