For Y/Y non-GAAP operating cost leverage, general & administrative (G&A), S&M 2 points and R&D 6 points. Fixed cost leverage, thanks to outperforming revenue, was the main driver here. Non-GAAP OpEx metrics remove the leverage benefit from lower stock comp as it moves further away from the IPO. I think that offers a better picture of structural margin expansion.

It sees growth in the second half of the year slowing as expected. It enjoyed a few unique tailwinds last year (Barbie and favorable FX) that will not recur during the second half. Still, its guide implies 28% Y/Y bookings growth, and leadership is always conservative. Q/Q EBITDA margin is expected to fall due to more hiring.

Elite execution came from the same recipe that has powered this annoying owl’s success since well before the IPO. It has the largest dataset in language learning by a country mile. Its team is as obsessive with split testing as I am with researching companies. It exhaustingly toys with variables to gain basis points of engagement, teaching efficacy and retention progress. These basis points add up.

This is why 90% of its revenue growth comes from organic channels rather than paid marketing. This is why it continues to see DAUs as a % of MAUs aggressively jump from 29% to 33% Y/Y and this is why results look so impressive. Chegg and Babbel fixate on paying for growth; Duolingo fixates on building a better product and letting growth naturally follow. It spends very little on marketing on a relative basis vs. these two, yet continues to compound while they falter.

The top-of-funnel remains wonderfully strong and Duolingo continues to enjoy a strong cadence of returning lapsed users. As an important aside, this is why alt-data for downloads (that shows recent slowing) is irrelevant; it’s a byproduct of Duolingo having well over 500M downloads and a larger portion of its DAU growth coming from people who have already downloaded the app.

Social media impressions rose 190% Y/Y and 20% of its DAUs now have year-long streaks. They’re hooked. All of this contributes to its expectation of maintaining 50%+ DAU growth for the foreseeable future. At this scale and in this environment… that growth is absurd.

Duolingo Max is live for 15% of its DAUs, and uptake so far has been very strong. This helped drive the beat and raise, but the vast majority of the financial impact here will not be felt until 2025. GenAI continues to be the secret sauce for this subscription tier’s early traction. In the coming months, it will introduce a new GenAI-powered tool to have real-time conversations with its characters. This has been among the most requested features for Duolingo, and GenAI makes the cost of building it possible. 

That leads us to another interesting point. Many assumed GenAI would kill this bird… I said not so fast… results point to GenAI augmenting operations, if anything. Like the new conversational tool and its radio podcasting project, the tech is shrinking years of content creation time down to a few months. DUOL’s preferred OpenAi partnership is helping too. Expedited timelines allow Duolingo to actually pursue these opportunities. It allows it to upgrade how fun, valuable and engaging its content is.

Some argue that real-time translators will kill it, but as leadership reminded us, that technology is a decade old. Furthermore, Duolingo caters to learners wanting to immerse themselves in a new language. Its users are trying to gain an edge in an interview room for a job or to impress a cute girl. Talking to someone through a phone is not a replacement for that.

“We’re not particularly worried about GenAI replacing us.”

Co-Founder/CEO Luis von Ahn

Since Duolingo added a dedicated Family Plan team last quarter, traction here has accelerated. 20% of its subscribers are now on this plan and retention benefits are notable.

It also keeps adding more social tools. It created the “Friend Streak.” This allows friends to share their learning journeys and even nudge each other with reminders to do their lessons. Duolingo does this nudging on its own, but now it has found a creative way to motivate passionate learners to do that for them. Making Duolingo more social has consistently delivered engagement gains and this time should be no different. Early results for this launch are “excellent.”

Duolingo has operationalized a global growth playbook in under-penetrated markets. It’s very similar to what Airbnb has done to accelerate volume in its less mature geographies. Duolingo starts with a bit of performance advertising, adds marketing directors, builds up its social presence and then lets things blossom from there. In Japan, this led to 93% Y/Y DAU growth and 58% Y/Y bookings growth. It sees many more markets like this one, and added marketing managers in France and Korea as a result. What a luxury it is to sell a digital product that is deeply relevant across the entire globe.

Duolingo’s multi-year push to deliver more advanced English content continues. It now has this content in all 20 of its English courses and added a standalone English course for its top learners. It’s getting better at starting students (with widely ranging experience levels) at the right level of difficulty, but thinks there’s more work to be done there.

“There's a lot of uncertainty in the external world. When we look at our numbers, we're not seeing anything. Our consumers don't seem to be reacting to anything. There's nothing worrying. In the past, we've had similar situations where there's uncertainty in the world. And when we look at our consumers, we just can't see anything. We don't really know why that is. I'm not going to say that we are recession-proof. We just don't know because we've never really gone through a recession. Because we have such a good free tier, people who can’t pay us are just using that.”

Co-Founder/CEO Luis von Ahn

Growth for both of these newer products was called strong. Duolingo does not monetize new products until a few years of content creation, product-market fit and scaling.

Duolingo is one of one in its combination of growth, GAAP margins, strong cash flow, rapid leverage and a large opportunity. When you add a sub-40x non-GAAP earnings multiple to that, I think there’s quite a bit to like. I continue to think this is the most underrated team in public markets and continue to be amazed by how well it performs every single quarter. Its core business is thriving, new products provide more growth prospects in the years ahead and this company keeps killing it. Go owl go.

Datadog is a dominant player in the data observability space. Observability simply refers to the practice of monitoring an entire software ecosystem to track issues, vulnerabilities and performance. Other players within this area include the hyper-scalers, Splunk, Elastic, CrowdStrike (through its Humio acquisition) and many more. Datadog splits its observability niche into 3 smaller buckets: infrastructure monitoring, log management and Application Performance Monitoring (APM).

Infrastructure monitoring: provides a holistic view of assets like servers and networks. It automates the collection of traffic and overall usage insights. That means it can more expediently fix and uncover infrastructure issues.

Log (or record of event) management: manages “timestamped records of events” occurring across the entire infrastructure. This also facilitates faster issue remediation and optimization of performance. These logs are organized and utilized within infrastructure monitoring and other use cases to identify things like customer service issues. Log management encompasses the collecting, maintaining, and leveraging of log data. This product routinely supports infrastructure monitoring, BUT there’s a key difference between the two. Log management handles event-based data, while infrastructure monitoring (as the name indicates) handles infrastructure-based metrics.

Application Performance Monitoring (APM): tracks app performance and uncovers/prioritizes performance issues to be remediated.

These three product categories closely tie together to form its “unified platform.”

Because Datadog already handles network viability, security is a wonderfully relevant growth adjacency. Products like Cloud Infrastructure Entitlement Management (CIEM) for example, ensure identity controls are strict and minimum access permissibility is in place. There’s a lot of competition here, but Datadog is no slouch. CIEM diminishes risk of identity attacks in a cloud environment. Its Security Information and Event Management (SIEM) product allows for “long term data log visualization for security investigations.”

“At Datadog, we're focused on helping our customers observe, secure, and take action on their complex systems."

Co-founder/CEO Oliver Pomel

“We base our guidance on trends observed in recent months and apply conservatism on these growth trends.”

CFO David Obstler

Next quarter guidance was slightly light on revenue, ahead on EBIT and ahead on EPS.

Datadog trades for 69x 2024 earnings. EPS is expected to grow by nearly 20% this year and by 22% next year.

Product expansion was the centerpiece of this call. Within GenAI and non-AI products, Datadog is quickly expanding its suite of use cases within various observability buckets, cloud and data security, cloud service management and GenAI.

Datadog Flex logs were fully rolled out during the quarter. As a reminder, this is its cost effective means to store and retain large batches of logs. They’re priced at $0.60 per 1 million annually and allow for separation of storage and query costs. This makes it ideal for long term data storage and regulatory compliance. With Flex Logs, storage and computation can scale in a parallel, independent manner. This separation for Datadog’s clients unleashes far more data scalability, customization and cost optimization. Conversely, querying from a flex log is slower than for Datadog’s standard log tier. That makes Flex Logs better suited for lower priority data.

Datadog is pushing further into digital experience monitoring. Datadog Synthetics is its way of testing interactions and processes in a zero stakes environment. It allows companies to predict and simulate usage patterns to test interface and product resilience. It can uncover issues before those issues ever are deployed as part of a software package. This is somewhat similar to real-time user monitoring (RUM), except RUM is tracking actual user actions, rather than a simulated user. This can scrape findings from how a user experience is progressing and if it needs to be tweaked. Datadog can provide churn analysis, engagement metrics and more. Each product now has $100 million in annual recurring revenue (ARR), which now gives the firm 5 products with that amount of business.

Within cloud security, there were several new product announcements. It added agentless cloud scanning, which allows Datadog’s security tools to run in a client’s client environment without needing to actually install any separate software (or agents). Some customers prefer this agentless approach, some don’t. Datadog now provides both. It added code security to allow “customers to detect and prioritize code-level vulnerabilities in their products and apps.” This also pushes Datadog “further left” in the DevOps space towards source code creation and maintenance. 

Its new data security tool (only for AWS for now) can uncover sensitive data, which is a perfect complement to its asset, app and infrastructure observability cores. Data is the engine driving every single asset and product a modern enterprise provides. Datadog now not only monitors the hygiene and usage of this data, but flags vulnerabilities too. Finally, Live Debuggger frees developers to “step through code in production environments & find the root cause of coding errors.”

The firm also added an OpenTelemetry integration with its agent. OpenTelemetry is a popular open-sourced means of collecting various kinds of data. Datadog now allows this to happen from within its environment without the need for a separate vendor.

Log Workspaces is its new collaborative environment for cross-department engineers to collect and combine assets. From there, they can build complex, “multi-stage” queries that can pick and pull needed data from different parts of an enterprise to combine for new use cases.

Its new Kubernetes Autoscaling tool handles resource usage and expansion optimization. It pulls from extensive usage data to tell customers where they can save on compute capacity and other areas. This is part of its cloud service management push. Another big piece of this push is the new Datadog App Builder. This allows developers to easily build applications with native integrations into the rest of their Datadog products. 

Toto is its first foundational large language model (FLLM). It utilizes Datadog’s enormous supply of data to augment customer workflows. It is “state of the art” in terms of performance across all major FLLM benchmarks.

Bits AI is the firm’s copilot. It can summarize incidents and conversationally field questions. The chatbot can also now be installed as an agent within a customer’s architecture to automate performance or security investigations.

Finally, Datadog debuted LLM observability. This allows developers to watch, upgrade and maintain their LLMs to accelerate the deployment of GenAI apps. And that’s a very important idea. As I’ve said many times, the first wave of GenAI monetization has been within chips and servers… all hardware. That foundation is being laid to ensure the proper amount of capacity is in place to support decades of high performance compute (HPC) applications. That wave is where Datadog will make its GenAI financial mark. LLM Observability sets the table for expediting and strengthening the GenAI app wave.

Sticking with GenAI monetization for a moment, Datadog now has 2,500 AI integrations vs. 2,000 Q/Q. AI integrations within the Datadog platform allow customers to place their first party data right into its ecosystem to accelerate model and app work. These integrations seem to be delivering momentum, as 4% of its ARR is now from AI customers vs. 3.5% Q/Q. It also continues to add observability support for data players like Snowflake, with data being the salt to any GenAI model’s pepper.

If all of this product news is the cause, more cross-selling, vendor consolidation and platformization are the effects. During the quarter, DDOG signed its largest contract ever with a South American Bank. They were struggling with overarching visibility, and now with Datadog, they aren’t. It highlighted three large 7 figure deals with a travel management firm and security firm to drive point solution displacement, better outcomes and $500,000 in annual savings for one of the clients. Another one of them is saving $1 million per year in OpEx. The European Central Bank (ECB) cut two more vendors in favor of DDOG and is now using 17 of its products.

Point solution elimination routinely allows customers to pocket efficiency gains and do “more with less.” Especially in today’s environment, that is working. It’s allowing Datadog to cut through continued customer hesitancy to deliver strong results. Gross revenue retention stayed in the mid-to-high 90% range and it continued to see usage growth from existing customers accelerate. That was most pronounced within its large enterprise segment, with smaller customers yielding stable Q/Q usage growth.

This was a good quarter. The guidance raises weren’t as explosive as we’ve seen with a small handful of enterprise software names, but this was still better than most. DDOG is successfully expanding beyond its core observability niche and laying the GenAI product foundation needed to support app monetization down the road. It continues to guide very prudently, continues to find steady growth and keeps delivering more EBIT leverage. Rock-solid quarter from a rock-solid company. I don’t see 70x earnings as compelling for a 20%-25% EPS compounder, but the actual quarter was quite good.

The Trade Desk is the leading buy-side player in open internet advertising. The firm’s two most compelling revenue segments are streaming, where it has relationships with most major players, and retail media, where it works with countless Fortune 500.

Its platform allows advertisers to bid on & purchase unique impressions with surgical precision, scale and open reporting. Purchases are essentially made on an impression-by-impression basis to uplift targeting efficacy and to double ad return metrics. Needed data is infused into every purchasing decision to ensure programmatically purchased ads provide optimal value. No longer do advertisers need to commit millions at annual upfront events to reach audiences; they can commit to smaller purchases in real-time with fantastic accuracy. No more “spray and pray.”

Kokai is the name of its data-driven, AI copilot-infused platform. It combines TTD’s leading open internet scale with its vast roster of 3rd parties to inject more context into each decision. It tells advertisers who they should be targeting. Kokai does so through TTD’s decade of experience that allows it to essentially find groups of high intent “copy-cat customers” with similar interests. Advertisers onboard their first party data (what TTD calls “concentrated data seeds”) and The Trade Desk does the rest.

 Kokai allows buyers to focus on whichever variable, key performance indicator or campaign objective they’d like to. It allows all of this to be done in a self-serve fashion, or in a fully managed environment. Up to them. Finally, Kokai emulates the ease of data onboarding that has made Alphabet and Meta so popular.

Unified ID 2.0 (UID2) is its open internet, omni-channel identifier. It uses hashed emails to responsibly ensure consumer and brand comfort. It knows exactly who is accessing what site or app. Kokai tells you who to target while UID2 tells you where they are. Other products include:

TTD beat revenue estimates by 1.1% & beat its guidance by 1.7%. Its 24.5% 2-year revenue CAGR compares to 24.8% Q/Q and 23.7% 2 quarters ago. Retention has been over 95% for a decade. North America stayed at 88% of its total business, as the international runway remains massive. The demand environment was called stable, while streaming advertising growth has actually accelerated year-to-date vs. 2023. Continued structural tailwinds and the best execution in the firm’s history (per the team) were cited as the reasons for outperformance.

The Trade Desk sells for 60x 2024 EPS. EPS is expected to continue compounding at a low 20% clip for the next few years.

As I talk through this section, just keep in mind that Jeff Green is always rather sharp in his commentary towards Meta and especially Google. Both of those advertising businesses continue to perform very well as of last quarter.

The predominant theme of the call is how the open internet, and especially TTD, continue to steal more attention and advertising budget away from walled gardens. How? In essence, uncertainty is great for TTD market share. It’s during these times when chief marketing officers are pressed to deliver efficient growth and to obsess even more over returns tied to each dollar spent. For decades, that has meant dollars flocked to digital advertising mega-caps.

That’s because they are both world-class in their ability to let advertisers easily onboard data, create campaigns and immediately enjoy “cheap reach.” In a world where data and open bidding aren’t driving the decisions, this is what looks best to marketers. It’s how they maximize impressions per dollar spent. But? That is not the variable to be optimizing for. You want to maximize conversions per dollar spent. Knowing this, walled gardens make measuring that difficult too. They have been left free to measure and report return metrics however they so choose. This has led to the big boys claiming last touch/click attributions as their own conversions when that’s not the case. This makes ad returns for their specific channels look better than they actually are.

The result? Over the last few years, there has been routine “mismatching” of expected return on ad spend (ROAS) and the actual financial results delivered from campaigns. It has led to what Founder/CEO Jeff Green calls “the illusion of growth.” That has left companies eager to find new channels, with honest reporting, and superior targeting to stretch budgets further. It has left CFOs frustrated and demanding that marketing teams actually prove their spending is working. TTD delivers by:

That’s how it’s winning more budget from the richest, deepest moat companies on the planet.

And now, thanks to years of success and reputation building, it has reached the critical mass needed to deliver the same scale benefits (across the entire internet) that mega-cap tech delivers on their own. TTD can now provide the best of both worlds – scaled reach + relevant, honest reach. For all of these reasons, it expects to continue to materially outgrow its sector and take more market share in the coming years. The world is embracing the open internet and, for the first time, spending more hours there than within walled gardens. TTD is poised to thrive amid this secular shift.

Besides TTD’s consistent ability to cut impression booking costs in half for buyers and double ad returns, all of these ideas above are qualitative and somewhat subjective. I think it helps to back this up with quantitative data. And just like every quarter, it highlighted compelling case studies to engrain its points. HP embraced The Trade Desk to cut cost per reach by 23% and juice targeting precision. Rossman (2nd largest pharmacy in Germany) used Kokai and UID2 to juice reach for a target product by 170%. TTD was able to confidently, openly and surgically calculate that the retailer enjoyed 20 sales per 1,000 impressions. That’s phenomenal and led to Rossman further leaning it. And when Rossman decided to use its first party data seed with Kokai, it saw another 3x-4x boost in purchase intent from customers.

More generally speaking, Kokai builds on the 2x ad return lead that its old platform (Solimar) delivered. Its reach efficiency is 70% better than Solimar, cost per customer conversion is 27% lower and data elements per impression are 30% higher. Overall, key performance indicators that go into return metrics are 25% better. Green is extremely pleased with Kokai’s launch and traction. 

Sell-side players source demand for publishers and control ad frequency for them too. A lot of publishers can do this themselves. It’s a lot harder to “buy the entire internet,” rank the value of all these impressions and buy accordingly. In my view, and general market consensus, the buy-side provides more unique value than the sell-side. Gigantic publishers are better at finding and organizing their own buyers than buyers are at finding the perfect publisher to work with. 

For evidence, TTD’s take rate has been wonderfully stable at an elevated 20% since it went public. For more evidence, its growth rate remained over 20% Y/Y while everyone else in the space saw growth tank through 2022 and into 2023. This is the structural growth story in an otherwise cyclical sector because it provides superior value than a buyer can get anywhere else. All of this is also why it commands a larger portion of ad budgets than any other piece of media plans. Times are good? Great, more budget and more revenue. Times are bad? Great, more turmoil and long term market share gains.

“This is a buyer’s market. We represent the buyer.”

Founder/CEO Jeff Green

Green reminded us that he’s been saying the search giant won’t deprecate 3rd party cookies for years. He is always right and this time was no different. Importantly, UID2 plans don’t change because of this. 3P cookies only work on display. None of the channels powering TTD’s growth are supported by cookies, but UID2 supports them all.

From a regulatory point of view, the DoJ search lawsuit that Alphabet lost is less meaningful to TTD than next month’s ad-tech lawsuit. He thinks the DoJ’s case here is even stronger than its issue with the search king paying Apple to be the default search engine. Green sees Google’s attempt at winning more open internet advertising dollars as being deprioritized. That’s why Google Network (aggregated supply where ads can be placed including 3rd parties) keeps shrinking and why it’s no longer focused on upgrading 3rd party cookies. He thinks that’s good news for TTD, but also thinks that they’ll win no matter what regulators or the search giant’s leadership decide to do. Green called Google a weaker competitor than it has been in years, as the search giant has dozens of other things to focus on to drive its own success, and this no longer seems to be one of them.

“We will win no matter what, but the lawsuit will still be fun to watch.”

Founder/CEO Jeff Green

Green sees Amazon Prime video as another walled garden. Like Netflix didn’t have the scale to make premium-only work, he doesn’t think Amazon will either. He fully expects them to open things up eventually and expects to be a big part of that. He also thinks Amazon’s conflicts of interest are even deeper than Google, as it has a massive marketplace of first and third party goods to jumble into these impressions. The influx of ad inventory from Prime creating an ad tier has had no material impact on overall cost per impression or the competitive environment for TTD. 

Conversely, he praised the Netflix team and called them rational. The financial impact from this new preferred programmatic partnership will be more noticeable starting next year.

This was yet another excellent quarter from an execution machine. The elite team, market share trends, secular tailwinds and runway all make me more comfortable with the lofty valuation. It will need to keep thriving to continue earning its multiple, but I’m quite confident that will happen. This is in the Shopify, ServiceNow, Zscaler and Cloudflare ballpark in terms of highest quality enterprise software disruptors.